Spanking and Child Behavior Problems
6 The Boomerang Effect of Spanking
Public approval of spanking has declined drastically in the past generation. Actual use of corporal punishment (spanking) on older children has dropped by about half since 1975 (see Chapter 17). Despite this, spanking remains an almost universal experience of children in the United States because, as was shown in Chapter 2 on spanking in the United States, at least 94% of parents spank and slap toddlers. Because every child was a toddler, this means that being hit by parents at least occasionally is an almost universal part of growing up in an American family. The 94% rate also suggests that parents are either not aware of, or choose to ignore, the possible harmful side effects of slapping and spanking. Surveys of pediatricians, psychologists, and sociologists have found that they also seem to give little attention to the possible harmful side effects of spanking. A clear indication is the virtual omission of spanking in child development textbooks shown in Chapter 1 and the surveys of pediatricians and psychologists (Anderson & Anderson, 1976; Schenck et al., 2000; White, 1993). Ignoring the harmful side effects of spanking flies in the face of a large and highly consistent body of research indicating that spanking is associated with an increased probability of physical aggression and other antisocial behaviors. Gershoff's meta-analysis documents the results of 27 studies of the relationship of corporal punishment to aggression by children (Gershoff, 2002). All27 found that corporal punishment was associated with an increased probability of aggression. Since then, all studies conducted on this relationship have found that spanking is associated with an increased probability of aggression and antisocial behavior.
There are many reasons this evidence has been ignored. One of the most important is the belief that spanking is more effective than nonviolent discipline and is, therefore, sometimes necessary, despite the risk of harmful side effects. The sometimes refers to occasions when a child engages in repeated misbehavior, dangerous behavior such running out into the street, or morally offensive behavior such as hitting other children. These situations occur in the lives of almost all children and that is why, as will be explained more fully in the chapter on why parents are driven to spank (Chapter 18), the belief that spanking is sometimes necessary means that almost all children will be hit by parents.
David B. Sugarman and Jean Giles-Sims are coauthors of this chapter.
As we pointed out in Chapter 3 on spanking throughout the world and in Chapter 4 on the relationship between family size and spanking, the child's behavior is only part of the explanation for spanking. As we previously suggested the evidence on harmful side effects has probably also been ignored because of the belief that spanking is more effective than other modes of correction. Spanking is widely believed to teach a lesson that children will not forget. The results of the study described in this chapter confmn that belief, but in a way that parents do not envision. In fact, they show that the lesson is often the opposite of what parents have in mind when they say, "I don't like to spank, but I had to teach him a lesson."
The specific questions to be addressed in this chapter are:
* What type of study could provide evidence that spanking causes antisocial behavior? This is an important question because, although many studies have found a correlation, a correlation does not provide evidence that there is a causal relationship.
* Does the relation of spanking to antisocial behavior apply to toddlers as well as older children? This is an important question because some argue that it has no harmful effects at early ages (Friedman & Schonberg, 1996a).
* Does the context in which spanking occurs make a difference? This is an important question because it is widely believed that if parents spank in the context of a warm and supportive relationship or that if spanking occurs within a culture where it is the prescriptive or statistical norm, it will not have adverse side effects.
* If parents spank only rarely, will it still cause harm? This question is important because many argue that the rare and judicious use of spanking could have beneficial outcomes.
Spanking and Behavior Problems
There are theoretical reasons to think that the link between spanking and an increased probability of physical aggression by the child is strengthened because parents use spanking for the socially approved and moral purpose of correcting behavior-and because spanking itself is socially approved and viewed as morally correct. Therefore, we could predict that when parents, albeit unintentionally, teach that it is morally appropriate to hit to correct misbehavior, their children will learn to hit others to correct their behavior. Children are continuously faced with situations in which other children are doing something they consider seriously wrong and who persist in that behavior, such as "he squirted water at me and wouldn't stop," "she took all the dolls and won't give me even one," or, later in life, "he made a pass at my girl." If children have learned that it is socially and morally appropriate for parents to hit others to change their behavior, they are more likely to hit others themselves. This is one of the reasons why study after study has found that the more spanking a child experiences, the more likely that child is to hit another child. In addition, other processes, such as those investigated in Chapter 12 examining the link between being spanked and partner violence in adulthood, contribute to the criminogenic nature of spanking.
The adverse effects of spanking have been observed as early as the beginning of the second year oflife. For example, Power and Chapieski (1986) found that toddlers whose mothers frequently spanked had a 58% higher rate of noncompliance with mothers' requests than did children whose parents rarely or never spanked. Among kindergarten children, Strassberg, Dodge, Pettit, and Bates (1994) found that those who were spanked that year had double the rate of hitting other children in school. Using children of widely varying ages, Straus (2001a, Chart 7.1) found that children in the National Family Violence Survey who experienced frequent spanking were twice as likely to severely assault a sibling as children of the same age who were not spanked that year. Analysis of another national survey found that, compared with children who were not spanked, those who were spanked the most were four times as likely to be delinquent (Straus, 2001b). Moreover, Chapter 12 on spanking and partner violence shows that the parents in that study who recalled having experienced corporal punishment during their early teens (about one half of the sample), were three times more likely to have hit their spouse during the previous 12 months.
Which is Cause and Which is Effect?
All of the studies just cited, however, used a research method that can only show that spanking is correlated with behavioral or emotional problems of children. These studies could not show that spanking causes those problems. It is just as plausible that the child's behavior problems caused the spanking. Our perspective is that there is a two-way causal process. This means that, although misbehavior causes spanking, when parents spank to correct misbehavior, it has the long-term effect of increasing the probability of aggression, noncompliance, and, as will be shown in Part IV, delinquency and, later in life, marital violence and other crime. Establishing whether this proposed boomerang effect is correct requires either experimental or longitudinal research that examines change in antisocial behavior subsequent to the spanking. The few longitudinal studies that had been completed prior to the study in this chapter did not measure change in children's behavior subsequent to spanking, so they did not permit inferring a cause-effect relationship (e.g., Simons, Burt, & Simons, 2008; Simons, Johnson, & Conger, 1994). Others, (e.g., Eron, Huesmann, & Zelli, 1991) included spanking in a scale measuring harsh disciplinary practices and therefore could not examine the effect of spanking per se. This problem continues to exist in many studies (e.g., Bender et al., 2007; Ehrensaft et al., 2003) that investigate spanking, but only as part of a harsh discipline scale.
Ethical and practical issues make it impossible to conduct experiments that randomly assign children to spanking or to non-spanking parents. However, longitudinal studies can follow children over time in their natural environments to see what happens in the months and years after they experience spanking. Such studies can provide information about the effects of spanking, but only if the study measures change in behavior-that is, in the months and years subsequent to the spanking, did the child's behavior improve or get worse? Unfortunately, none of the longitudinal studies conducted up to 1997 measured change in children's behavior. Without such a control, even longitudinal studies that follow children for several years are unable to determine which is the cause and which is the effect. When a study fmds that the more spanking a child experienced at Time 1, the more aggressive the child is at Time 2, this could reflect a situation in which parents were responding to a high level of aggression at Time 1. Because aggression is a relatively stable trait, it would not be surprising to find that the most aggressive children at Time 1 were still the most aggressive at Time 2. The most one can conclude from longitudinal studies that do not measure change in children's aggression is that the spanking did not reduce the level ofthat aggression.
The study in this chapter permitted us to overcome the causal direction problem because antisocial behavior was measured at the start of the study and then again two and four years later. This allowed us to determine whether spanking 'resulted in less, or more, antisocial behavior two years later. We tested tpe hypothesis that the more parents spanked at Time I, the greater the increase in children :S antisocial behavior from Time I to Time 2.
Sample and Measures
Sample
The sample consisted of the 807 children of women who were first interviewed in 1979 as part of the National Longitudinal Survey ofYouth who were 6 to 9 years old when their level of antisocial behavior was measured. The data that is reported here was collected in three different waves, two years apart, which enabled us to examine the potential impact of spanking on antisocial behavior over time. For information on the sample and data see Giles-Sims (Baker, Keck, Mott, & Quinlan, 1993; Giles-Sims et al., 1995; Straus, Sugarman, & Giles-Sims, 1997).
The information on spanking was obtained from observation by the interviewer of whether the mother hit the child during the course of the interview and from asking, "Did you find it necessary to spank your child in the past week?" Mothers who said they had spanked were asked: "About how many times, if any, have you had to spank your child in the past week?" We used these data to create a spanking scale that combined the observed and the interview measures. If the mother was observed hitting the child, it was counted as one instance of spanking and this was added to the number of times the mother reported spanking in the previous week. From this scale, we formed four categories of children: those who experienced no spanking (during the interview or the previous week) and those who experienced one, two, or three or more instances of being spanked.
Antisocial Behavior
The measure of child antisocial behavior consists of the following six behaviors: cheats or tells lies, bullies or is cruel or mean to others, does not feel sorry after misbehaving, breaks things deliberately, is disobedient at school, and has trouble getting along with teachers. The mothers were asked if, during the preceding three months, each of the behaviors was not true ofthat child (scored as 1), sometimes true (scored 2), and often true, (scored 3). The antisocial behavior score is the sum of the scores for these behaviors.
Three Other Methodological Problems
In addition to research that does not permit inferring whether spanking causes antisocial behavior, much of the existing research also fails to deal with one or more of three other methodological problems, each of which could undermine the validity of the results.
Overlap with other parental behaviors. What seems to be an effect of spanking could be a spurious relationship. A spurious relation is one that results from some other variable being the underlying cause. For example, if parents who spank also are harsh and rejecting, and lack warmth and affection, those characteristics rather than spanking itself, could be what explains the correlation of spanking with antisocial behavior. Many studies show that harsh and rejecting parents do tend to spank more (Herzberger, 1990; Pinto, Folkers, & Sines, 1991; Simons et al., 1994). However, parents who spank are not usually harsh and rejecting in other ways. Spanking is something done by good parents in the belief that it is necessary to correct misbehavior. Moreover, if parents who spank are harsh and rejecting, because over 94% spank toddlers, it would mean that over 94% of American parents are harsh and rejecting. That is very unlikely. Therefore, it is also unlikely that the harmful side effects of spanking are the result of these other parental behaviors rather than the result of spanking itself Still, it is very important to analyze spanking within the context of parents' child rearing styles. This means that research must separate the effects of spanking from the effects of other parental behaviors. Two crucial aspects of parenting that have been identified in previous research are warmth and cognitive stimulation, which refers to interacting with the child in a way that encourages the child to think and understand (Maccoby & Martin, 1983). Emotional support is especially important to take into consideration because it is widely believed that, if spanking is done by parents who provide emotional support, use of spanking in moderation and as a backup rather than a first resort, is beneficial to children (Baumrind et al., 2002). To deal with this problem, the statistical analysis for this study took into account the level of emotional support and cognitive stimulation.
Overlap with sex of child and socioeconomic status. Another methodological problem can occur if the study does not take into account whether the child is a boy or girl and whether the family is low or high in socioeconomic status. With respect to sex differences, boys engage in more disruptive behavior, school truancy, and verbal and physical violence than girls (Cullinan & Epstein, 1982; Hyde, 1984; Maccoby & Jacklin, 1980). This may be one of the reasons parents are more likely to spank boys more than girls, as was shown in Chapter 2 and by Simons et al. (1994). Thus, part of the relationship between spanking and a child's antisocial behavior may reflect the fact that boys misbehave more and parents are more likely to spank boys to correct the misbehavior. This is what is known as confounding of the relationship between spanking and antisocial behavior with the sex of the child (Hoghughi, 1992). There is also the possibility that spanking has different effects for boys and girls. This is what is known as an interaction effect or a moderator effect. We, therefore, examined whether the relation of spanking to antisocial behavior is different for boys and girls.
A similar problem applies to socioeconomic status and racial or ethnic group because, as was shown in Chapter 1, low socioeconomic status parents and parents from some minority groups spank more (see also Bank, Forgatch, Patterson, & Fetrow, 1993; Giles-Sims et al., 1995; Sears, Maccoby, & Levin, 1957; Straus, 2001a). Children from low socioeconomi.c status families, moreover, havl higher rates of antisocial behavior and delinquency (Bank et al., 1993; Feshback, 1970; Junger-Tas, Haen Marshall, & Ribeaud, 2003; Simons, Gordon',Simons, & Wallace, 2004). These two factors together can present a methodological problem. It is especially important to investigate whether the effect of spanking differs by racial or ethnic group because it has been argued that, in the context of a community where spanking is the cultural norm, spanking does not carry the same harmful side effects. This is believed to be the case because children who are raised with different norms concerning spanking do not equate spanking with parental harshness or rejection (Gunnoe & Mariner, 1997; Larzelere, Baumrind, & Polite, 1998; Polite, 1996) and therefore suffer no ill effects.
Effect of age of the child. Spanking might have different effects at different child ages. For example, it is widely believed that spanking toddlers, if done in moderation, does not have harmful side effects (Friedman & Schonberg, 1996a), whereas spanking older children and teenagers is thought to interfere with the transition to adulthood and autonomy. However, there does not seem to have been any empirical research on this issue. Thus, the advice to spank only young children is based on cultural tradition rather than scientific evidence. It is at least equally plausible to argue the opposite-that spanking toddlers will have a greater effect because it occurs at crucial developmental stages. Indeed, the results of the study described in Chapter 1 0 show that the effect of spanking on mental development is greater for younger than older children. Because there are plausible grounds for expecting age differences, we repeated the preliminary analysis for children of three age groups: 3 to 5 years, 6 to 9 years, and 10 years and oyer. The preliminary results were parallel for all three age groups. As a consequence, for the reasons given in the section of the Appendix for this chapter, the final analysis was conducted only for children aged 6 to 9.
Correlation of Spanking with Antisocial Behavior
The data permitted computing correlations between the frequency of spanking and the child's antisocial behavior. Fifteen of the correlations are on the relation of spanking to antisocial behavior in the same year ( contemporaneous correlations), and 15 examine the relationship of spanking at Time 1 to antisocial behavior two years later (time-lagged correlations). All 15 contemporaneous correlations, and all15 time-lagged correlations found that the more frequently a mother spanked her child in the week she was interviewed, the more antisocial behavior that year and also two years later. (The correlation coefficients are in Straus, Sugarman et al., 1997.
Following the recommendation of Bruning and Kintz (1987), we investigated whether the size of the correlations differed by year of measurement or by the age and sex of the child. We found that, with only two small exceptions, the relation between spanking and antisocial behavior was consistent across all ages and all years of interviews, and across gender (see Straus, 1997, Table 1 ). The fact that spanking was correlated with antisocial behavior to about the same extent for 3- to 5-year-old children as for older children is very important. It suggests that, contrary to both popular and professional beliefs (Friedman & Schonberg, 1996a), spanking is just as harmful for toddlers as it is for other age groups.
Spanking and Change in Antisocial Behavior
For the reasons explained previously, ordinary correlations may be misleading. What seems to be an effect of spanking on antisocial behavior could really be the effect of some underlying variable such as socioeconomic status or lack of parental warmth. The most important limitation of ordinary correlations is that they provide no information to distinguish cause from effect. Does spanking cause antisocial behavior or does antisocial behavior cause spanking? To deal with these problems, we used the statistical method known as analysis of covariance. This controlled for differences in family socioeconomic status, sex of the child, and the extent to which the home provided emotional support and cognitive stimulation. Most important, it controlled for the amount of antisocial behavior at the start of the study. The results of testing the hypothesis with this method can determine whether there is a change in children's antisocial behavior two years later (i.e., subsequent to the year in which the spanking occurred and whether that change is in the form of an increase or a decrease in antisocial behavior). It can also determine whether there is an effect of spanking that is over and above the effect of family socioeconomic status, sex of the child, and the extent to which the home provided emotional support and cognitive stimulation.
Chart 6.1 graphs the results of the analysis of covariance. At the left side of the chart are the children whose parents did not spank them during the week studied in the first year of the study. Those children had an average decrease of four points on the antisocial behavior scale. Moving from left to right in Chart 6.1 to the children who were spanked once during the Time 1 week, shows that their antisocial behavior score increased by an average of two points. The children who were spanked twice also had about a two-point increase in antisocial behavior. The biggest change in antisocial behavior was for children spanked three or more times that week. Their antisocial behavior score increased by an average of 14 points. Thus, the more frequent the spanking in 1988, the greater the increase in antisocial behavior over the subsequent two years. Only among children who were not spanked the week before the Time 1 interviews did antisocial behavior decrease. The F tests and other statistics for these results are in Table 2 of Straus, Sugarman et al., 1997.
Chart 6.1 Children Who Were Not Spanked Decreased Antisocial Behavior, whereas Children Who Were Spanked Increased Their Antisocial Behavior Two Years Later. (Chart not available)
*Mean adjusted for T1 antisocial behavior, T1 cognitive stimulation, T1 parental emotional support, child gender, race, and socioeconomic status.
It is important to keep in mind that the terms increase and decrease describe change compared with other children in the study at each point in time. In absolute terms, the average amount of antisocial behavior decreases as young children mature (Tremblay, 2003). The way we measured antisocial behavior takes this average improvement in children's behavior into account because it measures how far above or below the average antisocial behavior of all children in the sample each child is at each time point.
Does the Context Make a Difference?
the results also provide information on whether the presence or absence of the variables we controlled affected the relation of spanking to antisocial behavior. As explained previously, it is widely believed that, if parents are loving and supportive, spanking does not have a harmful effect. When we investigated that possibility, we did not find a statistically significant interaction or moderator effect for emotional support. That is, spanking was associated with an increase in antisocial behavior among children whose parents were high in emotional support as well as among children of low-support parents. The fact that, even when there is lots oflove and support, spanking is associated with an increase in antisocial behavior does not mean that love and support make no difference. Our study, like many others, found that the more supportive the parent, the lower the average level of antisocial behavior.
We did find that the relation of spanking to antisocial behavior was influenced by two other variables. Chart 6.2 shows that the tendency for spanking to be related to an increase in antisocial behavior two years later is stronger for boys than for girls, and Chart 6.3 shows that the relation between spanking and antisocial behavior is stronger for White children compared with minority children. Nevertheless, although the amount of increase in antisocial behavior associated with spanking may be smaller for girls and minority group children, both experienced an increase in antisocial behavior in proportion to the amount of spanking they had experienced two years earlier.
The result for minority group children is particularly important because many minority group parents believe that under the high crime conditions of inner city life, their children need (to use one of many euphemisms for spanking) "strong discipline" (Alvy & Marigna, 1987; Kohn, 1969; Peters, 1976; Polite, 1996; Young, 1970). Children growing up in those difficult circumstances no doubt need closer supervision and control, but these results suggest that attempting to do this by spanking increases rather than reduces the risk that children will get into trouble. Other studies that also found that the harmful effects of spanking apply in different cultural contexts are in Chapters 10 and 14. See also the review by Lansford (2010).
Summary and Conclusions
We found that the more spanking by the mothers in this sample, the greater the chances that two years later the child would have an increase in antisocial behavior. The tendency for spanking to be associated with an increase in antisocial behavior applies regardless of the extent to which parents provide cognitive stimulation and emotional support, and regardless of socioeconomic status, ethnic group, and sex of the child. These results are contrary to the idea that, if parents provide adequate warmth and cognitive stimulation, spanking will not harm children. Our results are also contrary to the idea that spanking works better among lower socioeconomic status families or in some minority families. Moreover, the findings are consistent across years (1986 to 1988, 1988 to 1990), across types of analysis (multiple regression and analysis of covariance), and for all three age groups (ages 3 to 5, 6 to 9, and 10 and over). The consistency across age groups is important because it contradicts the widespread belief, held by both the general public and professionals advising parents, that spanking is acceptable if confined to preschool-age children. As we noted in Chapter 2, the consensus statement from an American Academic of Pediatrics conference in 1996 permitted spanking for children age 2 to 6. If the results in this chapter, and those in Chapter 10 on spanking and cognitive ability, had been available at that time, the outcome of that conference might have been very different.
Chart 6.2 The Increase in Antisocial Behavior Associated with Spanking Is Much Greater for Boys than for Girls (Chart not available)
*Mean adjusted for T1 antisocial behavior, T1 cognitive stimulation, T1 parental emotional support, child gender, race, and socioeconomic status.
Another question that is often raised is, "Does only one spanking make a difference?" We found that the increase in antisocial behavior starts with children whose mothers spanked them only once during the week of the survey. This is consistent with the results of studies by Larzelere and Grogan-Kaylor. Larzeler (1986) found ''there is no evidence that a threshold frequency of spanking is necessary before it begins to influence child aggression" (p. 31). Grogan-Kaylor (2004) found that even relatively infrequent spanking is associated with an average increase in antisocial behavior compared with children whose parents did not spank to correct misbehavior.
Chart 6.3 Spanking Increases Antisocial Behavior by Both White and Minority Children, but the Effect Is Slightly Less for Minority Children (Chart not available)
*Mean adjusted for Tl antisocial behavior, Tl cognitive stimulation, Tl parental emotional support, child gender, race, and socioeconomic status.
Since the study in this chapter was done, there have been at least 14 other longitudinal studies that have found that spanking is associated with a subsequent increase in maladaptive behaviors. Chapter 19 reviews those that examined whether spanking is associated with an increase in the probability of antisocial behavior and crime. It is important to understand the implications of the phrase "an average increase." When there is an average effect, it means that there will be children for whom the effect of spanking is greater than the average effect and children for whom the effect is less than the average. Or putting it another way, spanking does not always lead to an increase in antisocial behavior. As explained in Chapter 1, the effect of risk factors such as spanking is always in the form of an increased probability of the harmful effect, not a one-to-one relationship. A wellestablished example of a risk-factor effect is heavy smoking. About one third of heavy smokers die of a smoking"related disease (Matteson et al., 1987). But this high mortality rate also means that two thirds of very frequent smokers can point out that they smoked more than a pack a day all their life and have not died from a smoking-related disease. Similarly, most adults who were spanked can say, "I was spanked a lot and I'm OK." Although those who say these things are factually correct, the intended implication that smoking or spanking are safe is not correct. The correct implication is that such individuals are one of the "lucky ones." Thus, although most children who are spanked will not be high in antisocial behavior, that does not mean that spanking is harmless-just as the fact that most heavy smokers do not die, it does not mean that smoking is harmless.
The chapters that follow, and much other research, show that the behavior problems associated with spanking are not confined to aggression and other antisocial behavior. In addition, these studies, like the present chapter, reveal a dose-response to spanking, starting with even one instance. The effect size for one instance is small, but it exists. The more frequent the spanking, the greater the probability of behavior problems. Taking the whole range of spanking as measured by this study, a rough estimate of the potential for reducing antisocial behavior can be obtained by comparing the change in antisocial behavior scores for children who were not spanked in the past week with scores of children who were spanked 3 or more times in the past week. The score on the antisocial behavior scale of the children who were spanked 3 or more times is 18-points higher than the none group. The 6- to 9-year-old children in the 3-or-more-times group, who are 1 0% of the children in the study, would have had the greatest chance of improved behavior if their parents had not spanked them. In addition, because Chart 6.1 shows that even one instance of spanking in the previous week is associated with an increased risk of antisocial behavior, an additional 19.8% of children could benefit by a reduction in spanking from once to never, and 14.1% could benefit by a reduction from twice to once in a week. This comes to a total of 44% of the children in this national sample whose antisocial behavior could have decreased if their parents did not spank them or spanked them less often.
7 Impulsive Spanking, Never Spanking, and Child Well-Being
Impulsive behavior consists of acts carried out with little or no forethought or control, hot-tempered actions, acting without planning or reflection, and failing to resist urges (Hoghughi, 1992; Lorr & Wunderlich, 1985; Monroe, 1970; Murray, 1938). Parents may spank impulsively, or they may follow the recommendations of some advocates of corporal punishment such as Baumrind, Larzelere, and Cowan (2002) who find spanking acceptable if done as well-regulated spanking, that is, in a calm and collected way. Others believe in striking while the iron is hot. John Rosemond, the author of best-selling books on child rearing says that he believes in "spanking as a first resort; spanking in anger" (Rosemond, 1994a).
Vera E. Mouradian is the coauthor of this chapter.
Violence is the use of physical force to cause pain or injury (Gelles & Straus, 1979). Corporal punishment of children (spanking) is a legal form of violence, provided it causes only pain and not lasting physical injury and is for purposes of correction and control. The child who is spanked impulsively experiences physical attacks as part of one of the most crucial social relationships in their lives. As these experiences become internalized, they provide a model for the child's own behavior. Thus, impulsive spanking may be an important risk factor for impulsiveness in the child. This process may be an important one for understanding the development of criminal behavior. The criminogenic nature of impulsiveness has been shown in a number of studies, including a longitudinal survey of 411 London boys that demonstrated that "low intelligence, an impulsive personality, and a lack of empathy for other people are among the leading individual characteristics of people at risk for becoming offenders" (Farrington & Welsh, 2006, page V). Studies by Aucoin, Frick, and Bodin (2006) and Olson, Sameroff, Kerr, Lopez, and Wellman (2005) found that the more spanking experienced by children, the higher their scores on a measure of impulsiveness. But because neither study distinguished between parents who spanked impulsively and those who did not, this correlation might reflect the part of the sample whose parents spanked impulsively. We believe that regardless of whether it is carried out impulsively or in a planned way, spanking increases the probability of antisocial behavior by the child. We also believe that this effect is even greater when the spanking is impulsive because, in that situation, spanking models not only violence but also impulsiveness. The study described in this chapter addressed the following questions:
* What percent of children were never spanked?
* When parents spank, how often do they spank impulsively?
* Are mothers who spank often more likely to do so impulsively?
* Is spanking related to impulsivity and antisocial behavior in children, even when the spanking is not done impulsively?
* Is spanking more strongly related to impulsivity and antisocial behavior in children when the spanking is impulsive than when it is planned?
* Are children who were never spanked less impulsive than those who have ever been spanked?
* Does maternal warmth and nurturance moderate the relationship of spanking to child antisocial and impulsive behavior?
Impulsi;ve Spanking and Child Behavior Problems
It has been 60 years since Sears et al. (1957) found that spanking by parents was associated with a less well-developed conscience and higher levels of aggression in children. Some argue that the relationship between spanking and aggression by children depends on whether spanking is used impulsively or in a controlled way (Dobson, 1988; Friedman & Schonberg, 1996a; Larzelere, 1994), reflecting a belief that only impulsive spanking has harmful side effects. Others, however, recommend that parents spank in anger (e.g., Rosemond, 1994a, 1994b ), and many parents have told us that to do otherwise is coldblooded. This discussion raises two questions. One question is to what extent parents spank impulsively. The other question is whether spanking in general, or only impulsive spanking, is associated with child behavior problems. These questions are examined by reviewing previous research and then by presenting new research fmdings.
Prevalence of Impulsive Spanking
Only three studies were found that provided data on the prevalence of impulsive spanking, and even these two did not use the term impulsive to describe their findings. However, their definitions are consistent with what we are calling impulsive spanking: spanking little or no forethought or control, hottempered actions, without planning or reflection. Carson (1986) studied 186 parents in a small New England city. Her findings can be interpreted as showing that about one third of those parents spanked impulsively. Holden and Miller (1997) differentiated between instrumental spankers and emotional spankers. Emotional spankers felt irritated, frustrated, and out of control when spanking their children. They constituted just under one third of the sample of 90 parents who used spanking. In addition to these two studies of actual incidents of spanking, we located two other studies that provide an indication of the potential for parents to use spanking impulsively. A Canadian national survey (Institute for the Prevention of Child Abuse, 1989) found that 80% of parents said that, at least on rare occasions, they came close to losing control when disciplining their children. A study by Frude and Goss (1979) found that of 111 mothers, 40% were worried that they could lose control and possibly hurt their children. A third and longitudinal study of 132 parents from a southwestern urban region examined the context in which spanking occurs (Vittrup & Holden, 2010; Vittrup, Holden, & Buck, 2006). The authors found that almost a third (29%) of parents reported feeling angry while they were spanking, which could also be related to impulsivity.
Effects of Impulsive Spanking
A search of electronic databases did not locate studies of specific effects of impulsive spanking on children. However, two of the three studies just cited provide some indirect evidence of the relation of impulsive spanking to specific child behaviors. Holden and Miller (1997) found that emotional spankers were less likely to believe that spanking would result in achieving their goals for the child, such as immediate compliance, good behavior, and respect for authority. Similarly, Carson (1986) found parents who spanked when they lost control tended to be more likely to see spanking as ineffective. Therefore, parents seem to believe that the impulsive administration of spanking undermines its effectiveness. In addition, impulsivity implies lack of consistency in punishment, and inconsistent punishment has been found to be associated with lower levels of parenting self-efficacy (Acker & O'Leary, 1988).
There are theoretical grounds for expecting adverse effects of impulsive spanking on children. First, we could expect that impulsivity in the parent would provide a model of impulsivity for the child. Impulsivity in children has been found to be associated with conduct disorders and antisocial behavior, among other problem behaviors (Hoghughi, 1992; Olson et al., 2005; Schweinle, Ickes, Rollings, & Jacquot, 2010). Moreover, impulsivity in the form of low self-control is the core explanatory variable of the general theory of crime (Gottfredson & Hirschi, 1990; Rebellon, Straus, & Medeiros, 2008).
A child who experiences impulsive spanking may come to believe that the punishment results from the characteristics of the parent, rather than for the child's good. When spanking is perceived as parent-centered, it may be more likely to create resentment and anger, undermining the parent-child bond that is critical to preventing antisocial behavior (see Chapters 8 and 9). Impulsive spanking may also be more stressful for children, because it is less predictable and associated with more negative parental emotion. In other words, erratic spankings by an angry parent are probably more stressful for the child than spankings that follow some kind of consistent guidelines (e.g., every time a child throws a ball inside the house after repeated commands to stop).
Effects of Never Spanking
At the other end ofthe spanking continuum from impulsive spankers are parents who never spank. These parents are important to furthering our understanding of the effects of spanking on children. It is widely believed that spanking is sometimes necessary and that without it, children's behavior will become , out of control and increasingly antisocial. Although the results of the study presented in the previous chapter demonstrate that spanking increases the probability of antisocial child behavior, they do not refute the prediction of those who believe that never spanking would be harmful because that sample did not include a group who never spanked. The no-spanking group in that study were mothers who did not spank in the previous week. But there are 51 other weeks in the year, to say nothing of previous years. The data presented in this chapter permitted us to study the children of mothers who never spanked, as well as those who spanked impulsively
The concerns that parents who never spank will lack effective control and that their children will have behavior problems were put forward at a community meeting iJt one of the cities where the present study was conducted. The meeting was held to discuss a proposed community initiative to end or reduce spanking. One of the parents attending said, "If you nuts have your way, we're going to be a town with kids running wild!" At the same meeting, those wanting a communitywide effort to reduce or end spanking made exactly the opposite prediction. They argued that, on average, children who are never spanked will have lower rates of delinquency and fewer psychological problems in childhood and as young adults.
The importance of this controversy cannot be overestimated. Some of the most sophisticated defenders of spanking, such as Baumrind and Larzelere (e.g., Baumrind et al., 2002) deny that they approve of spanking. But they say that spanking is a safe and necessary backup. As a consequence, the most crucial test, for both opponents and advocates of spanking, focuses on the children of parents who never spank. However, there does not seem to have been a study conducted that compared never-spanked children with those spanked very rarely. This is a crucial missing link in the research on spanking because spanking only as a backup is the advice, not just of Baumrind and Larzelere, but of almost all the pediatricians and parent educators with whom we have discussed this issue. Although most are now against spanking, they are unwilling to say that children should never be spanked.
The data for this study enabled us to identify children in the sample who were never spanked. As a consequence, we can make a start on resolving this critically important issue. Four hypotheses were tested:
1. The more spanking used by the mother, the greater the child's impulsiveness and antisocial behavior.
2. The more impulsive the spanking, the greater the child's impulsiveness and antisocial behavior.
3. Spanking is associated with antisocial behavior and impulsiveness by the child only when spanking is impulsive.
4. Children who were never spanked have the lowest levels of antisocial behavior.
Sample and Measures
These hypotheses were tested using data on a representative community sample of mothers ofl,003 children, aged 2 to 14, living in two counties in Minnesota. The data were obtained in 1993 by telephone interviews with the mothers. The children were primarily from two-parent families (95.1 %). They were about equally divided between boys (54%) and girls (46%), and their mean age was 8.6 (median 9). The mean and the median age of the mothers was 37. They had been married an average of 13.9 years and had a median of two children living at home (mean= 2.5). Consistent with census data on the socioeconomic composition of these two communities, the sample was almost entirely Caucasian, and 31% of the mothers and 35% of the fathers were college graduates. Additional information on the sample is in Straus & Mouradian (1998).
Measures
Spanking. The mothers were asked how often in the past six months they had spanked, slapped, or hit the child when the child "does something bad or something you don't like, or is disobedient." The response categories, which are taken from those used in the Conflict Tactics Scales (Straus et al., 1998; Straus & Mattingly, 2007), were never, once, twice, 3 to 5 times, 6 to 10 times, 11 to 20 times, and more than 20 times.
Children who were never spanked were identified on the basis of two questions that asked the mother the age at which spanking was first used and the age at which spanking was used the most. If the mother responded to both of these questions that she never spanked her child, we classified the child as not having been spanked. We used these data to classify the children into the following categories: never (189 children); not in the past six months (408 children); once in the past six months (98); twice in the last six months (81 children); 3 to 5 times (86 children); and 6 or more times (71 children). Thus, even in this low spanking community, only 20% of the children had never been spanked.
Other discipline. We measured non-corporal punishment discipline by asking the mothers how often in the past six months, when the child had done something bad, had done something the mother did not like, or had been disobedient, that she "Talked to him or her calmly about a discipline problem, sent him or her to his or her room, or made him or her do 'time-out,' took away something, or took away some privilege like going somewhere." Response categories were never, once, twice, 3 to 5 times, 6 to 10 times, 11 to 20 times, and more than 20 times. Response categories were transformed to the midpoints of the category (3-5 = 4, 6-10 = 8, 11-20 = 15, more than 20 = 25) and were summed. The resulting non-corporal punishment intervention scale scores ranged from 0 to 7 5 with a mean of26.6 and a standard deviation of19 .6. The alpha reliability was .71. The scores were grouped into the following four categories for use in the ANOVAs: 15 times or less frequently (n = 326), 16--30 times (n = 244), 31-45 times (n = 196), and more than 45 times (n = 167).
Antisocial behavior by the child. Antisocial behavior was measured by asking the mother about 11 behaviors that involved acting out against other people including the child's family, teachers, and peers. Eight of the items were asked regardless ofthe age of the child: How often in the past six months the child was cruel or meanto other kids, a bully; cruel, mean to, or insulting to the mother; in denial of doing something he or she really did; hitting a brother or sister; hitting other kids; hitting you or other adults; damaging or destructive to things; and stealing money or something else. The remaining three items depended on the age of the child. Mothers of preschool-age children (2 to 4) were asked how frequently their child "refuses to cooperate; repeats misbehavior after being told not to do it; and misbehaves with a baby sitter or in day care." Mothers of school-age children were asked how frequently their child "disobeys you; is rebellious; and has discipline problems at school." The response categories for all items were: 0 = never, 1 = rarely, 2 = sometimes, and 3 = frequently. The alpha reliability for this scale was .81.
Child impulsiveness. This was measured by two items asking how frequently in the previous six months the child had "temper tantrums, hot temper" and "acts in unpredictable, explosive ways, impulsive." These items were chosen to reflect two often-cited features of impulsivity: acting quickly without apparent thought or a lack of planning, or failing.to resist urges (Hoghughi, 1992; Lorr & Wunderlich, 1985; Murray, 1938) and being quick or hot-tempered (Hoghughi, 1992; Monroe, 1970). Response categories for these items were: 0 =never, 1 = rarely, 2 =sometimes, and 3 =frequently. The item scores were transformed to Z scores and summed. The alpha reliabilitY score for this scale was .56. Scale scores were normalized and transformed into ZP scores. The correlation between the child antisocial behavior scale and the child impulsiveness scale measure was .60. Although this is a substantial correlation, 64% of the variance is not shared, leaving open the possibility that the findings on child impulsiveness could differ from those for antisocial behavior. In fact, we expect that impulsive corporal punishment will be more strongly related to child impulsiveness than to antisocial behavior because that relationship could reflect modeling, which is a more direct linking process than the processes which might bring about a relationship with antisocial behavior, such as anger and resentment.
Control variables. The analyses controlled for five characteristics of the families and the children that might influence the relationship between spanking and child behavior problems: the mother's nurturance, the age of the child, the child's sex, the family's socioeconomic status, and the child's level of problem behavior. The interaction or moderating effect of these five variables was also tested using analysis of covariance. The tests for interaction provide data on whether the effect of spanking is different when one of these five variables is present or absent, or low versus high. An example of a test for an interaction effect is the analysis of the widely held belief that spanking is not harmful when it is done by loving parents. We tested that theory in the previous chapter and in the chapter on the link between spanking and risky sex in adulthood (Chapter 9) and found that the harmful side effects of spanking were present even for children with warm and supportive mothers. The implications for the relation of spanking to crime and violence in society are discussed in Chapter 19.
The Question of Causality
In the previous chapter, we pointed out that if a study finds a correlation between spanking and child behavior, this could mean that the spanking is affecting the child's behavior or that the child's behavior is eliciting spanking. We believe that both are true. That is, misbehavior can lead parents to spank, but if parents do spank, the longitudinal studies, such as the studies on child antisocial behavior (Chapter 6), mental ability (Chapter 10), adult crime (Chapter 15), and several other studies summarized in Chapter 19, have found that spanking tended to make children's behavior worse. We can draw that conclusion because the longitudinal studies included information on the child's behavior at the time of the spanking and then two years later, making it possible to determine whether the problem behavior decreased or increased among children who were spanked. All of the longitudinal studies conducted to date have found that, on average, spanking makes children's behavior worse.
The study described in this chapter did not have follow-up data on the children. However, we were able to take other steps to control for the level of misbehavior that led the parents to spank. We developed a scale to measure the extent to which the mother used discipline methods other than spanking, such as deprivation of privileges, explaining, and time-out. Use of this scale is based on the assumption that parents would not engage in these disciplinary interventions if there were no misbehavior (as perceived by the parent). Therefore, it is plausible to assume that the frequency of these disciplinary interventions reflects the extent of the child's misbehavior. To the extent that this is correct, the nonviolent interventions scale controlled for the misbehavior that led to the corporal punishment.
Prevalence of Spanking and Impulsive Spanking
Prevalence of Spanking
Consistent with all other studies, such as the one on the use of corporal punishment in the United States (Chapter 2), we found that the younger the child, the larger the proportion of mothers who hit their child during the previous six months:
59.3% of mothers of children aged 2 to 4
45.8% of mothers of children aged 5 to 9
20.5% of mothers of children aged 10 to 12
14.4% ofmothers of children aged 13 to 14
These are high rates, but they are much lower than those found for other representative samples of American children (see Chapter 2 and Giles-Sims et al., 1995; Straus, 2001a). Part of the reason for the lower rate of spanking may be because this study asked about spanking in the previous six months, as compared with the previous year for the other studies. It may also reflect the fact that the study was done in the state of Minnesota, which is a state that has long had a social and cultural climate that is favorable for children. Ever since the Annie E. Casey Foundation started annual publications that compared states on 10 indicators of child well-being, Minnesota has been in the top group. In the 2006 ranking, Minnesota ranked fourth in the nation (Annie E. Casey Foundation, 2006). Minnesota is also a state with a high average level of education, and it is the only state in the United States that has required every county to collect a tax to pay for parenting education. Perhaps most directly related to the relatively low rate of spanking for the mothers in this study is that one halflived in a city that had a Positive Parenting community-wide program to end the use of spanking sponsored by the Minnesota Cooperative Extension Service.
Relation of Spanking and Impulsive Spanking to Child's Antisocial Behavior
Chart 7.1 shows that the more spanking by the mother, the greater the antisocial behavior by the child. This is the same result as was shown in the previous chapter for children in a national sample. However, this study goes beyond the previous chapter in two important ways. The first advance is that it provides data on impulsive spanking. Chart 7.2 shows that impulsive spanking is even more strongly related to antisocial behavior than is spanking in general. The analyses of covariance used to obtain the results in these two charts are in Straus (1998, Tables 1 and 2).
The second advance over the study in the previous chapter is that it identified children who were never spanked. This enabled an important issue to be investigated for the first time in any study of spanking. This is whether there is a difference between children whose parents spanked only very rarely and those who did not spank at all. Both Chart 7.1 and 7.2 show that the neverspanked group at the left side of the chart had much less antisocial behavior than any other group, including the second group: children whose mothers did not spank at all in the past six months but had spanked previously. Three of the other five variables we examined were also significantly related to antisocial behavior: (1) the more nurturing the mother, the lower the child's antisocial behavior, (2) girls had lower antisocial behavior scores than boys, and (3) the more nonviolent discipline, the higher the antisocial behavior. This is probably because nonviolent discipline like spanking tends to occur in response to the amount of antisocial behavior.
Chart 7.1 The More Spanking, the More Antisocial Behavior
Context Effects
The analysis for this study was done in a way that investigated whether the effect of spanking depends on whether the spanking was impulsive. The results showed that:
* The relation of spanking to antisocial behavior was greatest for the children whose mothers were most impulsive in using spanking-those who spanked impulsively one half or more of the time (upper line of Chart 7.3).
* The children in the never-spanked group had the lowest antisocial behavior. Thus, any amount of spanking, even when it occurred prior to the past six months, was associated with greater antisocial behavior than that by children of mothers who never used spanking.
* Among mothers who reported only rare impulsive spanking (dotted line in center of Chart 7.3), all five comparisons with children who never experienced spanking showed that the never-spanked children had lower antisocial behavior scores. This means that when there was any impulsive spanking in the past, even though there was none in the past six months, or when there was even occasional impulsive spanking, it is associated with more antisocial behavior than by children of mothers who never spanked.
* The solid line at the bottom of Chart 7.3 is for children of mothers who reported no impulsive spanking. This line shows that, even when there was no impulsive spanking, spanking is associated with more antisocial behavior. The bottom line in Chart 7.3 shows a decrease in antisocial behavior for the two highest levels of spanking. However, because mothers who never spank impulsively also are infrequent spankers, there are very few children in those two categories. As a consequence, the seeming decrease is not statistically dependable-see footnote to Table 1 in Straus (1998), and we do not think any importance should be attached to that seeming decrease with more spanking.
Chart 7. 2 The More Impulsive Spanking, the More Antisocial Behavior
We also investigated whether the effect of spanking and impulsive spanking depends on one or more of the five other variables listed in the Hypotheses section. The only one of these five that made a difference was the amount of nurturance provided by the mother. With the exception of one data point, within each level of nurturance, the more spanking and the more impulsive the spanking, the higher the average antisocial behavior. Thus, although children of nurturing mothers had lower antisocial behavior scores, spanking still had an important adverse effect.
Chart 7.3 Spanking Is Most Strongly Related to Child Antisocial Behavior When Spanking Is Impulsive
*Means adjusted for mother's impulsive spanking, mother's nurturance, mother's non-spanking interventions, child's age, child's sex, and family socioeconomic status.
Child's Impulsive Behavior
The three previous charts are about antisocial behavior by the child. In this section, the focus is on impulsive behavior by the child. Chart 7.4 shows that the more spanking, the more impulsive the behavior of the child. Chart 7.5 shows that the more impulsive the spanking, the more impulsive the child. For impulsive spanking, there is an almost one-to-one increase in child impulsiveness as the mother's impulsive spanking increases, and the differences are large. The dashed line at the top of Chart 7.6 shows that the relation of spanking to impulsive behavior by the child is greatest when mothers use spanking impulsively half or more of the time. For children of these mothers, even one instance of spanking in the prior six months was strongly associated with children being much more impulsive than among children who did not experience spanking during the six month referent period or who were never spanked. Among mothers who only rarely were impulsive when spanking, the dotted line in the center of Chart 7.6 shows that as spanking increases, there is a step-by-step increase in child's antisocial behavior. For the most part, any spanking, past or present, was associated with more child impulsiveness than for children in the never-spanked group.
Chart 7.4 The More Spanking, the More Impulsive the Child
The relation of spanking to impulsiveness of children was weakest, but still present for children of mothers who spanked nonimpulsively (solid line at the bottom of Chart 7 .6). As the frequency of spanking increased, impulsiveness increased, but only up to the "twice in the last six months category" of spanking, and then decreased. These decreases were not statistically dependable because there were very few cases in those two groups of children and are best regarded as chance occurrences.
Do the other five variables affect the relation of spanking to impulsive behavior by the child? The statistical tests found that the sex of the child and the socioeconomic status of the family made a difference in the relation of spanking to children's impulsiveness. There was a stronger relation between spanking and impulsive behavior by boys than by girls. Although the statistical analysis showed a significant interaction of family socioeconomic status with spanking, we were unable to identify a meaningful difference between low and high socioeconomic status families in the relation of spanking to impulsive behavior by children.
Chart 7.5 The More Impulsive Spanking, the More Impulsive the Child
Summary and Conclusions
This study found that the more spanking is used, and the more impulsive it is, the more likely children are to be impulsive and antisocial, even after controlling for five other variables that could influence the effects of spanking. These results cast serious doubt on the recommendation of a conference of pediatric and other child behavior specialists that endorsed the use of spanking with children age 2 to 6, if done by loving parents (Friedman & Schonberg, 1996a). If this view were correct, we should have found that spanking was related to child misbehavior only among older children, and only among children with mothers who were less nurturing. Instead, the results of this study show that the tendency for spanking to be associated with more antisocial behavior and impulsiveness by the child applies to all age groups, regardless of the level of maternal nurturance---see Straus and Mouradian (1998) for tables giving the detailed results of the analyses of covariance.
Although the relationship between spanking and child behavior problems is weaker for children of mothers who were not impulsive spankers, it was strong enough for there to be a statistically dependable relationship between spanking and child behavior problems, even when spanking is not done impulsively.
Chart 7. 6 Spanking Is Most Strongly Related to Child Impulsiveness When Spanking Is Impulsive
*Means adjusted for mother's impulsive spanking, mother's non-spanking interventions, child's age, child's sex, and family socioeconomic status.
Limitations and Strengths of the Study
One type of limitation is the fact that this study measured only two aspects of spanking: frequency and impulsiveness, and there are a number of other aspects needed for a complete understanding of spanking. For example, since conducting the study in this chapter, we developed the Dimensions of Discipline Inventory (Straus & Fauchier, 2011), which includes measures such as the degree to which explanation and support accompany spanking, as well as measures of eight other methods of correcting misbehavior.
The most important limitation of this study is that the findings are based on cross-sectional, rather than longitudinal or experimental, data. We attempted to take into account the fact that spanking is typically a response to child misbehavior by using a scale to measure nonviolent discipline as a proxy for child misbehavior. The results showed that spanking, and especially impulsive spanking, was associated with more child antisocial behavior and impulsiveness regardless of the level of misbehavior that led to the spanking. We interpret this as evidence that when spanking is used in addition to other disciplinary strategies, it tends to make things worse. This interpretation is strengthened by the findings from the longitudinal studies on child behavior (Chapter 6), cognitive ability (Chapter 10), crime in adulthood (Chapter 15), and other longitudinal studies summarized in Chapter 19 and Straus (2001c). These studies found that the more parents respond to misbehavior at Time 1 by spanking, the greater the increase in undesirable behavior from Time 1 to Time 2.
This study also has two unique strengths. The first is that, rather than treating spanking as present or absent, or differentiating only on the basis of the chronicity of spanking, the study took into account another of the many other dimensions of spanking-whether it was done impulsively.
A second unique feature of this study is that it identified children who, at least according to the mother, had never experienced spanking. It is particularly important to learn about children who have not been spanked because it is so widely believed that spanking is sometimes necessary (see Chapter 18) and that if parents are not prepared to spank as a backup when other methods have not worked, the child will grow up being out of control. Thus, the inclusion of a never-spanked group in this study begins to fill a void in the literature. It can be considered a starting point for addressing individual and societal concerns about the effects of ending all spanking. With these strengths and weaknesses in mind, what can be concluded from this study?
The findings suggest that spanking and impulsive spanking increase the risk of children developing a pattern of impulsive and antisocial behavior. Moreover, we believe that spanking, especially impulsive spanking, are part of the etiology of the high level of violence and crime in society because about one half of parents used spanking impulsively at least some ofthe time and because the long-term risks associated with corporal punishment have been demonstrated by the longitudinal studies in this book and the many other longitudinal studies listed in Chapter 19.
At present, parents of toddlers who do not spank make up a small but growing portion of the population (see Chapters 2 and 17). What will happen when, as seems likely, no-spanking becomes more widespread? That is likely to happen because of the changes in society described in the concluding chapter. By 2011, 30 nations have chosen to follow the Swedish example and the recommendations of the European Union and United Nations by enacting laws against spanking by parents. Will it produce a generation of antisocial and out of control children as believed by those who think that spanking is sometimes necessary? The results in this chapter and the experience in Sweden described in Chapters 19 and 20 suggest that the opposite is more likely. The decrease in youth crime and drug abuse in Sweden since passage of the no-spanking law is consistent with that implication.
8 The Child-to-Mother Bond and Delinquency
There are a number of ironic aspects of spanking. One of the most frequent occurs when a parent spanks a child for hitting another child-which they are more likely to do than for most other misbehaviors. A national survey of 1,012 parents asked what punishment these parents would consider appropriate for three types of misbehavior. For ignoring a request to clean up their room 9% thought spanking was appropriate, for stealing it was 27%, and for deliberately hurting another child 41% approved of spanking (Kane-Parsons and Associates, 1987; see also Sears et al., 1957). The irony is that when parents hit a child to correct the child's hitting, they are inadvertently modeling the very behavior they want the child to avoid. This is one of reasons for the almost complete consistency between studies like those in Part IV and by many longitudinal studies documented in Chapter 19, which have found that the more spanking used by parents, the greater the physical aggression by the child.
Kimberly A. Hill is the coauthor of this chapter.
Studies that have found that spanking is associated with an increased probability of property crime as a child or adult (Grogan-Kaylor, 2004) are less easily understood than is the link between spanking and aggression. Except for the relatively rare parents who themselves engage in criminal acts, modeling is probably not an important part of the process linking spanking and delinquency. If it is not modeling, what social and psychological processes could account for the link between spanking and delinquency? The study in this chapter was designed to provide data on one of several possible processes. This is the possibility that, when parents spank, it tends to undermine the bond between child and parent, and the weakened bond in turn makes the child less receptive to parental guidance and more vulnerable to peer pressure and other factors that increase the probability of delinquent acts. In this chapter, we address the following questions:
* When spanking is used, does it weaken the bond between a child and his or her parents?
* If the parents who spank show warmth and support, does this avoid weakening the child-to-parent bond?
* Do children whose parents correct misbehavior by spanking engage in less or more delinquent behaviors?
* If spanking is done by warm and loving parents, does it override the tendency for spanking to be related to an increased probability of delinquency?
Does Spanking "Teach Him a Lesson"?
The caning of an American teenager in Singapore in the 1990s (Elliott, 1994) received national headlines and public discussion. It brought to the surface the widespread belief in the United States that spanking reduces delinquency. Many letters to the editor and comments on talk shows argued that if parents would go back to the good old-fashioned paddle, there would be less delinquency. The research evidence, however, suggests that the opposite is more likely. For example, in this book:
* Chapter 6 showed that the more spanking used by parents, the greater the probability that the child's antisocial behavior would become worse. However, tliis study was for young children and for antisocial behavior rather than statutory delinquency. * The 'five studies in Part IV all found that spanking was associated with an increased probability of crime later in life as an adult.
A few of the other studies include:
* A longitudinal study by Grogan-Kaylor (2005), which found that, after controlling for six other variables such as family income and cognitive stimulation by the mother, spanking was associated with an increase in antisocial behavior by the child, and that this applied across racial and ethnic groups.
* Welsh (1978) found that almost all the delinquent children in his sample had experienced a great deal of spanking by their parents.
* A study that followed up a large sample of boys from a high-risk area for 35 years found that spanking was associated with an increased probability of conviction of a serious crime as an adult (McCord, 1997).
* Straus (200 la, p. 1 08) found that, even after controlling for a number of other family characteristics, including socioeconomic status and whether there was violence between the parents, the more spanking the parent reported using, the greater the probability of the child being delinquent.
* Gove and Crutchfield (1982) found a number of parenting variables to be related to delinquency, including spanking, and that after controlling for all other variables, spanking remained significantly related to delinquent behavior.
* Conger (1976) and Rankin and Wells (1990) found that high parental punishment is associated with higher levels of juvenile delinquency. However, their measures of parental punitiveness included non-corporal punishment and yelling at the child. As a consequence, they do not provide direct evidence that spanking per se is related to delinquency.
* Simons, Lin, and Gordon (1998) found that adolescent boys who had experienced spanking were more likely to hit a dating partner, even after controlling for a number of factors such as parental involvement and support.
These are just a few of the 88 studies reviewed by Gershoff (2002). Fortynine of the studies tested the hypothesis that corporal punishment is associated with an increased probability of physical violence and antisocial or criminal behavior as a child and as an adult; of these, 97% found that corporal punishment was associated with an increased probability of physical violence, antisocial behavior, and crime. This degree of consistency of results is extremely rare in any field of science, and perhaps even rarer in child development.
Nevertheless, many of these studies have important limitations. In additi~n, there are a few well-designed studies that found no relationship between spanking and delinquency (e.g., Agnew, 1993; Simons et al., 1994). Therefore, one purpose of this chapter was to reexamine this issue by testing the hypothesis that spanking is associated with delinquency. The main purpose, however, was to test the theory that one of the reasons spanking increases the probability of delinquency is because spanking undermines the bond between children and parents.
The Child-to-Parent Bond and Delinquency
Spanking and other forms of corporal punishment may stop a specific undesired behavior at the moment, but in the long run, it may teach children to avoid particular behaviors only when they are in the presence of a parent or other authority figure who can impose a penalty, or some other circumstance where the probability of punishment is high. Obviously, that leaves a great deal of opportunity to engage in delinquent and criminal acts. As children grow older, they become less subject to parental observation and, in adolescence, they are usually too big to control by physical force. When this happens, avoiding delinquent behavior is largely dependent on the child having internalized behavioral standards (i.e., on the development of conscience).
Two aspects of spanking are likely to interfere with development of such internalized standards. First, as just noted, spanking places the focus on behaving correctly as a means to avoid punishment, rather than on behaving correctly in order to follow principles such as honesty, courtesy, compassion, prudence, and responsibility-the cognitions and emotions that together make up conscience. When parents spank to teach these principles, children do learn them, but we believe it is despite the spanking, not because of it. There is also a greater risk that some or all of what the parent wants to teach will be inadequately learned because fear and anger interfere with learning (see Chapter 10). The same moral instruction without spanking is likely to be more effective in facilitating a well-developed conscience.
A second aspect of spanking that interferes with the development of conscience is the tendency of spanking to weaken the bond between child and parent (Azrin & Holz, 1966; Barnett, Kidwell, & Leung, 1998; Bugental, Johnston, New, & Silvester, 1998; Mulvaney & Mebert, 2010; Parke, 1969). A strong child-to-parent bond is important because when there is a bond of affection with the parent, children are more likely to accept parental rules, restrictions, and moral standards as their own. These ideas are central to the social control theory of delinquency (Akers & Sellers, 2008; Hirschi, 1969). This theory links delinquent behavior to a weak social bond with parents and with other key institutions of society, such as the church, schools, and the workplace. The central idea of the social control theory is that law-abiding behavior depends, to a considerable extent, on the bond between a child and persons and organizations that represent the moral standards of society. A bond with parents refers to ties of affection and respect that children have for a parent. Hirschi argued that the bond or attachment to parents is the most important variable insulating a child against deviant behavior. It enables the child to internalize the rules for behavior and develop a conscience. The stronger the bond with the parents, the more likely children are to rely on internalized standards learned from parents when tempted to engage in a delinquent behavior (Hirschi, 1969, p. 86).
Hirschi's research and many empirical studies since then have found a link between a weak parent-child bond and juvenile delinquency. To take just four of these studies, Hindelang (1973) replicated Hirschi's study using rural adolescents 'and found that a weak bond to parents is also significantly related to juvenile delinquency in that environment. Wiatrowski and Anderson (1987) found that the weaker the bond with parents, the higher the probability of delinquency. Rankin and Kern (1994) found that a bond to both parents provides more insulation from delinquency than does a bond with one parent. Eamon and Mulder (2005) studied 420 Hispanic children and found that each increase of one point on the 15-point scale of parent-child attachment was associated with a 9% decrease in the probability of the child being in the high antisocial behavior category.
The relationship between the child-to-parent bond and delinquency is likely to be moderated by other variables, that is, to be contingent on other variables. Seydlitz ( 1990) found that the relation of some aspects of the bond with parents to juvenile delinquency depends on the age and gender of the adolescent. Several studies have also found an indirect relationship between bonds with parents and delinquency. For example, Agnew (1993) found that a weak bond with parents is associated with an increase in anger or frustration and an increased likelihood of association with delinquent peers, which in tum increases the likelihood of juvenile delinquency. Warr (1993) found that part of the reason for the link between bonding and delinquency occurs because a close child-toparent bond inhibits forming delinquent friendships. Marcos and Bahr (1988) found both a direct link between bonding and drug abuse and an indirect link through the relation of bonding to greater educational attainment, conventional values, inner containment (the ability to withstand pressure from peers), and religious attachment.
Although there is a large body of evidence showing a link between weak child-to-parent bonds and delinquency, there does not seem to have been a study investigating whether spanking weakens the child-to-parent bond. In fact, Hirschi himself doubts that it does (Hirschi & Gottfredson, 2005). This study was designed to help fill that gap in the chain,Of evidence linking spanking and delinquency.
Spanking and the Child-to-Parent Bond
If spanking is typically carried out in the hope of raising a well-behaved and lawabiding child, what could explain why so many studies have found that spanking is instead associated with delinquency? Given the many studies just reviewed showing that a weak child-to-parent bond is associated with delinquency, it is plausible to suggest that one of the reasons for the relationship between spanking and delinquency is that spanking undermines the bond between parent and child.
If spanking and physical abuse are conceptualized as low and high points on a continuum of violence against children, a link between spanking and attachment to parents can be inferred from a number of studies that show that physical abuse has an adverse effect on child-to-parent attachment.
* Lyons-Ruth, Connell, Zoll, and Stahl (1987) compared 10 maltreated infants, 18 non-maltreated high-risk infants, and 28 matched low-income controls on the Ainsworth Strange Situation test and found less adequate infant attachment among the maltreated children. (The Strange Situation experiment examines the attachment between very young children and parents.)
* Crittenden (1985) studied 73 mother-infant dyads referred by the welfare department and found that infants who were abused showed an avoidant or ambivalent pattern of attachment, and adequately treated infants showed a secure pattern of attachment.
* Kinard's (1980) review of the literature on child abuse and emotional problems identified other studies showing that child abuse is associated with weak attachment.
Spanking children may be at the low end of the same continuum of parent-to-child violence as physical abuse, but it is also different. Spanking and other forms of corporal punishment are legal and socially approved parent behaviors and, in many communities, are expected parent behaviors (see Chapter 17 and Alvy, 1987; Carson, 1986; Walsh, 2002). Walsh, for example, found that in the previous six months, 61% of a sample of 1,003 mothers of children age 2 to 4 in two Minnesota counties had been advised to spank. Moreover, the children themselves accept the right of parents to spank-but they also resent it (Willow & Hyder, 1998). Similarly, when Cohn and Straus (Straus & Donnelly, 2001a, p. 149) asked 270 students at two New England colleges for their reactions to "the first time you can remember being hit by one of your parents" and also the most recent incident, 42% checked that they "hated him or her" from a list of reactions. Similar views were expressed by the 8- to 17-yearold children in a study by Saunders and Goddard (2010). Rohner, Kean, and Cournoyer's ( 1991) study of children in St. Kitts is particularly relevant because the right and the obligation of parents to spank is strongly embedded in the culture of that society, and many of the children shared this cultural belief. Despite that, research by Rohner and colleagues of children in St. Kitts, West Indies (Rohner et al., 1991) and research in a poor biracial Southern community in the United States (Rohner, Bourque, & Elordi, 1996) found that spanking was asso~ ciated with feelings of rejection regardless of whether the children accepted the cultural belief that spanking is appropriate. Feeling rejected by a parent is close enough to a weak child-to-parent bond to suggest that is one of the processes explaining why spanking undermines the child-to-parent bond.
We believe that each spanking chips away at the bond between parent and child. Although a single instance will rarely undermine the bond, Chapters 2 and 10 show that a single instance, ever in a child's lifetime, is very rare. The typical pattern is about 3 times a week for toddlers, and for one half of American children it continues, although with lower frequency, for another six years. These repeated chips cumulatively weaken the strength of a parent-child bond. That does not mean that all children who are spanked have a weak bond with their parents, only that a larger percentage of these children than non-spanked children will have a weak bond (see the section in Chapter 1 on "Risk Factors: The Real Meaning of Social Science Results").
Although the studies we reviewed provide a plausible basis for the theory that spanking is related to delinquency, because it undermines the bond between children and parents, the evidence is far from definitive. For example, Baurnrind et al. (2002) argue that what seems to be the effect of spanking is really the result of the combination of a lack of parental warmth and nurturance combined with very frequent or severe spanking. The effects of spanking may also depend on the age of the child and the socioeconomic status of the family. For example, as noted previously, it is widely believed that spanking harms only older children (Friedman & Schonberg, 1996a). The research described below, although it is also not definitive, can at least rule out these plausible rival interpretations.
Hypothesis
The studies and theories just reviewed led to the hypothesis that .spanking is associated with delinquency because spanking is associated with a reduction in the bond between child and parent, and the weakened bond is in tum associated with an increased level of delinquency.
Sample and Measures
The data for this chapter come from interviews with the random sample of 1,003 mothers of children age 2 to 14 in two small cities and their surrounding counties in Minnesota described in the previous chapter and in Straus (1998). The analysis of the link between spanking and bonding is based on all 915 children for whom complete data was available. The part of the study that examined the link between spanking and delinquency used only the 411 children age 10 to 14 because the concept of delinquency is primarily appropriate for children this age or older.
Measures
Spanking. The mothers were asked how often they· spanked, slapped, or hit their child in the past six months: none, once, twice, 3 to 5 times, 6 to 10 times, 11 to 20 times, or more than 20 times, and a parallel question on how often their husband or partner had spanked, slapped, or hit the child in the past six months. We summed these two items to obtain the measure of spanking.
Child-to-mother bond. We used a scale consisting of two items believed to reflect the aspect of child-to-parent bond that Hirschi identifies as effectual identification. The mothers were asked how often in the past six months their child showed affection or closeness toward them, and how often their child did things to please them. The response categories were: none = 0, rarely = 1, sometimes= 2, and frequently= 3. These two items were combined to form the child-to-mother bond scale. It should be noted that this scale is based on interviews with mothers. Thus, the scale measures the mother's perception of the child's behavior. This may be a less adequate measure than one based on the child's report. If so, it would mean that the results to be presented probably underestimate the relationship between spanking and the child-to-mother bond.
Delinquency. We asked the mother about the extent to which the child was rebellious; cruel or mean to other kids, a bully; cruel, mean to, or insulting to her (the parent); damaging or destructive to things; hitting other kids; hitting her or other adults; having discipline problems in school; stealing money or something else; drinking or using drugs; hanging out with kids who get into trouble; having school problems; and getting into trouble with kids he or she hangs around with. The response categories were: never= 0, rarely= 1, sometimes= 2, and frequently= 3. The scale is the sum of the scores for these 12 behaviors and could range from 0 (never for alll2) to 36 (frequently for all12). A limitation of this measure is that mothers do not know about all delinquent behavior. As a consequence, the results may underestimate the relationship between spanking and delinquency. The alpha coefficient of reliability for the delinquency scale is .79.
Support and nurturance provided by the mother. Many other things could affect the relation of the child-to-mother bond and delinquency and some of these were controlled when we tested the hypothesis that spanking weakens the bond and that a weaker bond is part of the explanation for the link between spanking and delinquency. One of the most important of these other variables is the amount of support and nurturance provided by the mother. To investigate the role of nurturance, we used a scale composed of three questions about the mother's behavior toward her child in the past six months, whether she: comforted and helped him or her when the child had some kind of problem; hugged or kissed the child or did something else to show love; or talked to the child about things that bothered the child. The response categories were: 0 = never, 1 = rarely, 2 =about one half the time, 3 =usually, and 4 =always or almost always. The responses were summed to create the nurturance scale, with the possible range of scores between 0 and 12. Because ofthe small number of cases with scores below 7, scores of 4 through 7 were combined into one category.
Control variables. The analysis also included three demographic variables: gender, age of child, and socioeconomic status. Each of these variables is known to be related to spanking and delinquency, so leaving them uncontrolled could result in spurious findings. We also did an analysis to find out if the relation between spanking and delinquency is different among boys than among girls.
Spanking, Bonding, and Delinquency
Chart 8.1 shows the tendency for spanking to be associated with higher delinquency that has been found in many previous studies. Considering that spanking is used to correct misbehavior and, at the same time, intended to encourage correct behavior, the question of why it seems to do the opposite needs to be answered.
Chart 8.1 The More Spanking, the More Delinquency, Even after Controlling for Nurturance, Age, Gender of the Child, and Socioeconomic Status
Chart 8.2 provides a start to answering that question. Reading from left to right, it shows that the more spanking by the mothers in this study, the weaker the bond of the child to the mother. Chart 8.2 begins to provide part of the explanation of why spanking is associated with more rather than less delinquency. The results in this chart suggest that it is because spanking undermines a bond to parents that, as we explained earlier, is extremely important for moral development and for avoiding criminal behavior.
Defenders of spanking such as Baumrind and Larzelere agree with the importance of the bond between child and parents for the development of conscience. They do not, however, think that spanking itself undermines the bond between parent and child. They propose that the weak bond is because some of the parents who spank are also cold and harsh parents. According to them, the inclusion of these parents with all other parents who spank explains the relationship. Chart 8.3 shows that the more nurturing the mother, the stronger the child-to-mother bond. As a consequence, to conclude that there is something about spanking itself that explains the link between spanking and a weak child-to-parent bond requires controlling for the amount of nurturance by the parents, and this is what we did, as explained in the next section.
Chart 8.2 · The More Spanking, the Weaker the Child-to-Mother Bond
Chart 8.3 The More Nurturing the Mother, the Stronger the Child-to-Mother Bond*
*Controlling for spanking, age, gender of the child, and socioeconomic status.
Does a Weak Child-to-Parent Bond Help Explain the Link between Spanking and Delinquency?
We used a statistical technique called path analysis to test the theory that a weakened child-to-parent bond resulting from spanking is one of the processes that explain why so many studies have found that spanking is a risk factor for delinquency. Chart 8.4 summarizes the results of the path analysis we conducted to test that theory (see Appendix for the details). Each of the arrows or paths in Chart 8.4 indicates a relationship that was hypothesized and was found to apply to this sample of children. An important aspect of Chart 8.4 to keep in mind is that the paths represent relationships that hold after controlling for all the other variables in the chart. That is, each is over and above the effects on delinquency of all four other variables in the analysis.
Chart 8.4 Paths to Delinquency
Two paths bear most directly on the theory that a weakened child-to-parent bond is one of the reasons spanking is associated with an increased probability of delinquency. The first is the path with a coefficient of -.13 going from spanking in the upper left to child-to-mother bond in the middle. The second is the path with a coefficient of -.12 going from child-to-mother bond to delinquency. The negative coefficient of -.13 in the path from spanking to child-to-mother bond shows that, among the children in this study, the more often they were hit by their parents, the weaker the child-to-mother bond. The -.12 path from child-to-mother bond to delinquency means that the stronger the child-to-mother bond, the less delinquency. In short, as hypothesized, spanking weakens the child-to-mother bond, and because the bond has been weakened, it cannot have the protective effect that it would have otherwise.
The path at the top of Chart 8.4 with a coefficient of .17 shows that spanking, in and of itself, has a direct relation to delinquency, even when controlling for all four of the other variables. This means that, in addition to the five variables we investigated, there are other processes that are set in motion by spanking that make children more vulnerable to delinquency, which we did not include in this study. For example, we found that spanking is associated with an increased chance that the child will be impulsive (Chapter 7), and that spanking is related to lower mental ability and lower school achievement (Chapter 10). Impulsiveness and doing poorly in school both increase the risk of delinquency (Farrington & Welsh, 2006). In addition, spanking is related to low self-esteem (Coopersmith, 1967), more feelings of alienation (Straus & Gimpel, 2001), and a higher level of anger and rage (Tsang, 1995). These are plausible intervening processes because each can increase the probability of delinquency. They were not included in the statistical analysis for this chapter because those variables were not available for this particular sample. They may be what accounts for the upper path in Chart 8.4.
The path with a value of .32 from nurturance to child-to-mother bond shows that the more nurturing the mother, the stronger the bond. It is the largest path in this chart, which nieans that nurturing behavior by the mother has more influence on the child-to-mother bond than any of the other variables. The -.17 path from nurturance to delinquency shows that the more nurturing the mother, the less delinquency. These results are consistent with many other studies that have found that warm and positive parenting is associated with a lower risk of delinquency.
The path from being a girl to delinquency (-.17) at the bottom of Chart 8.4 is consistent with many studies showing that girls are much less likely than boys to engage in delinquent behavior. Finally, the -.20 path from child's age to child-to-mother bond shows that older children have a weaker bond to their mothers than do younger children.
Does the Relation of Spanking to Bonding Depend on Other Variables?
It is widely believed that if spanking is used by warm and nurturing parents and is restricted to preschool-age children, it has no harmful side effects (Friedman & Schonberg, 1996a). One of the advantages of a path analysis such as in Chart 8.4 is that each path represents a result that is in addition to the effect of all the other factors. As a consequence, we can say that both spanking and parental nurturance have their own separate relationship to delinquency; or put somewhat differently, spanking is related to delinquency regardless of the age of the child and the nurturance of the parent (in this study, the mother).
Because the question of whether the adverse effects of spanking are really just a reflection of insufficient parental warmth and support, we also tested the interaction of spanking with parental nurturance. This determines if the relation of spanking to the child-to-mother bond and delinquency depends on the extent to which the mother is warm and loving. The analysis did not find a contingent effect that is statistically dependable. Keeping the limitations of cross-sectional data in mind, this indicates that, even when used by highly supportive parents, spanking is linked to a weakened bond and higher delinquency. This is an extremely important result because it contradicts the almost universal belief in many societies that if spanking is used by warm and loving parents, it does not have harmful side effects.
Similarly, we found that the relationship of spanking to child-to-mother bond was parallel for boys and girls, but the relationship was stronger for boys. For both boys and girls, the more spanking, the weaker the child-to-mother bond. This is in addition to the finding that the bond of boys to mothers is, on average, lower than the bond of girls to mothers. The combination is likely to be part of the explanation for the higher delinquency rate among boys.
Summary and Conclusions
The results show that:
* The more spanking used by mothers; the lower the child-to-mother bond. The results in Chapter 9 for an entirely different sample show the same tendency for spanking to weaken the bond between children and parents.
* The tendency for spanking to be associated with a weak child-to-mother bond applies to children whose mothers were high in nurturance as well as those with low-nurturance mothers.
* The lowered child-to-mother bond associated with spanking is one of the processes linking spanking and delinquency.
* The analysis controlled for nurturance by the mother. Thus, it is unlikely that the underlying link between spanking and delinquency is explained by lack of nurturance.
One of the important results of this study is the parallel, but opposite effects of spanking by mothers and nurturance by mothers. Both spanking and parental nurturance have an indirect effect on delinquency. Both are related to the bond between the children and their mothers in this sample, but in opposite directions. Spanking is associated with a weaker bond, and nurturance with a stronger bond. So, by weakening the bond, spanking is associated with an increased probability of delinquency, whereas by strengthening the child-to-mother bond, nurturance is associated with a decrease in the probability of delinquency. In addition to these indirect effects on delinquency, spanking and nurturance have about the same direct effect on delinquency (the path coefficients are .17 for both), but in the opposite direction.
The relationship between child-to-mother bond and delinquency, although significant, is not very large. This is consistent with the principle that delinquency determined in multiple ways, and therefore a single risk factor cannot have a strong relationship. To take a simple example, the data on child-to-parent bonds in this study included only the child-to-mother relationship. Based on Rankin and Kern (1994), who found that a bond to two parents provides more insulation from delinquent behavior than attachment to one parent, it is likely that there would be a stronger effect if the study had also been able to include data of the child-to-father bond.
Some Cautions
The results should be interpreted with caution because they are based on crosssectional data and depend entirely on the report of the mothers who were interviewed. As noted, there is also no information on the bond of the children in the study to their fathers. The limitation of cross-sectional data is an especially important limitation because the relationship between spanking and delinquency could reflect the fact that parents use spanking in response to delinquency (i.e., that delinquency causes spanking). We believe that is what happens. However, we also believe that using spanking, although it may work in the immediate situation, is counterproductive in the long run. That is what was found by the longitudinal studies in the chapters on child behavior (Chapter 6), mental ability (Chapter 10), adult crime (Chapter 15), and several other longitudinal studies described in Chapter 19. These studies followed up children to examine the long-term effects of spanking and found that when parents use spanking to correct misbehavior, it boomerangs and is associated with a subsequent increase in problem behavior. As a consequence, we discuss the findings from that perspective and focus on the processes that make spanking counterproductive in the long run, even though spanking may work in the immediate situation.
Other Processes Linking Spanking and Delinquency
The focus of this chapter emphasizes the notion that one of the reasons why spanking is related to more, rather than less, delinquency is the tendency of spanking to undermine the child-to-parent bond, which in tum makes the child more vulnerable to delinquent influences. As pointed out previously, there are many other variables that are related to both child-to-parent bonds and to delinquency. For example, this chapter shows that there is a strong relationship between maternal nurturance and child-to-mother bonding. Thus, parenting strategies to insulate a child from delinquency must include far more than just not spanking. It must also include warm and nurturing interaction with the child. But this does not mean that the combination of warmth and avoiding spanking are sufficient. For example, studies of other aspects of the control theory of delinquency indicate the need to establish clear standards and rules, and to monitor the child's behavior and enforce these rules. Without clear rules and standards for behavior, children can interpret warmth and absence of spanking as parental approval for whatever a child does. In respect to physical aggressiveness by the child, for example, Sears et al. (1957) found that children of parents who avoided spanking were less aggressive to other children, but only if the parents also had clear rules about not hitting others.
It is also crucial to keep in mind that a strong bond to parents will only decrease delinquency if the parents exemplifY socially legitimate behavior. A strong child-to-parent bond can increase the probability of deviant behavior if the parents exemplifY antisocial and criminal behavior. This is illustrated by the research of Jensen and Brownfield (1983) who found that attachment to nondrug using parents decreases the likelihood of adolescent drug use, whereas attachment to drug-using parents increases the likelihood of the child's drug use. Similarly, Foshee and Bauman (1992) found that among the children of smokers, the stronger the attachment, the more likely the child was to smoke.
Finally, we want to stress that physical assaults in the form of spanking are only one mode of punitive child rearing. The Dimensions ofDiscipline Inventory (Straus & Fauchier, 2011 ), for example, measures three punitive methods of correcting misbehavior (corporal punishment, verbal/psychological aggression, and deprivation of privileges), five nonpunitive methods (diversion, explain/teach, ignore misbehavior, monitoring, reward), and one that combines punitive and nonpunitive methods of correction (penalty tasks and restorative behavior). Psychological attacks on a child in the form of yelling, nagging, scolding, swearing at the child, demeaning the child, or frequent grounding for long periods are as highly or more strongly correlated with delinquency than physical attacks on a child (Briere & Runtz, 1988; O'Hagan, 1993; Vissing, Straus, Gelles, & Harrop, 1991 ). One of the probable reasons for such fmdings is that these approaches, like spanking, undermine the bond of children to parents and, therefore, the ability of parents to supervise and influence the child.
Many factors affect the relationship between spanking and delinquency. The evidence in this chapter, however, (as well as in the next chapter), suggest that the tendency of spanking to weaken the child-to-parent bond is an important part of the process leading from spanking to an increased, rather than a decreased, probability of delinquency.
9 Spanking and Risky Sex
The consequences of unprotected sex, such as unwanted pregnancy and sexually transmitted diseases, may be greater for youth than for adults because these events can drastically alter a young person's entire life trajectory. This possibility strikes understandable fear in the hearts of parents and concern on the part of the nation. The actions taken based on those fears and concerns, however, can be counterproductive, such as when the promotion of sexual abstinence denies youth information and access to contraceptives (Kirby, 2002). In families, parental fear and anxiety over sexual activity may lead parents to lash out physically to control sexual behavior, such as slapping a child for engaging in activities that might have involved sex. This can be counterproductive because, as shown in the previous chapter, the side effects include weakening the childto-parent bond, and this can decrease the influence parents have and, therefore, increase rather than decrease the probability of risky sex-sexual behavior that puts a young person at greater risk for an unwanted pregnancy or a sexually transmitted disease.
Research on the etiology of risky sex has found that it is related to experiencing childhood sexual abuse (Messman-Moore, Walsh, & DiLillo, 2010) and childhood physical abuse (Elliott, Avery, Fishman, & Hoshiko, 2002). Elliott found that girls ages 14 to 17 who experienced physical violence from parents had a 3.5 times greater rate of risky sex. However, that study used a measure of violent parent-child interactions that included a broad range of violent acts in addition to corporal punishment. A 30-year follow-up study of abused children (Wilson & Widom, 2008) also found that abuse was related to risky sex. It is possible that the relationship between broad measures of physical abuse and risky sexual behavior also applies to spanking, even though spanking is a culturally approved form of parental violence.
Rose A. Medeiros is the coauthor of Study I in this chapter.
Although there has been a great deal of research on the side effects of spanking, we did not find any empirical studies that examined whether spanking was associated with an increased probability of risky sex. This chapter presents the results of two studies that did investigate whether there is a link between spanking and risky sex. Perhaps more important, both also investigated processes that might explain why spanking is related to risky sex. Thus, using these two separate studies, we addressed the following questions.
Questions Addressed
Study I: Risky Sex by High School Students
* How frequent is risky sex among a sample of high school students?
* Were the students who were spanked more likely to have risky sex?
* If spanking is linked to risky sex among high school students, is it because spanking is associated with one or more of the following:
• Weakened child-to-parent bond
• Reduced self-esteem
• Lower academic achievement
• Increased risk of sexual victimization
• Traditional gender roles
• Low self-control
• Approval of violence
Study 2: Unprotected Sex by University Students
* Are university students who were spanked in childhood more likely to have risky sex?
* Is spanking associated with a child who is low in self-control and high in violence approval?
* Are low self-control and high approval of violence part of the explanation of why spanking is associated with an increased probability of risky sex?
Study 1: Risky Sex by High School Students
We began the research in this chapter by drafting a causal diagram that included the variables and links just discussed, as well as a number of others. Then we looked for a data set that included as many of the variables of interest and related variables as possible. By a stroke of luck, we found such a data set in our own backyard. It was a study of students in the Durham, New Hampshire high school. This study was conducted at the request of parents and students to provide information that might help address the concerns about teen pregnancy and sexually transmitted diseases mentioned in the introduction to this chapter. The study was intended to provide information on a variety of student health-related behaviors (i.e., substance use, drinking, etc.) and serve as a baseline for future programs intended to encourage healthy behavior by students.
The sample consisted of 440 students in the 9th through 12th grades of a single high school in New Hampshire. This school is located in a university community and serves an almost all-White, high-average-education population. The students were surveyed during the 1992-1993 school year. The sample included equal numbers of males and females and approximately equal numbers of students from each grade level. Most students were 16 or 17 years old. A little over two thirds lived with both parents. Eleven percent lived with a parent and a stepparent. About one third of students said they had average grades of B+ or better, and about an additional third of students reported a grade average of B or B-. About one quarter of students received average grades of C or C+, and less than 10% received grades that averaged lower than a C.
The Child-to-Parent Bond
What especially drew us to this sample was that it included a measure of alienation from parents, which is the opposite end of the continuum from a close child-to-parent bond. If there is a link between spanking and risky sexual behaviors, it is not necessarily traceable to a single incident of being hit as a teenager. Parents who slap a child who has failed to come home at a reasonable hour probably had used spanking for years when the child was younger. As a consequence, as evidenced by the results in Chapter 8, the child-to-parent bond may have been weakened. If spanking interferes with the quality of parent-child relationships as was shown in that chapter, it suggests that spanking will be linked to an increased probability of children engaging in problematic behaviors, including risky sex. This is because a weakened bond between children and parents decreases the influence of parents in setting standards for many problematic behaviors (Akers & Sellers, 2008; Hirschi, 1969). The results of several studies are consistent with this theory because they find that a weak parent-to-child relationship is related to an increased probability of sexual intercourse during the teenage years (Danziger, 1995; Metzler, Noell, Biglan, Ary, & Smolkowski, 1994; Moore, 1998; Resnick et al., 1997) or a pregnancy at an early age (Danziger, 1995; Pick & Andrade Palos, 1995).
A weakened parent-to-child bond may also be linked to risky sex by less direct but still important processes. For example, as was shown in the previous chapter, spanking is associated with an increased probability of delinquency. Numerous studies have found that delinquent children tend to associate with delinquent peers (Hirschi, 1969; Matsuenda & Anderson, 1998), and other research shows that associating with delinquent peers increases sexual risk taking by adolescents (Metzler et al., 1994). Thus, spanking may also be linked to risky sex because spanking is associated with an increase in delinquent peers, which then may then lead to an increased likelihood of risky sexual behavior.
Measure of Risky Sex
Risky sex can be broadly defined as sexual activities that increase the likelihood of two life-altering events: pregnancy and contracting a sexually transmitted disease. Because of the negative impact of both sexually transmitted diseases and childbearing on adolescents' lives and development, any behavior that increases the likelihood of these events can be considered risky. Frequently used measures of risky sex include: age at first intercourse (Danziger, 1995; Murray, Zabin, Toledo-Dreves, & Luengo-Charath, 1998; Pick & Andrade Palos, 1995), number of sexual partners (Lucke, 1998; Metzler et al., 1994), and nonuse of contraception (including condoms; Kowaleski-lones & Mott, 1998; Lucke, 1998; Metzler et al., 1994; Pick & Andrade Palos, 1995). The items used to measure risky sex in this chapter are consistent with the literature on risky sex; this study does, however, include more items than was typical of studies in our literature review. Rather than relying on just one indicator of risky sex, nine behaviors were used to create a measure of risky sex.
* Ever had sexual intercourse
* Age at first sexual intercourse
* Number of sex partners
* Frequency of condom use
* Frequency of birth control pill use
* Frequency of the use of other contraception
* Number of times respondent has purchased condoms in the past year
* Number oftimes the respondent has had sex in the past year
* Whether or not the respondent has had or caused a pregnancy
The more of these nine indicators a student reported, the higher the score on the risky sex scale. (Further information on the scale is in the Appendix.)
Prevalence of Spanking
Just over two thirds of these high school students reported that their parents had spanked at some point in their life (see Appendix Table A9.1). This is a high percentage, but it is also a much lower percentage than the 90% to 98% found in Chapter 1 and those typically found by surveys of college students (Berger, Knutson, Mehm, & Perkins, 1987; Bryan & Freed, 1982; Deley, 1988; Graziano et al., 1992). The difference may stem from the fact that New England is the region with the lowest rates of use and approval of spanking (Flynn, 1994, 1996b), and the fact that the students are from a high school serving a university community with parents who, on average, tend to have a high level of education. Even allowing for these regional and community characteristics, because ages 2 to 4 are peak ages for spanking, it is likely that at least some of the almost one third of students who reported never experiencing spanking did in fact experience some spanking, but were too young at the time to recall the experience.
The peak age of spanking reported by these students was around 8 years of age. Again, that is likely influenced by the fact that few people remember events that happened when they were only 2 to 4 years old, which are the ages when the most parents report using spanking (see Chapter 1). On average, spanking for this sample ended around age 10, which is two or three years younger than the national figure. During the year they were spanked the most, these students reported that it occurred an average of 2.9 times a year by their father and 2.8 times a year by their mother. This is an atypically low frequency of spanking. As in other studies, boys were spanked slightly more often than girls (3.0 versus 2.7 times) by their father, but reported the same average number of incidents of spanking by their mother.
For the analysis, students were divided into five groups. Those who reported experiencing no spanking were the first group (39% of the sample). Those who reported spanking were divided into quartiles based on their score on the spanking scale. Group 2 consisted of the one quarter of students who experienced the least spanking. Each consecutive quartile group experienced increasing duration and frequency of spanking. On average, the students in category five (those who experienced the most spanking) were spanked until almost 13 years of age (until 12.6 years old by mother and 12.8 years old by father) and were older the year they were spanked most often (10.5 years). The frequency of spanking in the year that they were spanked the most was about 4 times by their father and about 3 times by their mother.
Prevalence of Risky Sexual Behavior
Almost one half of students (42%) reported having had sexual intercourse at least once. This is very close to the national average for 2009 of 46% (Centers for Disease Control and Prevention, 201 Ob ). The percentages for boys and girls were quite similar. The average age for first having sex was about 14.5 years old. Of the students who reported having had sex, the average number of partners was three. The average number of times they had sex in the past year was 31.5, with males reporting a slightly higher number than females (33.5 versus 29.5). Except for the fact that boys reported somewhat more frequent sex, the sexual behavior ofthe boys and the girls in this sample was remarkably similar, as was also found by the 2009 Youth Risk Behavior Survey (Center for Disease Control and Prevention, 201 Ob ). Students who had sexual intercourse typically had been in their current dating relationships longer (an average of 11 months versus an average of 3 months).
Of the students who had sex at least once, 17% had never used a condom, 58% had never used birth control pills, and 83% had never used other contraceptives. These categories were not mutually exclusive, and there is overlap between them. The mean number of times that they purchased condoms in the past year was 2.7. Only 8% had never used any form of contraception. This is a remarkable figure because it means that almost all sexually active students in this school (92%) had used contraception at least once.
Despite the high use of contraception, 8% of the students who had sex experienced pregnancy, but they were not necessarily the students who had never used contraceptives. This included 5.3% of boys who said they had impregnated a sexual partner and 10.4% of girls who reported having been pregnant. This large difference in pregnancy rate may have occurred for a number of reasons; for example, the male partner may not have known because the female partner did not disclose the pregnancy to the boy who was responsible, the sexual partners of some of the girls who became pregnant may have attended a different school, or they may have no longer been in high school.
Links between Spanking and Risky Sex
Chart 9.1 graphs the relationship of the score on the spanking scale to the risky sex scale. It shows that students who experienced the most spanking (the two groups at the right of the chart) also had the highest risky sex scores. Statistical tests comparing those two groups found that, after holding constant the sex of the student and socioeconomic status of the student's family, students in the two groups who experienced the most spanking had risky sex scores that were higher than students who experienced little or no spanking c (Groups 1, 2, and 3).
Chart 9.1 Students Who Experienced the Most Spanking Were Most Likely to Engage in Risky Sexual Behavior
Chart 9.1 shows that spanking is related to risky sex, but it does not provide information on the process that could explain why spanking increases the probability of risky sex that was discussed in the introduction to this chapter. For example, spanking might be related to risky sex if spanking weakens the bond to parents. To investigate these issues, we used a method called structural equation modeling. The data that were available made it possible to examine five explanations for the relation of spanking to risky sex. We used this method to test the theory that spanking is associated with an increased probability of the following five variables, each of which in turn is associated with an increased probability of risky sex.
* Alienation from parents
* Lower school performance
* Belief in traditional gender roles
* Sexual victimization
* Low self-esteem
Information on the measures used to obtain the data on these five variables is in the Appendix. It should be noted that the data on sexual victimization was obtained by questions that are more applicable to victimization by peers rather than molestation by an adult or incest.
Paths from spanking. Chart 9.2 shows the interrelationships among the variables included in our study. We had expected spanking to be related to all five variables in the center of Chart 9.2 and to sexual risk taking at the right side of the chart. However, only some of those hypothesized relations were found. The heavy black arrows in Chart 9.2 show that spanking is directly related to beliefs in traditional gender roles and to alienation from parents. Because alienation from parents is very similar to a weak child-to-parent bond, these results confirm those reported in the previous chapter-that spanking is associated with a weakened child-to-parent bond. The present findings suggest that spanking increases the probability of alienation from parents (.46), which in tum is associated with an increased probability of risky sex (.57). These are the largest coefficients in the model.
Sexual victimization, self-esteem, and risky sex. The fmdings for this sample also show, as expected, that students who had been sexually victimized (i.e., had a sexual behavior forced on them) were more likely to engage in risky sex (see Appendix for the measure of sexual victimization). It is important to note that, although having been sexually victimized may lead to risky sex, it is also possible that engaging in risky sex increases the probability of being sexually victimized.
Chart 9.2 Paths from Spanking to Risky Sex
The results for self-esteem are the reverse of what was predicted. The arrow from self-esteem to risky sex indicates that high self-esteem is linked to more risky sex when all other variables are controlled. But when only self-esteem and risky sex are examined, without controlling for other variables, self-esteem is linked to less risky sex. At the time this study was planned, we were aware only of studies showing that self-esteem is associated with positive outcomes in many spheres oflife (e.g., Covington, 1989; Rosenberg, Schooler, & Schoenbach, 1989). Since then, however, two critiques of the research on self-esteem have been published disputing the idea that self-esteem has across-the-board benefits. They point to research showing that high self-esteem can be associated with more problem behavior (Baumeister, 2001; Baumeister, Heatherton, & Tice, 1993; Emler, 2002). These reviews suggest that some people with very high self-esteem may not engage in enough self-appraisal of their behavior. This hypothesis might help to explain our unexpected finding that higher selfesteem was linked to more risky sex. Although some of the results may be difficult to understand, the key result from Study 1 is that spanking is associated with an increased probability of alienation from parents, and alienation from parents, in turn, is associated with an increased probability of risky sex.
Study 2: Unprotected Sex by University Students
Study 2 was designed to test two hypotheses. Both hypotheses test the idea that the more spanking experienced as a child, the greater the probability of risky sex as a young adult. The difference between the two hypotheses is in the process or mechanism that could explain why spanking is associated with risky sex later in life. The first hypothesis is that spanking leads to low self-control, and that low self-control is associated with risky sex. This hypothesis is based on assuming that risky sex tends to be an impulsive act that is more likely to occur when there is low self-control.
Our second hypothesis was that spanking is associated with approval of violence, and that the more violence approval, the greater the probability of coercing a partner into sex without a condom. This hypothesis reflects the fact that we measured risky sex as the percent who reported one or more instances in which they said they, "Made my partner have sex without a condom." This is both a measure of sexual coercion and risky sex because it is about coercion to engage in a risky sexual behavior. The question is from the sexual coercion scale of the revised Conflict Tactics Scales (Straus, Hamby, Boney-McCoy, & Sugarman, 1996).
The sample was composed of the 14,252 university students who participated in the International Dating Violence Study and were in a romantic relationship. This is the same sample as was analyzed for Chapters 3, 13, and 16. A detailed description of the study, including the questionnaire and all other key documents and previous publications is available on the website (pubpages.unh.edurmas2) and from the Inter-university Consortium for Political and Social Research where the data file has been deposited. Most of the data were obtained by administering a questionnaire during regularly scheduled classes. The analyses either controlled for gender or were conducted separately for male and female students.
Corporal punishment was measured by a question from the Personal and Relationships Profile (Straus et al., 2010), "When I was less than 12 years old, I was spanked or hit a lot by my mother or father," with the following response categories: 1 = strongly disagree, 2 = disagree, 3 = agree, and 4 = strongly agree. The measure of self-control was a six-item scale described in Rebellon et al. (2008). For the present study, we used a cutoff score of the lower boundary of the top fifth ofthe sample as the criterion measure of high self-control. Violence approval was measured by the eight-item violence approval scale of the Personal and Relationships Profile (Straus et al., 201 0). High violence approval participants were those in the top scoring fifth ofthe sample.
Relation of Spanking to Risky Sex
Chart 9.3 shows that the more a student was spanked before age 12, the greater the likelihood that he or she would insist on having sex without a condom. It also shows that this relationship applies to both men and women.
Chart 9.4 provides the results of the test of the two hypotheses about the processes or mechanism that might explain why having been spanked is related to this aspect of risky sex. The numbers on the path arrows are the percentages by which the variable at the left are associated with an increased or decreased probability of the variable at the right. These are based on a logistic regression analysis that also controlled for the age and sex of the student, education of the mother and father, length of the dating relationship, and the score on a scale to measure a tendency to avoid disclosing socially undesirable characteristics (the limited disclosure scale ofthe Personal and Relationships Profile (Straus et al., 2010; Straus & Mouradian, 1999). High self-control and high violence approval students are those in the high scoring fifth ofthe samples. The middle arrow in Chart 9.4 shows the same relationship as was graphed in Chart 9 .3-that the more spanking, the greater the probability of insisting on sex, without a condom. The additional contributions of the results in Chart 9.4 are the path arrow at the top and the bottom of Chart 9.4. They indicate two of the processes that explain why spanking is related to risky sex.
Chart 9.3 The More Spanking, the Greater the Percent Who Insisted on Sex without a Condom
Chart 9.4 Links between Spanking as a Child and Unprotected Sex by University Students
Self-control. The top left path of Chart 9.4 shows that spanking is associated with a 38% reduction in the percentage of students who were high in self-control. This relationship is shown in more detail in Chart 9.5. The top right path of Chart 9.4 shows that high self-control is associated with a 39% decrease in the percent who insisted on sex without a condom. This relationship is shown in more detail in Chart 9.6. Thus, spanking is related to insisting on sex without a condom in part because spanking is associated with a lower probability of high self-control.
Violence approval. The lower left path in Chart 9.4 shows that spanking is associated with a 32% increase in violence approval. This relationship is described in more detail in Chart 9. 7. The lower right path of Chart 9.4 shows that violence approval is associated with a 56% increase in insisting on sex without a condom. This relationship is shown in more detail in Chart 9.8. Thus, spanking is related to sex without a condom in part because spanking is associated with an increased probability of approving violence in some circumstances. See Chapters 5 and 12 for other studies that found that the more spanking, the more approval of violence.
Chart 9.5 The More Spanking, the Lower the Self-Control as a Young Adult
Chart 9. 6 The Higher the Self-Control, the Lower the Percent Who Insisted on Sex without a Condom
Summary and Conclusions
The results of the two studies described in this chapter show that spanking is associated with an increased probability of risky sex by adolescents and young adults; Despite a very different sample, a different source of data, and different measures, the results of Study 1 also confirm what was found in the previous chapter-that spanking weakens the parent-child bond. The tendency of spanking to undermine the parent-child bond helps to explain why spanking is linked to problems such as antisocial behavior (Chapters 6 and 7), delinquency (Chapter 8), lower academic achievement (Chapter ll),'and crime (the chapters in Part IV).
Other Paths from Spanking to Risky Sex
Although we examined several processes that could explain why spanking is related to risky sex, there are other important influences on sexual behavior that are affected by spanking and need to be investigated in future research. The studies in Part III of this book show that spanking slows cognitive development and lowers academic achievement. Low academic achievement has been shown to be associated with risky sex (Abma, Driscoll, & Moore, 1998; Murray, 1938; Resnick et al., 1997) and adolescent pregnancy (Scaramella, Conger, Simons, & Whitbeck, 1998). Further, Coopersmith (1967) found that spanking is associated with low self-esteem, and low self-esteem is associated with decreased use of contraception (Kowaleski-Janes & Mott, 1998). For young women, low self-esteem has been found to be associated with an increased risk of pregnancy (Berry, Shillington, Peak, & Hohman, 2000; Kowaleski-Janes & Mott, 1998).
Chart 9. 7 The More Spanking, the Higher the Violence Approval as a Young Adult
Although our results demonstrate that spanking is associated with risky sex, it must be remembered that these data are cross-sectional, so we cannot draw conclusions about the direction of causality. However, the fact that most spanking tends to occur before the age at which risky sex occurs makes a causal relation plausible. On the other hand, the fact that spanking occurs before the risky sex does not rule out the possibility that the relationship between spanking and risky sex is spurious, that is, caused by a third variable that is associated with both. For example, parents who are under great stress may be more likely to spank and also less likely to exercise guidance and supervision over their adolescent children. It may be the stress variable that explains both more spanking and less adequate guidance.
Chart 9.8 The More Approval of Violence, the Higher the Percent Insisting on Sex without a Condom
But the case for believing that spanking causes the harmful side effects found by these two studies, and the many other correlational studies published during the last 30 years, rests on new evidence obtained from the growing number of longitudinal studies conducted since 1997-the year the first studies were published that showed that spanking is associated with a subsequent change for the worse in the child's behavior (Chapter 6 and Gunnoe & Mariner, 1997). The longitudinal studies in the chapters on child antisocial behavior (Chapter 6), mental ability (Chapter 10), and adult crime (Chapter 15) controlled for the misbehavior that presumably elicited the spanking, and found that spanking was associated with a subsequent increase in the problematic behavior. Chapter 19 summarizes many other longitudinal studies that have found that spanking is associated with a subsequent increase in the probability of physical aggression, antisocial personality, and crime as a child and as an adult.
In addition to identifying an additional important correlate of spanking, the two studies in this chapter provide another example of why it is difficult or impossible for parents to see the harmful effects of spanking-because it is linked to behaviors that do not occur until years after most parents have ceased to use spanking and behaviors of which parents might not even be aware.
Part III Spanking and Human Capital
10 Mental Ability
A guest opinion editorial in The New York Times in 2011 advocated spanking (Williams, 2011). The reader reaction was mostly critical and opposed spanking. This is the opposite of the reader response to the Time magazine article described in Chapter 1. The Time article summarized a study showing that spanking was associated with an increase in children's antisocial behavior. The letters to the editor of Time were almost entirely defenses of spanking, and many dismissed the research as junk science. Perhaps the antispanking letters to The New York Times compared with the pro-spanking letters to Time was a reflection of the characteristics of readers of The New York Times. It could also be the tendency for more people who disagree with an article to write a letter to the editor than those who l:J,gree with the article.
Mallie J. Paschall is the coauthor of this chapter.
Regardless of the explanation, one of the letters to The New York Times that favored spanking was particularly relevant for this chapter. The writer said: "I'll never forget watching a woman trying to reason with a two-year-old, explaining all the reasons not to do something. At two, the child does not have the capacity for critical reasoning, but he or she understands a swift spank on the bottom." We doubt that a swift spank on the bottom, without explanation or some other means of conveying what specifically was wrong, will teach the child right and wrong, and it may teach entirely wrong lessons, such as "Mommy doesn't like me." Moreover, not explaining to a child is a form of cognitive deprivation. Research shows that talking to an infant or toddler is associated with enhanced mental ability. That research suggests that the more parents talk to a child, including talking to infants, the larger their vocabulary and the more different type of words they use (Huttenlocher, Waterfall, Vasilyeva, Vevea, & Hedges, 2010). Presumably this is because these are formative ages for the development of neural connections in the brain (Blakeslee, 1995; Dawson & Fischer, 1994). This has implications for understanding the side effects of spanking because, as the letter just quoted indicates, more spanking usually means less talking to a child and less talking to a child can result in less brain development. This line of thinking led us to look into the following questions that are addressed in this chapter.
* What process might link spanking and mental ability?
* Is there any evidence from previous research that spanking adversely affects mental development?
* What percent of children age 2 to 4 and 5 to 9 were spanked during the two sample weeks, and how many times per week were they spanked?
* Is spanking associated with a subsequent change in the child's mental ability, and does this apply to toddlers as well as school age children?
* Do contextual factors, such as socioeconomic status and whether mothers are loving and supportive, affect the relation of spanking to mental development?
* Do the adverse effects of spanking apply as much to toddlers as to older children?
* What are the theoretical and policy and practice implications of the results of this study?
Processes that Could Link Spanking and Mental Ability
Less Verbal Interaction
As suggested in the previous paragraph, when parents spank or slap a child's hand for touching a forbidden object, they are less likely to engage in cognitive methods of behavior control, such as explaining to the child why the object should not be touched. Of course, they usually do both, but some parents have told us that they spank because they don't have time for all that explaining stuff, and many think that just a quick swat will take care of things. The reduction in talking to a child by parents who spank could also come about if, as other parents have told us, they believe that you can't reason with a two-year-old. For these parents, spanking often replaces reasoning and explaining. The other side of the coin is more clear because the less spanking used by a parent, the more verbal interaction is needed to teach and correct the child. The increased level of verbal interaction may enhance the growth of neural connections in the brain, and with it the child's mental development.
Spanking and Stress
In addition to limited verbal interaction, spanking could adversely affect mental ability through other processes. Being slapped or spanked by someone 2 or 3 times the size of the child, and from which there is no escape, is a frightening and threatening event that many children experience as highly stressful (Saunders & Goddard, 201 0; Turner & Finkelhor, 1996; Willow & Hyder, 1998). For some parents, fright and fear is an explicit part of their discipline strategy. A father recently told us that he wants his son to fear him so that he will obey. "Respect will come later," he said. Relatively few parents are this explicit about fear. But regardless of whether it is intentional, when parents spank, they are using fright as a means of teaching. Ironically and unfortunately, fear and fright can result in cognitive deficits such as erroneous or limited coding of events and diminished elaboration (Heuer & Reisberg, 1992; Perry, 2006). The physiological arousal from being hit by a parent tends to narrow the range of cues in the environment to which the child is sensitive (Bruner, Matter, & Papanek, 1955; Easterbrook, 1959).
There is now evidence that spanking is associated with adverse changes in brain structure. Bugental et al. (2003), Bugental, Schwartz, and Lynch (2010), and Tomoda et al. (2009) found that the 34% of infants in their study who were spanked "showed high hormonal reactivity to stress." They concluded that "the hormonal responses shown by infants may alter the functioning of the hypothalamic-pituitary-adrenal (HPA) axis in ways that, if continued, may foster risk for immune disorders, sensitization to later stress, cognitive deficits, and social-emotional problems" (p. 237).
Moreover, to the extent that spanking is experienced as stressful, it is a stress that, for many children, continues for several years. Those who defend spanking typically approve of spanking only with young children, for example, ages 2 to 6 (see the consensus statements and personal statements in Friedman & Schonberg, 1996b). In addition, as shown in the chapters on spanking and the child-to-mother bond (Chapter 8) and risky sex (Chapter 9), spanking tends to undermine the attachment and bond between the child and the parent and reduce a child's motivation to learn from parents. Whatever the intervening processes, if spanking influences mental ability, it has broad implications because as shown in Chapter 2 on spanking in the United States, at least one third of U.S. children are hit as infants and over 90% as toddlers, and for one third, it continues into the early teenage years.
Previous Research
Both physical abuse and spanking are acts of physical assault. A key difference is that, by definition, spanking does not cause physical injury. However, the difference between spanking and physical abuse is not as large as one would gather from newspaper stories. News articles are almost always about physically injured children, whereas empirical research has found that most physical abuse cases dealt with by Child Protective Services also do not involve an injury serious enough (Trocme, MacMillan, Fallon, & De Marco, 2003) to require medical treatment. Spanking and physical abuse are also similar in that both increase the probability of psychological injury. For psychological injury, the difference is that the effect size of physical abuse (the probability of psychological damage occurring) is much smaller for spanking, but still present. For example, Strassberg et al. (1994) found that both physical abuse and spanking were associated with more aggressive behavior observed in kindergarten. Spanking was associated with twice the number of acts of physical aggression compared with the children who were not spanked six months earlier. But physical abuse was associated with 4 times as many physically aggressive acts. A study by Afifi, Browmidge, Cox, and Sareen (2006) of a U.S. nationally representative sample found that spanking was associated with 2.2% more cases of externalizing problems such as antisocial personality, whereas physical abuse was associated with 15.3% more cases. Thus, the adverse effect of spanking was much less, but still present. As a consequence, the results of studies showing that maltreatment at an early age can have enduring negative effects on a child's brain development and function (Craig, 1986; Kinard, 1999; Teicher, Andersen, Polcari, Anderson, & Navalta, 2002; Widom, 1989) can also apply to children who experience spanking, but with a lower probability of it happening.
A possibly tragic aspect of cultural beliefs about spanking is reflected in the advice given to parents by many professionals such as pediatricians and child psychologists. It is the belief that spanking is acceptable if restricted to early childhood. For example, a consensus statement drawn up at the conclusion of a conference on corporal punishment sponsored by the American Academy of Pediatrics, recommended limiting spanking to children between 2 and 6 years old (Friedman & Schonberg, 1996a). However, there is no research evidence for that recommendation. It is a part of folk beliefs about spanking, even though put forth by professionals. Contrary to this folk belief, the risk of psychological damage from spanking may be greatest in early childhood. We found that the increase in antisocial behavior that occurred subsequent to spanking applied to toddlers as well as older children (Chapter 6). This chapter investigates whether that also applies to a slower rate of mental development for toddlers as well for older children.
Hitting children at a young age may be worse than hitting them when they are older because the neural connections in the brain are being formed more rapidly by young children. Another reason why hitting young children may be the worst possible age is that spanking is associated with a weakened bond to the mother, as shown in Chapter 7 on impulsive spanking and child well-being and Chapter 8 on spanking, the child-to-mother bond and delinquency, and by Afifi et al. (2006) and Coyl, Roggman, and Newland (2002). If the ages below 6 are the most crucial for developing bonds of attachment to the ·parent, the weakened bond may reduce a child's motivation to learn from the parent who is hitting them.
Studies Suggesting that Spanking Might Adversely Affect Mental Ability
There are several studies that found spanking to be related to characteristics that are related to mental ability. A study by Bodovski and Youn (20 1 0) of a large, nationally representative sample of U.S. elementary school students measured spanking when the children were in kindergarten. They found that spanking in kindergarten was associated with lower 5th-grade math achievement even after controlling for many other variables such as parental depression, parental warmth, and racial or ethnic group. These results bear on the issue of this chapter because one of the causes of poor math ability is low IQ.
A study of a national sample of American adults found, even after controlling for the education and occupation of the respondent's parents and other potential confounds, that the more spanking, the lower the probability of the study participant being in the top fifth of the occupational and income distribution for the United States (Straus,& Gimpel, 2001). The next chapter presents the results of a study of another nJtionally representative sample of U.S. adults and found that the more spanking experienced, the lower the percentage who graduated from college.
A unique natural experiment was studied by Talwar, Carlson, and Lee (2011). They compared the executive functioning of children in two West African private schools: one high in use of corporal punishment and one nonpunitive. Executive function is an umbrella term for mental processes such as planning, problem solving, verbal reasoning, and mental flexibility. Both schools served high-education families. The study controlled for verbal IQ. In kindergarten, the children did not differ in executive functioning. By Grade 1, however, the children in the punitive school had lower executive functioning scores (i.e., they had fallen behind the growth in executive functioning of the children in the school without corporal punishment). Although this study was of corporal punishment by teachers, not parents, we believe the same processes are involved in spanking by parents.
Studies of Spanking and Mental Ability
We found six studies that examined the relation of disciplinary practices to standard measures of mental ability. Bayley and Schaefer (1964) studied the children in the Berkeley Growth Study whose mental ability had been tested at frequent intervals from birth to age 18 (approximately 25 boys and 25 girls). From age four on, especially for boys, the more punishment was used, the lower the mental ability. However, the adverse effect on mental ability may not have been the result of spanking because their measure of punishment was not restricted to spanking.
Smith and Brooks-Gunn (1997) studied 715 low birth weight children. Discipline was measured at 12 and 36 months. The Stanford Binet intelligence test was administered at 36 months. They found that the children who experienced harsh discipline had the lowest IQ, even after controlling for many socioeconomic factors, including birth weight, neonatal health status, ethnic group, mother's age, family structure, mother's education, and family income. One limitation of this study is that the harsh discipline measure included scolding the child. Therefore, it is not possible to separate out the effect of spanking alone. Another limitation is that there was no Time 1 measure of mental ability to enable testing whether spanking at Time 1 was followed by a subsequent decrease in IQ scores.
Power and Chapieski (1986) interviewed and observed the interaction of 18 upper middle class mothers with their 12- to 15-month-old children. They compared children whose mothers relied on spanking with children whose mothers rarely or never spanked. The outcome measure of interest was children's score on the Bayley infant development scale when the children were tested at an average age of 21 months. The mental ability of the small proportion of children whose mothers rarely or never spanked, averaged 20 points higher than the rest of the children. Two limitations of this study are the small number of cases and lack of differentiating "rarely" spanking from "never."
Aucoin et al. (2006) compared children who experienced no corporal puriishment over approximately a 2-week period, those who had experienced mild levels of corporal punishment (i.e., 1 or 2 instances), and those who had experienced high levels of corporal punishment (i.e., 3 or more instances), and found differences between groups in IQ. Children in the low corporal punishment group scored significantly higher on a brief measure of intelligence than children in the high corporal punishment group. Although the statistical analysis did not control for family income, race, or gender, the groups did not differ significantly on these variables.
Two recent and comprehensive studies provide the most definitive evidence on the extent to which spanking harms the development of mental ability. A study of a sample of 2,573 low-income, White, Black, and Mexican American children ages 1 to 3 (Berlin et al., 2009) found that, after controlling for a number of other variables, spanking at all three ages predicted lower Bayley mental development scores at age 3. Another longitudinal study found similar results. This study examined an urban U.S. national sample of779 children (MacKenzie et al., 2011). It found that frequent spanking by mothers at age 3 was associated with a large subsequent decrease at age 5 in the probability of the child being high in cognitive ability relative to other children as measured by the Peabody Picture Vocabulary Test. This study controlled for 30 other variables, including child characteristic variables such as low birth weight and difficult temperament, maternal and family characteristics such as mother's age at the time of the child's birth, and pre-natal risk variables such as substance abuse during pregnancy. This may be more than in any other study of the effects of spanking and included variables that are particularly relevant for child mental ability such as maternal depression, maternal intelligence, and an observational measure of the extent to which the home environment was cognitively stimulating. This study is also important because it provides information on the belief that the adverse effects of spanking do not apply in cultural contexts where it is the norm, such as among Blacks. This study, like most of those summarized in Chapter 14 on cultural context effects, found that the adverse effects of spanking on mental ability applied to Black children as well as Hispanic and White children.
The findings of these studies lead us to hypothesize the following:
1. Spanking is associated with lower mental ability relative to other children of the same age.
2. When data is analyzed over a 4-year time period, spanking at Time 1 is associated with an average decrease in mental ability at Time 2, relative to other children of the same age.
3. The decrease in mental ability is greater for preschool-age children (age 2 to 4) than for children age 5 to 9.
Sample
The sample consists of the children of women who were first interviewed in 1979 as part of the National Longitudinal Survey ofYouth (see Chapter 6 for a description of the sample). At the start ofthe study in 1979, the women were 14 to 21. Starting in 1986, those women who had children were interviewed periodically about child rearing practices and behavior problems of their children. This included 806 children age 2 to 4 and 704 children age 5 to 9 whose mental ability was tested. Additional information on the characteristics of the sample is in (Straus & Paschall, 2009).
We studied two groups of children because each can help address a different issue. The younger group was studied because the theory underlying this study is most applicable to young children because the development of neural connections is greatest for infants and toddlers. If that is the case, the adverse effects of spanking should be greatest for the younger children. In addition, the effect of what the parents do may be greater for younger children because, on average, they have fewer nonfamily experiences that could be related to mental ability (e.g., school experiences) than older children. We studied the second group of children so that we could replicate the hypotheses with children who were age 5 to 9 at Time 1. Many parents continue spanking into this age range (see Chapter 2). Also, the American Academy of Pediatrics (Friedman & Schonberg, 1996a) advises against using spanking with children older than 6, implicitly because they believed that spanking of older children is more likely to result in harm than is spanking of preschool-age children. If that is correct, it suggests that the adverse effects of spanking on mental ability should be greater for the 5- to 9-year-old children than for the 2- to 4-year-old children.
Measures
Mental Ability
For both age groups, mental ability was measured at both Time 1 and four years later at Time 2-using at Time 1 as many of the following tests as were appropriate for each age child: body parts recognition, memory for locations, and motor and social development. At Time 2, mental ability measure was the Peabody Individual Achievement Tests (PlAT) for math and reading recognition (see Baker et al., 1993, p. 3043). The scores from the mental ability measures were transformed so that the resulting scores indicate how far above or below the average level of mental ability each child was relative to other children in this study of approximately the same age (see Straus, 2009a).
Spanking
Spanking was measured during one sample week in 1986 and again in 1988 using two types of data. The first is observation by the interviewer of whether the mother spanked or hit the child during the course of the interview. The second was two interview questions: "Did you find it necessary to spank your child in the past week?" Mothers who said they had spanked were asked: "About how many times, if any, have you had to spank your child in the past week?" We used these data to create a scale that combined the observed behavior and mothers' reported behavior for Time 1. If the mother was observed hitting the child, it was counted as one instance of spanking in addition to any that the mother reported as having occurred in the past week. Next, we grouped the children into four categories: (1) those who experienced no spanking in either of the two weeks, (2) those who experienced either one, (3) two, or (4) three or more instances. The fact that a score of zero identifies children who were not spanked in either of the two sample weeks over a 2-year time span, makes it plausible to consider the zero group as children for whom spanking was extremely rare, or in some cases, nonexistent. However, the spanking scale used for this study does not eliminate the possibility that the children in the zero category experienced spanking on rare occasions.
Control Variables
Because many other things that influence mental ability may also be correlated with spanking, it is necessary to take those other influences into account to pin down the effect of spanking. To do this, analysis controlled for the amount of cognitive stimulation and emotional support by the mother. Cognitive stimulation was measured by the Home Observation for Measurement of the Environment Short Form (Caldwell & Bradley, 1984). This includes questions and observations on whether the mother read to the child or whether the mother helped the child learn colors, numbers, shapes, or the alphabet; and how many books the child had of his or her own. Emotional support was measured by behaviors such as how often the child had dinner with both parents, whether the mother caressed or kissed the child, and whether the mother's voice showed positive feeling toward the child. See Straus & Paschall (2009) for additional information on these measures.
Other variables controlled for this study included the child's birth weight, child's age, child's ethnicity, child's gender, number of children of the mother in the home, mother's age at child's birth, mother's education, and if the father was living with the mother at Time 1.
Prevalence and Frequency of Spanking
• Only 6.6% of the 2- to 4-year-old children were not hit at all in either of the two sample weeks; thus 93% were hit at least once in those two weeks. This is almost identical to the 94% of parents who reported hitting children in this age group in our national survey of U.S. children (Chapter 2).
• The percent of 5- to 9-year-old children who were not hit is much greater, but more than one half (58.2%) were spanked in that period.
• Almost one half of the 2- to 4-year-old children were hit 3 or more times in those two weeks.
• Mothers of children age 2 to 4 years who had spanked in the past week, did so an average of 3.6 times that week. One third of the mothers spanked 4 or more times, and 12.8% spanked 7 or more times that week.
• The mothers of children age 5 to 9 who had spanked in the past week reported doing so an average of 2.5 times that week.
As pointed out in Chapter 2 on spanking in the United States, these may be low estimates because spanking is used so frequently and is taken for granted as being sometimes necessary. As a consequence, parents do not realize how often they do it. One indication of the taken-for-granted nature of spanking children is that, among the children who were 26 months old at Time 1, 18% of the mothers hit the child during the interview. Although we cannot be sure of the number of times these children were hit, it is clear that they experienced a lot of spanking.
Spanking and Development of Mental Ability
We hypothesized that spanking slows the rate of further mental development. Therefore, we should find that 4 years down the road, the more children who were hit by their parents, the more likely they are to fall behind the average for children their age. To test this, it was necessary to have data that could determine if spanking is associated with change in mental ability, and specifically, whether the more spanking experienced, the slower the growth of mental ability. The results of testing this hypothesis are presented in Chart 1 0.1 and, in more detail, in Straus and Paschall (2009). Chart 10.1 shows that the 2- to 4-year-old children who were not spanked gained an average of5.5 points more than the average child, and the 5- to 9-year-old children gained an average of almost 2 points more than the average child.
At the other extreme of the spanking categories, the 2- to 4-year-old children who were hit 3 or more times in the two sample weeks neither gained nor lost relative to other children their age. This is consistent with the fact that they are the typical children in this age group, for whom we found that 48 were hit 3 or more times. Thus, 2- to 4-year-old children who experienced 3 or more instances of spanking were typical of all children in the nation in respect to both being spanked and growth in mental ability.
Chart 10.1 The More Spanking, the More the Child's Mental Ability Fell Behind that of Other Children Four Years Later
For children age 5 to 9, the statistical norm for spanking was quite different. Instead of most children that age being hit 3 or more times in those two weeks, as was true of the younger children, only 15% of the 5- to 9-year-old children were hit 3 or more times in those two weeks. The relation of spanking to mental ability, however, was similar to the results for the 2- to 4-year-old children. The mental ability of the children whose parents did not spank in either of the two sample weeks was greater than the children who were hit even once in those two weeks. They gained an average of almost 2 points more than the average child. On the other hand, the 5- to 9-year-old children who were hit once neither fell behind nor gained compared with other children. In other words, their score stayed at about 100. The 5- to 9-year-old children who were hit 2 or more times in those two weeks fell slightly behind the average child in mental ability in the 4 years following the initial testing.
To understand these results, it is important to realize that the decreases shown in Chart 10.1 do not mean that spanked children became less smart. Unless there is some developmental impairment, all young children become much smarter than they were 4 years earlier. A mental ability score of 100 indicates a score at the average for children of the same age. To maintain a score of 100 over a 4-year period, a child's ability must increase during those years at the average rate of increase. Thus, the decreases associated with spanking do not indicate an absolute reduction in mental ability, only that spanking is associated with failing to keep up with the average development of mental ability.
Does the Harmful Effect of Spanking Depend on the Social Context?
As noted earlier, on the basis of both theory and the results of empirical research, there are grounds for expecting that the effect of spanking depends on the presence or absence of other variables, or as it is sometimes put, the effects of spanking may be context specific. In order to examine this, we considered spanking in combination with other variables in this study. Chart 10.1 shows that the adverse effect of spanking is greater for toddlers than for early schoolage children. We examined each of the other child and family characteristics available for this sample to see if they reduced or exacerbated the relation of spanking to mental ability. For example, it is often claimed that spanking is not harmful in a sociocultural context where it is the norm, such as among Blacks (see Chapter 13). We tested this idea and found that the results for Blacks parallel those for Whites. Spanking was found to be related to a slowing of mental ability development among both Blacks and Whites and also among each of other nine parenting and demographic context variables.
Is "Just Once" Harmless?
Defenders of spanking believe it is harmless if done only rarely. They do not indicate how often rarely is, so their belief cannot be tested exactly. For this study, the best approximation to only rarely was the children who were spanked only once in the two sample weeks. We consider this rare for U.S. parents because only 10.5% of the children in this sample were spanked this rarely. We compared the 6.6% of the children who were not hit at all during the two sample weeks with the 10.5% who were hit only once, those hit twice, and those hit 3 or more times. The mental ability of children of mothers who hit them even once in these two weeks was slower than the development of the children whose mothers did not hit them at all. However, the difference was just short of the usual standard for being statistically dependable for either the 2- to 4-year-old children (p = .062) or 5- to 9-year-old children (p = .057).
Summary and Conclusions
This study investigated the extent to which mothers spanked in large national samples of children age 2 to 4 and 5 to 9. It tested the hypothesis that spanking experienced by these children is associated with slower development of mental ability over a 4-year period.
Prevalence of Spanking
Ninety-three percent of the mothers of children age 2 to 4, and 58% of mothers of children of the 5- to 9-year-old age group spanked in the two-week referent period. These prevalence rates are consistent with the other studies cited in the introduction to this chapter. Among those who spanked, it occurred an average of3 .6 times per week. This figure is consistent with the average of2.5 times per week for toddlers found by Holden, Coleman, and Schmidt (1995), provided one takes into account that these authors studied college-educated mothers wllo tend to spank less than mothers with less education (Day et al., 1998). If the average of3.6 per week is extrapolated to a year, it results in an estimated 187 instances per year per child. This is at least 1 0 times higher than the average number of times based on studies that ask parents about the number of times they SQanked in the past year (the method used for Chapter 2). We suggest that the much lower chronicity of spanking (frequency of spanking among those who spanked) found by studies that use a 1-year recall period occurs because, for many parents, spanking is such an everyday and taken-for-granted occurrence that parents do not realize how often they use it. This interpretation is consistent with findings from a pioneer study by Goodenough ( 1931) that found that when mothers used a diary to record their disciplinary tactics, the chronicity of spanking was 6 times greater than when the figure was based on recall during an interview.
We believe that the public and most service providers and social scientists do not realize the high prevalence and how chronic spanking is in the lives of U.S. children. We further suggest that misperception ofthe extent of spanking is an example of selective inattention (Dexter, 1958) by members of a society in which spanking is the norm (Straus & Mathur, 1996). Selective inattention may be one of the mechanisms that enables our society to continue to support spanking because it avoids the necessity of facing up to the fact that almost all children are hit, and most are hit frequently. Without the information on prevalence and frequency, the results about the effects of spanking could be dismissed as applicable only to atypical, high-spanking parents. Indeed this was a basis used by defenders of spanking to dismiss the general applicability of the results reported in Chapter 6 on spanking and children's antisocial behavior; spanking increases rather than decreases antisocial behavior of children (Ambati, Ambati, & Rao, 1998).
Relation of Spanking to Development of Mental Ability
Although almost all U.S. children experience at least some spanking, the differences in how often mothers spank provided sufficient variation in the use of spanking to test the hypothesis that the more spanking experienced by a child, the slower the development of mental ability. The results from this study are consistent with the hypothesis. We found that 2- to 4-year-old children who experienced no spanking in either of the two sample weeks gained an average of 5.5 more mental ability points over the next 4 years than children whose mothers spanked. Similarly, children in the 5- to 9-year age group, whose mothers did not spank in either week, gained an average of about 2 points more than children whose mothers spanked. Conversely, for both age groups, spanking was associated with a decrease from Time 1 to Time 2 in mental ability test score.
These findings remained, even after adjusting statistically for 10 variables that could be the real underlying causes of the link between spanking and slower development of mental ability. This includes the child's birth weight, child's age, child's ethnicity, child's gender, number of children of the mother in the home, mother's age at child's birth, mother's education, mother's cognitive stimulation with the child, mother's emotional support, and if the father was living with the mother at Time 1. In addition, the results of our analysis are probably minimum estimates of how closely spanking is related to slower development of mental ability because of the relatively low reliability of cognitive testing of children as young as those in this sample at Time 1.
Contextual effects. The question of whether there are circumstances or contexts that make spanking appropriate has been the subject of much debate. For example, it has been argued that if parents are loving and supportive, children will know parents are spanking them for their own good, and spanking will have no adverse effect. Given the debate and theoretical importance of contextual effects, we tested the interaction with spanking of the 10 variables listed in the previous paragraph. These analyses indicate whether each of the 10 made a difference in the degree to which spanking is related to the development of mental ability. For example, if mothers are loving and supportive, is there still a harmful effect of spanking? The answer is yes there is. Or, more generally, none of the 10 tests of moderator effects found that these variables either mitigated or enhanced the relation of spanking to mental ability. The lack of any context effects indicates that the relation of spanking to slower development of mental ability applies even when done by loving and attentive parents, even when it occurs among a sector of the population with cultural norms that approve of spanking such as Blacks, among children of low and high socioeconomic status, etc. These results can also be interpreted as showing that spanking has a unique harmful effect, that is in addition to the effect of the 10 variables that were statistically controlled.
The fact that none of these 10 variables mitigated the tendency for spanking to be related to slower development of mental ability does not mean that these variables have no effect on mental ability. Table 3 in Straus and Paschall (2009) shows that many of them were related to the child's mental ability. For example, the more children in the family, the lower the average mental ability. This result is consistent with much other research on the effect of number of children (Blake, 2011 ). Many readers will find this implausible, just as many will find it hard to believe that spanking lowers mental ability. They might say "I was spanked and my IQ is 120" or "I was one of five children and my IQ is 120." To understand the results on the relation of spanking or number of children to mental ability, it is necessary to understand that almost all social science and medical research results are typically in the form of risk-factor relationships rather than one-to-one relationships. See the section of Chapter 1, Risk Factors: The Real Meaning of Social Science Research.
Limitations
Although we studied a large and nationally representative sample of children, contr,olled for many potential confounds, and examined many contextual effects, there are limitations to keep in mind. The data are more than 20 years old, and many changes in parent practices have probably taken place during this period. However, as shown in the chapters on the prevalence of spanking in the United States (Chapters 2 and 17), there has been virtually no change in the use of spanking children of the ages of those in this study. Moreover, the issue of this study is not the prevalence of spanking but the effect of spanking on children. That is likely to be the same even if spanking had decreased. Second, this study did not have a method for screening out severe forms of corporal punishment that might be considered physical abuse. Thus, it is possible that some of the parental reports of spanking might actually have been in the form of physical attacks that are severe enough to be physical abuse rather than legal spanking. Third, because this is a longitudinal study, the measures of mental ability had to be different at Time 1 and Time 2 in order to be age-appropriate. We took this into account by standardizing the scores so that for each testing, the scores indicate how much above or below the average of the children tested at that time point. Fourth, the relatively small effect size of spanking needs to be kept in mind. If future studies confirm the findings of this study, it means an average gain of about 5 points for children who do not experience spanking. For an individual child, a 5-point gain in a 1 00-point mental ability test is good but, given measurement error, not a major difference. Nonetheless, it is a well-established principle in epidemiology that reducing a widely prevalent risk factor with small effect size, such as spanking, can have a much greater impact on public health than reducing a risk factor with a large effect size, but low prevalence, such as physical abuse (Rose, 1985; Rosenthal, 1984, p.131). An example showing the greater effect in reducing antisocial behavior of ending spanking than of ending physical abuse is given in the concluding chapter.
Fifth, the children who were not spanked in either ofthe two sample weeks could have been spanked in the other 50 weeks of the year. Similarly, children who experienced 1 or 2 instances of spanking in the two-week sample period may have experienced much more spanking on a regular basis. Finally, no data on the behavior of the fathers is available for these children, and the measure of the mother's emotional support is minimal.
Perhaps the most important limitation of the study is that we did not control for the Time 1level of behavior problems or externalizing behavior. Research indicates that children with low IQ scores also often have externalizing behavior problems and impulsivity, which may in tum lead to a higher likelihood of spanking. Therefore, it is possible that those with lower IQs had more behavior problems that lead to more spanking, rather than spanking leading to lower IQ. That is why the longitudinal studies that did control for Time 1 behavior (Chapters 6, 10, and 15, and the longitudinal smdies reviewed in Chapter 19) are so important.
Implications for National Level of Mental Ability
A review of data on mental ability found an increase in scores on many different intelligence tests in a number of countries (Neisser, 1997). The evidence compiled by Neisser and others leaves little doubt that intelligence test scores have been increasing, and that the increase is not an artifact of the tests used. What is in doubt is why this has occurred. Neisser identifies a number of plausible contributing factors. For example, there is abundant evidence that children of educated parents obtain higher scores on intelligence tests (Flynn, 1999; Neisser et al., 1996). Because the level of education of parents has been increasing worldwide, this is likely to be an important part of the explanation. Another strong possibility is that nutrition levels have been improving because better nutrition is associated with greater mental ability (Rizzo, Metzger, Dooley, & Cho, 1997).
Chapter 4, which shows that the more children in a family, the higher the probability of spanking, suggests another change that could be contributing to the increase in IQ. Around the world, birth rates have been declining. That means less use of spanking around the world. Therefore, as shown in this chapter, less spanking means a higher average IQ. It does not mean that everyone is smarter. The section of Chapter 1, Risk Factors: The Real Meaning of Social Science Results, explains how that could be correct if, as is probably the case, most readers of this book were spanked as children and do not suffer from low IQ.
Reductions in spanking and its replacement by cognitive forms of correction might also explain part of the worldwide increase in IQ. When parents spank less, they tend to use more cognitive methods of correction. They also tend to shift away from the idea that children should be seen and not heard. Explanation, rather than the fear of being spanked, becomes the reason the child should engage in socially appropriate behavior. If this theory is correct, and if there has been a worldwide decrease in spanking, then that decrease could have contributed to the worldwide increase in scores on mental ability tests. We do not have data on change in spanking to test that theory. However, we do have data on national differences in spanking and national differences in IQ for 32 nations. An analysis of this data (Straus, 2009a) found a tendency for the larger the percent of children under 12 who were spanked, the lower the national IQ (r = .21 ). The correlation with IQ of or slapping and spanking older children was much higher (r = .43). However, with a sample of only 32, after introducing controls for such variables as the level of economic development and the average education, the regression coefficient was not statistically dependable. These results suggest that future research with a larger sample of nations might provide more definitive evidence that prevalence of spanking in a nation is associated with a lower national IQ. However, for a definitive test, there needs to be data on change in the percent spanked to see if, when that decreases, IQ goes up.
Policy Implications
Although a smaller percent of parents of older children in the United States now spank, almost all U.S. parents continue to spank and slap toddlers (see Chapters 2 and 17). There is a cruel irony to this because both the theoretical basis and the findings of this study suggest that it is precisely at early stages of development that avoiding spanking and using cognitive modes of correction may be most beneficial for the development of mental ability (Doyle, Harmon, Heckman, & Tremblay, 2009). Moreover, it is even more ironic that most defenders of spanking have reformulated their position to oppose spanking of older children and accept spanking of toddlers (Friedman & Schonberg, 1996a). It is ironic because that is precisely the age group this study suggests is most vulnerable to adverse effects on mental ability. Moreover, there are many studies that, although they did not compare age groups, have found serious adverse effects of spanking for toddlers. The combination of our findings on mental ability and the many other studies showing that spanking toddlers is harmful. suggest that media and educational programs explicitly focused on not hitting toddlers, and making clear the benefits of avoiding spanking could help bring about a national increase in mental ability, as well as many other benefits that are discussed in this book. That includes less crime because, as Farrington and Welsh (2006) conclude from their own and other longitudinal studies, low intelligence is one of the leading individual level risk factors for crime. It may be one of the mechanisms that explains the links between spanking and crime shown in Chapters 12 through 15 and 19.
11 College Graduation
One of our previous studies analyzed a large and nationally representative sample of American adults and found that the more spanking they experienced as a child, the lower the probability of being in the top fifth of the U.S. occupational and economic achievement distribution (Straus & Gimpel, 2001 ). This relationship remained valid even when controlling for other family characteristics that overlap with parents spanking, such as the educational level and race or ethnic group of the parents. That is, spanking made a unique contribution to lowering a child's chances of being in the top fifth. Many people thought the results were implausible, perhaps in part because that study provided no empirical data on the processes that explain why spanking is related to lower occupational achievement and income. The previous chapters in this book suggest some of the possible processes, including impulsivity (Chapter 7), lower mental ability (Chapter 10), an increased probability of antisocial behavior and delinquency (Chapters 6, 7, and 8), and crime as an adult (all the chapters in Part IV).
Anita K. Mathur is the coauthor of this chapter.
Depression is also likely to interfere with academic achievement. The next chapter shows that, as in at least 14 other studies, spanking is associated with an increased probability of depression. Examples of such studies include (Afifi et al., 2006; Bordin et al., 2009; Leary, Kelley, Morrow, & Mikulka, 2008; Turner & Muller, 2004). The study in Chapter 9 found that spanking is associated with an increased probability of risky sex. This is relevant because risky sex as a high school student means an increased probability of a teenage birth, which in turn risks completion of high school and of going on to higher education, especially for girls. Each of these adverse effects of spanking is related to low academic achievement. Therefore, it seems plausible that one of the reasons spanking is associated with lower occupational and economic achievement is that spanking reduces the probability of college graduation. The research reported in this chapter examined that issue. Specifically, we present data on the following questions:
* Is spanking associated with a lower probability of college graduation among two large and nationally representative samples of U.S. adults?
* Does that relation of spanking to college graduation apply to both men and women?
* Has the relationship of spanking to college graduation become stronger or' weaker during~ a 10-year period in which there was a large increase in the percent of the U.S. population who attended college?
* Does the relation of spanking to college graduation apply after taking into account four family characteristics that might be the underlying cause: the education of the parents, race or ethnic group, violence between the parents, and the age of the participants in the study?
We did not find any study of the relation of spanking to college graduation. However, two studies are relevant. A study by Bodovski and Youn (2010) of a large, nationally representative sample of U.S. elementary school students measured spanking when the children were in kindergarten. They found that spanking in kindergarten was associated with lower 5th-grade math achievement even after controlling for many other variables such as parental depression, parental warmth, and racial or ethnic group. Poor academic achievement in 5th grade is associated with a lower probability of completing higher education. A longitudinal study by Margolin, Vickerman, Oliver, and Gordis (2010) found the spanking was associated with an increased probability of academic failure in the 5th grade. This study is particularly important because it is longitudinal. Many parents spank when a child gets failing grades in school in the hope that fear of being spanked again if there is no improvement will motivate the child to work harder and avoid failing again. This study shows that, although failing in school may have prompted some of the incidents of spanking, the effect was to increase rather than decrease the chances that the child will have failing grades again. These two studies are relevant to the issue of the relation of spanking to college graduation because the same processes that result in spanking being related to poor academic achievement in 5th grade, could also apply to lower educational achievement later in life.
A college degree is a direct and strong predictor of high occupational level and income. As a consequence, it is important to determine if spanking is related to a reduced probability of graduating from college. If so, not graduating from college could be another effect of spanking that could help explain why our previous study found that spanking is associated with lower economic and occupational achievement. A college degree is particularly important in a post-industrial society because it is almost a prerequisite for being in the top fifth of the income distribution. If college gradation is virtually a prerequisite, a crucial step in the process of understanding the link between corporal punishment and economic achievement is determining whether corporal punishment is related to the probability of college graduation. We, therefore, tested the hypothesis that the more spanking experienced, the lower the probability of college graduation.
Sample and Measures
Sample
We used data on two large and nationally representative samples to test the hypothesis just stated: the 1975 and 1985 National Family Violence Surveys. The 1985 survey is briefly described in Chapter 4. Both studies are described in more detail in Straus and Gelles (1986). Testing the hypothesis with two different samples a decade apart can provide greater confidence in the findings than either one by itself in the results, if the results are similar.
The data for this study are 25 and 35 years old. Does that mean they are outdated and not worth analyzing? They certainly are outdated as far as the percent of men and women who graduate from college. The percentages are now much higher. But the purpose of this study is not to find out what percent of men and women graduate. That is easily found in the annual Statistical Almanac of the United States. Rather, the purpose of the study is to test the theory that spanking is associated with a lesser chance of graduating from college. If the results show that spanking was related in 1975 and again in 1985, that is likely to still be the case.
Measure of Corporal Punishment
The extent of corporal punishment was assessed by asking the adults who participated in the survey, "I'd like to ask you about your experiences as a child. Thinking about when you were a teenager, about how often would you say your mother or stepmother used physical punishment like spanking or slapping or hitting you? Think about the year in which this happened the most." The response categories were never, once, twice, 3 to 5 times, 6 to 1 0 times, 11 to 20 times, and more than 20 times. The question was repeated for corporal punishment by fathers. The responses to the two questions were added to obtain a measure of how many times the respondent experienced corporal punishment. This variable was then grouped into seven categories ranging from never to 30 or more times.
Asking about spanking or slapping in the early teenage years, however, means depending on the ability and the willingness of the study participants to recall these events. However, there is empirical evidence indicating that adults' recall of events in childhood can provide a valid measure (Coolidge, Tambone, Durham, & Segal, 2011; Fisher, Bunn, Jacobs, Moran, & Bifulco, 2011; Morris & Slocum, 201 0). Another limitation is that the measure is about spanking during the teenage years. This raises the possibility that the data would apply to only a small and atypical group of children who were being hit at as teenagers. That turns out to not be the case. More than one half of the respondents reported being hit as a teenager. This percentage is roughly consistent with what we reported in the chapter on spanking in the United States (Chapter 2) for a national sample of U.S. parents. Almost 40% of the parents of children age 13 and 14 were still spanking or slapping. The percentage of the study participants for this chapter who were hit when they were 13 and 14 is very likely to be higher than that because they represent earlier generations than in the chapter on spanking nationwide (Chapter 2). For those generations, being' hit by their parents when they were age 13 to 14, rather than being a small and possibly deviant group of families, probably represent a typical child of their generation. In fact, for those interviewed in 1975, almost two out of three reported one or more incidents of being hit by parents when they were in their early teenage years (Straus et al., 2006).
A decade later in 1985 when the second survey for this chapter was conducted, the prevalence rate had dropped substantially. But even the 1985 figures indicate that just over one half of that sample of adults in the United States that year (52%) said they had been spanked or slapped as a teenager (Straus, 2001a). Moreover, among those who were hit during their teenage years, it was not usually a rare event. The average was 8 times in the year they were 13 and 14, and the median was 5 times (Straus, 2001a).
Men and Women
Women not only have now achieved parity with men in college enrollment but, in recent years, have considerably higher rates of college enrollment and degrees granted than men (U.S. Bureau of the Census, 2011, Table 226). The two national samples for this study, however, were interviewed in 1975 and 1985 and, on average, were college students years before they were interviewed. They, therefore, represent earlier generations of college students when more males than females graduated from college. In either case, if spanking makes a difference in college graduation and if boys are hit more than girls, it is important to examine the relation of spanking to college graduation separately for men and women. We, therefore, repeated the analysis for men and women in the 1975 and the 1985 samples, making four tests of the theory that spanking is associated with a reduced probability of college graduation.
College Graduation
Those who participated in the study and had 4-year college degree or higher were coded as 1 and all others as 0. Among the 1975 sample, 28% of the men and 16% of the women were college graduates. For the 1985 sample, it was 31% of the men and 20% of the women who were college graduates.
Are Other Variables the Real Cause?
When we tested the hypothesis that the more spanking, the lower the probability of graduating from college, we took into account four other variables that are known to affect the probability of graduating from college. Because these four variables are also correlated with spanking, they could be the underlying cause of a relationship between spanking and completing college. One of the four controls was violence between the parents of the study participant. As shown in Chart 5.7 in Chapter 5, which addresses violence approval and spanking, and in Straus, Gelles, and Steinmetz (2006), Silverstein et al. (2009), and Taylor, Lee, Guterman, and Rice (2010), parents who hit each other are more likely to hit their children. In addition, witnessing violence between parents has been found to be associated with a number of social and psychological problems (Davies, DiLillo, & Martinez, 2004; Holden, Geffner, & Jouriles, 1998; Kitzmann, Gaylord, Holt, & Kenny, 2003; Straus, 1992). As a consequence, unless violence between the parents is controlled, the results for college graduation might merely be showing the harmful effect of growing up in a family where the parents are violent to each other.
Another variable that needs to be controlled is the age of the respondent. The percent of the population who attend college has been increasing each decade and the percent of school-age children who were spanked has been decreasing. Unless the age of the respondent is held constant, this combination could result in statistics seeming to show that spanking is associated with a lower chance of graduating from college, but which really only shows the effects of these two trends.
Third, there is a large amount of research showing that the more educated the parents, the higher the chances the child will complete college. Better-educated and higher socioeconomic status parents also spank less (see Chapter 2 and Taylor, Lee et al., 2010). As a consequence, to rule out the possibility that a relationship of spanking to college graduation is really the result of bettereducated parents spanking less, we controlled for parent education. Finally, race or ethnic minority parents are more likely to spank their children are less likely to complete college. As a consequence, both the race and ethnic group .of the parents was controlled in these analyses.
Other Measures
Violence between parents. Study participants were asked, "Now thinking about the whole time you were a teenager, were there occasions when your father/stepfather hit your mother/stepmother or threw something at her." If the response was yes, respondents were asked how often that happened and presented with the same response categories as was used for the question on corporal punishment. The same questions were repeated for whether the mother had hit the father. A code of 1 was assigned if either parent was reported as having hit the other and zero if not. Depending on the year of the survey and whether the information was obtained from interviews with male or female study participants, from 13% to 16% of the participants reported one or more physical assaults between their parents.
Age. Age was controlled because, as explained previously, the parents of older study participants were part of a generation that was less likely to be college graduates and more likely to spank.
Parent's education. The education of the parents of the study participants is related to whether the study participants complete a college degree. Education of parents is also related to the frequency of using corporal punishment,' as shown in the chapter on spanking in the United States (Chapter 2) and by GilesSims et al. (1995). The 1975 survey included information on the education of each parent of the study participants. A two-item parental education index was computed by adding the number of years of education completed by the father and mother of the participants in the 197 5 survey.
Ethnic group. Disadvantaged ethnic groups have substantially lower average education and income than Whites, making it important to control for ethnic group when relating corporal punishment to college graduation. The 1975 study did not include enough minority group respondents for a reliable analysis of ethnic groups. However, the 1985 study oversampled Blacks and Hispanic Americans. This permitted taking minority ethnic groups into account by coding. Blacks, Hispanic Americans, and Native Americans were coded as 1 and Whites as 0. Seventeen percent of the study participants identified themselves as one of these racial or ethnic groups.
Corporal Punishment and College Graduation
The left side of Chart 11.1 shows the relationship between spanking and the percent who graduated from college in the 1975 sample, and the right side shows this relationship in 1985. The dotted lines are the actual graduation rates, and the solid lines are the relationship predicted on the basis of the statistical analysis that controlled for other possible causes of educational attainment such as the educational level of the parents. (See the Appendix for the variables controlled and their effects.) It was important to control for these variables because, to take the example of parent's education, more educated parents spank less and the children of more educated parents do better academically. Despite these controls, both the 1975 and the 1985 charts show that the more spanking, the lower the percent who graduated from college. Both charts also show that this relationship applies to both men and women.
Although spanking was associated with a lower probability of graduating from college for both men and women, in 1975 the adverse effect was greater for men than for women. For men, each increase of one category of the spanking categories was associated with an 11% decrease in the probability of graduating from college, whereas for women, each increase in the spanking scale was associated with an 8% decrease. A decade later, although more men than women continued to attend college, there was no difference between men and women in the adverse effect of spanking on the chances of college graduation. For both men and women, each increase of one unit in spanking was associated with an 8% decrease in the probability of college graduation for both men and women. Perhaps the increase in gender equality, including more women attending college, had something to do with the adverse effects of spanking becoming more gender-equal.
Chart 11.1 The More Spanking as a Teenager, the Lower the Percent Who Graduated from College
Other Variables Related to College Graduation
Our focus in this chapter on the relation of spanking to college graduation does not mean that other variables are unimportant as determinants of college graduation. The other independent variables, although they were included in the analysis as controls, are also interesting in their own right. Moreover, Table All. I in the Appendix shows that some, such as the education level of parents and violence between the parents, have a much stronger relation to college graduation than spanking. Although other variables have a stronger relationship, controlling for them did not erase the adverse effect of spanking. Moreover, controlling for these other variables shows that spanking had a unique effect, over and above the effect of parent's education and violence between the parents. In addition, Appendix Table All.2 confirms what has been shown by many studies: That position in society, not just individual ability or family experiences such as spanking, is related to the probability of completing a college degree. Thus:
* Children from low-education families in these two studies, as in the nation as a whole, were much less likely to graduate from college. This applies to both men and women (see Appendix).
* Minority men in the 1975 survey were less than one half as likely as White men to have graduated from college. Minority women were about one third less likely than White women to have graduated from college. Thus, although minority women were more likely than minority men to graduate from college, compared with Whites, they were also less likely to be college graduates.
* Women, in this sample, as in the nation as a whole at the time the participants in this study were college age, had lower rates of college gradation than men.
* To our surprise, after controlling for the other variables, we found that violence between the parents was not related to college graduation in any of the four replications.
Summary and Conclusions
We tested the hypothesis that spanking is associated with a reduced chance of graduating from college in four different samples: nationally representative samples of men and women interviewed in 1975 and in 1985. All four analyses found that spanking was associated with a lower probability of college graduation. These analyses controlled for variables that could be the underlying real reasons spanking is associated with a lower probability of college graduation: educational level of the study participants' parents, ethnic group of the family in which the participant grew up, physical violence between the respondent's parents, and the age and sex of the study participant. The study found that after taking account of the effect of these control variables, spanking makes an additional independent contribution to predicting college graduation. Nevertheless, there could be other variables that need to be controlled to provide more confidence in the conclusion that spanking is associated with a reduced chance of college graduation.
It would be especially important to control for parental warmth and supportiveness and cognitive stimulation, as we have done in other chapters, because spanking might be just one way in which lack of parental involvement, warmth, and support manifest themselves. It could be a lack of warmth and support by parents who spank, rather than the spanking itself, that is the problem. We do not agree because, although harsh parenting and lack of warmth are correlated with spanking, it is not a strong correlation. Most warm, loving, and involved parents also spank. Probably most parents believe they do it out of love and concern for the child. In fact, for the sample of mothers studied to examine the effect of impulsive spanking (Chapter 7), there was a tendency for more nurturing mothers to spank slightly more than less nurturing mothers. Moreover, we controlled for parent support and warmth in four other chapters: those that studied the relation of spanking to child antisocial behavior (Chapter 6), impulsiveness (Chapter 7), delinquency (Chapter 8), and mental ability (Chapter 10). The results can be summarized by saying that, when data was available to control for positive parental behaviors such as explaining and warmth and supportiveness, the controls reduced the size of the link between spanking and child behavior problems; but spanking continued to have the harmful side effect studied. In addition to controlling for other variables, longitudinal studies are needed to provide a stronger basis for concluding that spanking is one of the causes of failing to complete a college degree. We were able to do longitudinal studies for three of the chapters in this book, and a number of other longitudinal studies are described in Chapter 19. The three longitudinal studies in this book found that spanking led to more antisocial behavior (Chapter 6), slower development of cognitive ability (Chapter 10), and an increased probability of crime as an adult (Chapter 15). The study in this chapter, although it is not longitudinal, is consistent with those studies in finding that spanking is associated with an increased probability of a harmful effect. Other chapters suggest some of the processes that explain why spanking is related to lower educational attainment, such as antisocial behavior as a child (Chapter 6), impulsiveness and lack of self-control (Chapter 7), and slower development of cognitive ability (Chapter 1 0). In addition, still other research has found that spanking is associated with alienation (Straus & Gimpel, 2001) and depression (Chapter 12 and DuRant, Getts, Cadehead, Emans, & Woods, 1995; Straus, 2001a). Finally, college graduation requires a self-directed commitment to learn, but in our opinion, spanking teaches obedience more than self-direction.
Whatever the processes that link spanking to a lower probability of completing higher education, the results in this chapter are another of the many cruelly ironic aspects of spanking. Many parents spank if the child comes home with poor grades, as in the example of a 13-year-old boy in Texas who did badly on a test. His teacher told the boy he was going to paddle him, in the same fashion as his father did to him. He then hit the boy twice with a large wooden paddle (Harrison, 2011 ). The irony of course is that the paddling probably decreased, rather than increased, the chances that the boy will do better on the next test. Although this book is about spanking by parents, corporal punishment by teachers probably has parallel adverse effects. A study by Tal war, Carlson, and Lee (20 11) compared children in the first grade in two private schools in a West African country. Both schools served families who lived in the same urban neighborhood, and the parents were largely civil servants, professionals, and merchants. One school used corporal punishment and the other did not. They found that the children in school that used corporal punishment performed significantly worse in tasks involving executive fimctioning-psychological processes such as planning, abstract thinking, and delaying gratification. These are all characteristics needed for college graduation.
Both the research in this chapter and the results in previous chapters on the relation of spanking to characteristics that are inimical to completing a college degree, make it reasonable to conclude that spanking is associated with a decreased probability of college graduation. It is important to also keep in mind that a decreased probability means just that. That is, this adverse side effect actually affects only a minority of spanked children, just as only a minority of heavy smokers die of a smoking related disease (see the section on Risk F acto:r:s Relationships in Chapter 1 ). It is important to also keep in mind that this study also provided data on some many other life experiences and characteristics that influence academic achievement. Spanking is only one of these harmful experiences and not nearly as important as being the child of low-education parents or parents who physically assault each other. If we want more children to complete higher education, we need to move toward a society in which everyone is educated and no one assaults their partner and no one assaults their child under the guise of discipline.
With those qualifications in mind, our finding that spanking is associated with a decreased probability of graduating from college has important theoretical and practical implications. One theoretical implication is based on studies that show that low-education parents tend to spank more often than middle class parents (see Chapter 2 and Giles-Sims et al., 1995; Socolar & Stein, 1995; Straus, 2001a, Chart 4). Ironically, although parents spank to raise children who will follow rules and who will do well in school, the results of our research suggest that spanking may be a mode of discipline that more often has the opposite effect. If that is correct, it is as we said a cruel irony because more spanking, and with it less use of cognitive modes of correction, by loweducation parents makes it particularly difficult for the children of the poor to attain a level of education that will enable them to surmount their difficult life circumstances and the economic and social barriers to upward social mobility.
In the final months of his life Professor Straus handed out free copies of this book and took other actions to indicate that he wanted to make it available, presumably because he recognized that his work could help teach how to reduce violence if made available freely. Therefore I'm posting the book in four parts.
The Primordial Violence 2014-1
The Primordial Violence 2014-2
The Primordial Violence 2014-3
The Primordial Violence 2014-4
Wikipedia: John Rosemond Others believe in striking while the iron is hot. John Rosemond, the author of best-selling books on child rearing says that he believes in "spanking as a first resort; spanking in anger" (Rosemond, 1994a).
No comments:
Post a Comment