Critique of Anti-Spanking Study
This is a critique of the study by Straus, Sugarman, and Giles-Sims (1997) Spanking by Parents and Subsequent Antisocial Behavior of Children in the Archives of Pediatrics and Adolescent Medicine, and its implications for the broader topics of nonabusive spanking and parental discipline by Robert E. Larzelere.
THE GOOD NEWS (for anti-spanking advocates): This is the first study that finds any detrimental child outcome of nonabusive or customary physical punishment by parents using a design that would not also tend to find detrimental outcomes of most alternative discipline responses (or, for that matter, a wide variety of other interventions, including marital therapy, psychotherapy, or cancer treatment).
Larzelere's review of the literature (Pediatrics, 1996 (Oct.), pp. 824-828, 858) found that the strongest previous design with "detrimental" child outcomes was a longitudinal association without controlling for child misbehavior at Time 1. If managed care were using the same reasoning, they would deny payments for marital therapy because those who received marital therapy last year were not now doing as well maritally (on the average) as those who had no need of such therapy last year. This problem is called the intervention selection bias (i.e., the tendency of any intervention for an enduring problem to be positively associated with subsequent problems associated with the original problem being intervened for). The Straus et al. study is the first to correct for that problem and find any detrimental effect on children.
THE BAD NEWS: The only thing that Straus et al. (1997) have proven is that spanking 6- to 9-year-olds at the rate of 156 times a year has a small, but detrimental effect (accounting for 1.3% of subsequent variation in anti-social behavior). Most children spanked from 1 to 25 times annually were in their most-improved group (because they weren't spanked in the particular week they asked about in the survey; see their Figure 1). Yet media reports are reporting the exaggerated claim that this study has proven that even one spanking has detrimental outcomes. Instead, their results are consistent with Larzelere's (1996) review, which reported that spanking 6- to 9-year-olds once a week was more detrimental than was less frequent spanking.
The evidence for spanking from 26 to 155 times annually is more ambiguous. I would like to see a pairings comparison between the subsequent levels of antisocial behavior comparing those spanked 0 vs. 1 and between those spanked 1 vs. 2 times per week. Straus has not done that, but the only significance tests included the most extreme high frequency group. If spanking is as detrimental as it is being made out to be, it should be easy to get a statistically significant difference between those spanked at the rate of 104 times annually vs. those spanked from 0 to 25 times annually. Given a total variance accounted for of 1.3%, I don't think Straus et al. can obtain that even with their overall sample size of over 800.
THE OTHER NEWS: The first review of child outcomes of nonabusive or customary physical punishment was published by Larzelere in Pediatrics in Oct. 1996). Dr. Diana Baumrind (Cal-Berkeley) summarized the implications in the first sentence of her published response: "a blanket injunction against spanking is not scientifically supportable" (p. 828). The review found 8 studies that, like Straus et al. (1997), distinguished parental effects on children from the opposite causal direction. All 8 studies found beneficial outcomes in children, generally of nonabusive spanking of 2- to 6-year olds, when used as a back-up for time out or reasoning.
Thus the evidence to date suggests that nonabusive spanking has generally beneficial effects on children under the following limited conditions at least:
Age: 2 to 6 years
How: 2 open-handed swats to the buttocks, leaving no bruise
How: Primarily as back-up for less aversive discipline responses (e.g., reasoning and time out). Using it as a back-up should make reasoning and time out more effective so that the spanking back-up can be phased out.
Who: by loving parents
Eight of the 9 best studies support this, and the 9th (Straus et al., 1997) has no evidence against it.
Immediately following the article by Straus et al. (1997) in the latest issue of Archives of Pediatrics and Adolescent Medicine is a study by Gunnoe and Mariner (1997). It improves upon the Straus et al. study in several ways, and reconciles the differential findings between their study and the other 8 best studies to date. Improvements include a longer time until outcome variable (5 years), a more representative sample, and using different informants (parent and child) for the spanking and the outcome variable. They also add some analyses to account for the differences in results between their study and Straus et al.'s.
Basically they find that the effects of spanking vary from significantly beneficial effects to significantly detrimental effects according to four factors: Age, sex, race, and marital status. At one extreme, spanking frequency increases the subsequent level of antisocial aggression for 8- to 11-year-old white sons of single mothers. At the other extreme, spanking frequency decreases the subsequent level of antisocial aggression for 4- to 7-year-old Black daughters in intact families. Other combinations in between have SEM coefficients that vary from the one extreme to the other, with the most beneficial coefficients for Blacks and for younger children and mostly near-zero coefficients for Whites. This illustrates why African-Americans distrust social science.
This is now the third adequate study that has looked separately at the outcomes of spanking in African-Americans compared with European-Americans. The three are unanimous that the effects are different by ethnic group and that the outcomes are generally beneficial to African-American children with the exception of those 6- to 9 year-olds spanked at the rate of 156 times a year (Straus et al., 1997; and Deater-Deckard et al., 1997, Devel Psych are the other two studies).
The effects of nonabusive spanking have to be considered in the context of the effects of alternative discipline responses and the broader parenting context. My research program for the last 11 years has sought to find alternatives that are more effective than is spanking, especially for 2- and 3-year-olds. The effectiveness of the most typically recommended alternatives depend on being backed up by more aversive discipline tactics such as spanking. For example, my presentation at APA (Aug. 1997) found that reasoning alone (i.e., without accompanying punishment) works with 2- and 3-year-olds only for parents who back reasoning up with punishment at least 1/10 of the time. That punishment could be physical or, preferably, nonphysical (time out or withdrawal of privileges). For the most noncompliant preschoolers, Mark Roberts' series of studies have shown that time out's effectiveness depends upon being backed up with either a 2-swat spank or a brief room isolation, with one working better for some preschoolers and the other better in come cases.
Another caution concerns the case of Sweden: Since they outlawed parental spanking in 1979, child abuse has increased at least 4-fold and teenage violence has increased at least 6-fold. Being a less violent country than the USA, perhaps they can handle a 6-fold increase in teenage violence. The United States cannot. An SRCD paper by Palmerus and Scarr (1995) suggested a possible reason: Swedish parents are less likely to use reasoning and behavior modification techniques than are USA parents, and they are more likely to use physical restraint and coercive verbal techniques. This is consistent with my research and Mark Roberts's in that the most positive use of a nonabusive spanking with 2- to 6 year-olds is to back up less aversive discipline tactics so that they become sufficient and effective by themselves as the child gets older. Loss of the spank backup, as in Sweden, may lead to less effective use and thus less use of the very alternatives being recommended as preferable alternatives to spanking.
As Gunnoe & Mariner and Patterson have noted in 1997, there is a lot about parental discipline that we do not understand adequately yet. I have written this critique primarily as a caution against making too big of a leap from the first good evidence that parental spanking leads to detrimental child outcomes under some conditions.
Robert E. Larzelere
Boys Town, Nebraska
Spank With Love - controversial and almost pornographic
Biblical Brickbats - using scripture on both sides of the spank divide
The erotic angle - a legitimate research concern: sexual excitement in adult entertainment
Both Sides Wrong? - a potpouri of spanking thoughts pro, con, and corpun
FastCounter by bCentral