American Psychologist
© 2018 American Psychological Association
0003-066X/18/$12.00
2018, Vol. 73, No. 5, 626 – 638
http://dx.doi.org/10.1037/amp0000327
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
The Strength of the Causal Evidence Against Physical Punishment of
Children and Its Implications for Parents, Psychologists, and Policymakers
Elizabeth T. Gershoff
Gail S. Goodman
University of Texas at Austin
University of California, Davis
Cindy L. Miller-Perrin
George W. Holden
Pepperdine University
Southern Methodist University
Yo Jackson
Alan E. Kazdin
Pennsylvania State University
Yale University
The question of whether physical punishment is helpful or harmful to the development of children
has been subject to hundreds of research studies over the past several decades. Yet whether causal
conclusions can be drawn from this largely nonexperimental research and whether the conclusions
generalize across contexts are issues that remain unresolved. In this article, the authors summarize
the extent to which the empirical research on physical punishment meets accepted criteria for
causal inference. They then review research demonstrating that physical punishment is linked with
the same harms to children as is physical abuse and summarize the extant research that finds links
between physical punishment and detrimental outcomes for children are consistent across cultural,
family, and neighborhood contexts. The strength and consistency of the links between physical
punishment and detrimental child outcomes lead the authors to recommend that parents should
avoid physical punishment, psychologists should advise and advocate against it, and policymakers
should develop means of educating the public about the harms of and alternatives to physical
punishment.
Keywords: physical punishment, spanking, discipline, parenting, causality
When experimental studies are not feasible or ethical,
when are alternative research approaches sufficient to conclude that a causal relation between variables exists? At
what point can that evidence be used to inform individuals,
psychologists, and even public policy? These fundamental
questions are illustrated in the controversy over the physical
punishment of children. Physical punishment refers to the
use of physical force to cause a child to experience some
degree of pain or discomfort with the intention of modifying
the child’s behavior (United Nations Committee on the
Rights of the Child, 2007). Spanking is the most common
form of physical punishment and typically involves an adult
striking a child on the bottom with an open hand. Decades
of research have consistently linked physical punishment
with risks of harm to children (Gershoff & Grogan-Kaylor,
2016a), yet over 80% of American parents continue to
spank their children (Gershoff, Lansford, Sexton, DavisKean, & Sameroff, 2012).
Why do parents persist in a behavior that may increase the
risk of harm to their children? A primary reason is that they
believe it to be effective at improving child behavior and are
unaware of its potential harms to children (Holden, Miller,
& Harris, 1999), in part because they themselves were
spanked as children and believe that they “turned out OK.”
These beliefs are not often challenged by the media, which
typically frames stories about spanking as a “debate” among
Elizabeth T. Gershoff, Department of Human Development and Family
Sciences and the Population Research Center, University of Texas at Austin;
Gail S. Goodman, Department of Psychology, University of California, Davis;
Cindy L. Miller-Perrin, Social Science Division, Pepperdine University;
George W. Holden, Department of Psychology, Southern Methodist University; Yo Jackson, Department of Psychology, Pennsylvania State University;
Alan E. Kazdin, Department of Psychology, Yale University.
The lead author’s time writing this article was supported by grant
P2CHD042849 (PI: Umberson) from the Eunice Kennedy Shriver National
Institute of Child Health and Human Development.
This article is based on a report written by the authors for the Task Force
on Physical Punishment of Children sponsored by APA Division 7: Developmental Psychology and APA Division 37: Society for Child and
Family Policy and Practice. We thank the following colleagues who
provided constructive feedback on that report: Ronald Dahl, Nancy Eisenberg, David Finkelhor, Milton Fuentes, Joan Grusec, Michael E. Lamb,
Jennifer Lansford, Mary Ann McCabe, and Anita Thomas.
Correspondence concerning this article should be addressed to Elizabeth
T. Gershoff, Department of Human Development and Family Sciences,
University of Texas at Austin, 108 East Dean Keeton Street, Stop A2702
Austin, TX 78712. E-mail: liz.gershoff@austin.utexas.edu
626
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
627
APA Divisions 7 (Developmental Psychology) and 37 (Society for Child and Family Policy and Practice). The Task
Force (active 2013–2016) was charged with summarizing
the literature on spanking and physical punishment, in part
to inform APA’s Council of Representatives as it decides
whether to approve a resolution opposing all physical punishment of children, including that by parents. As of early
2018, APA had yet to consider such a resolution.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
What Is the Evidence About Causality Regarding
Physical Punishment?
Elizabeth T.
Gershoff
researchers without adequately vetting the strength of the
evidence (e.g., Moyer, 2016). It is also the case that not all
psychologists refute the notion that spanking is ineffective
and inadvisable. In a 2016 survey of 843 members of the
American Psychological Association (APA), the majority of
these psychologists agreed that spanking children was
harmful and would not recommend it to parents, yet substantial minorities held favorable views. For example,
nearly 30% of survey respondents did not agree that spanking is harmful to children, 28% did not agree that spanking
is a form of violence against children, 17% did not agree
that spanking is a bad disciplinary technique, 14% said they
advise parents with whom they work to spank with a hand
occasionally, and 15% to 36% were unsure whether research has linked spanking with particular negative outcomes (Miller-Perrin & Rush, 2018). There are also a few
psychologists who argue that, because experiments of
spanking are, for the most part, not possible, we can never
definitively conclude that spanking causes harm to children
(Larzelere, Gunnoe, Roberts, & Ferguson, 2017). In short,
there remains an information gap between what researchers
in psychology, medicine, and other disciplines have confirmed about the harms linked to spanking and what the
majority of parents and a substantial minority of psychologists continue to believe.
The goal of this article is to fill the information gap by
directly addressing critiques about whether the relevant
research warrants causal conclusions and to thereby illustrate a case in which a large body of largely nonexperimental research can, when taken together, support causal inference. This article developed out of a 2016 report from the
Task Force on Physical Punishment of Children created by
Ethical concerns have meant that very few studies have
examined physical punishment with experimental designs:
Researchers cannot randomly assign children to “hit” and
“no hit” conditions, nor can they randomly assign parents at
the birth of their child to use physical punishment or not.
The fact that the body of research on physical punishment is
largely correlational leads to two key challenges to causal
conclusions. The first is the classic “chicken or egg” problem: Are children’s behavior problems the consequence or
the cause of physical punishment? Some critics of the
physical punishment literature have termed this “intervention selection bias”: Individuals who most need a treatment
are the ones most likely to get it, and consequently the
intervention can be linked with more problems, at least
initially (Larzelere, Kuhn, & Johnson, 2004). It is true that
children with behavior problems elicit more physical punishment from their parents over time (Gershoff et al., 2012),
and so research that measures physical punishment months
or years before the child outcomes is key to establishing
causality.
A second potential problem with physical punishment
research is selection bias, which is the possibility that links
between physical punishment and child outcomes are determined in part by parent or child characteristics that predict
whether parents use, or in the terminology of economists
“select into,” physical punishment (Duncan, Magnuson, &
Ludwig, 2004). This possibility is often referred to as the
“third variable” explanation— other variables, such as the
parent’s level of warmth, may predict both the amount of
physical punishment and the level of the child outcome in
question, and it is this third variable that may account for the
association between them.
To guide our review of research on physical punishment,
we relied in part on Sir Austin Bradford Hill’s (1965)
classic criteria that have been used to establish causal connections such as those between smoking and lung cancer
and between certain occupational hazards and pulmonary
disease (Fedak, Bernal, Capshaw, & Gross, 2015). We used
seven of Hill’s criteria (two of his criteria are not necessary
for causality, namely, his criteria of specificity of effects
and of analogy) and combined them with the three criteria
identified by Shadish, Cook, and Campbell (2001) into five
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
628
GERSHOFF ET AL.
hypothesis goes against common knowledge does not mean
we should dismiss it.
Research on physical punishment is coherent with the
hypothesis that it predicts detrimental outcomes for children. Decades of correlational research have linked more
physical punishment with more problematic behavior in
children (Gershoff, 2002; Gershoff & Grogan-Kaylor,
2016a). Physical punishment is a form of hitting (Gershoff,
2013), and given that other forms of hitting are already
acknowledged as harmful to children (bullying; American
Psychological Association [APA], 2004) and adults (violence between romantic partners; APA, 1999), the hypothesis that physical punishment should predict detrimental
outcomes for children is consistent with our society’s beliefs about the negative outcomes that can be attributed to
hitting.
Gail S.
Goodman
criteria for establishing causal relations between a predictor
and an outcome (Hill’s criteria are italicized): (a) the hypothesized causal pathway must be plausible and have
coherence with existing facts about the predictor and the
outcome; (b) the behavior must be strongly and consistently
correlated with the outcome of interest; (c) the behavior
must temporally precede the outcome; (d) a gradient in the
association between the predictor and the outcome is observed; and (e) experiments or statistical methods establish
that the association between the two cannot be attributed to
spurious factors. We will summarize relevant research on
physical punishment for each criterion.
Plausibility and Coherence
Hill (1965) observed that it would be “helpful” if the
hypothesized causal pathway is plausible and that the observed data are coherent with what is generally known about
the predictor and outcome. The hypothesis that parents’ use
of physical punishment should affect children’s behavior is
plausible and indeed widely accepted: Parents use physical
punishment under the working hypothesis that their action
(physical punishment) will cause a reaction (change in
behavior) in the child. However, there is less agreement
about the valence of the effect. Parents’ beliefs about effective parenting, their own histories with physical punishment, and their cultural milieu lead many parents to assume
that physical punishment will yield positive changes in the
child. Yet if we accept as plausible that parents’ use of
physical punishment can affect children’s behavior at all,
we must be willing to accept that any effects could be
positive, negative, or neutral. As Hill noted, just because a
Strength and Consistency
Hill (1965) urged researchers to judge the strength of
potential causal relations; the direction of the association is
of course also important to assess. For years, the field
judged strength of effect sizes, which statistically characterize the association between two variables, according to
criteria for small, medium, and large effect sizes suggested
by J. Cohen (1988) in his writings on power analysis. Yet on
their own, effect sizes can be large but not practically
significant, as in the case of studies with small samples and
large standard errors. Confidence intervals (CIs) around
effect sizes are needed, as they help researchers establish the
amount of error in the effect size and indicate whether an
effect size is statistically different from zero.
Meta-analysis is an ideal tool for determining the general
strength and direction of the associations between key variables. We can obtain an overall sense of the strength,
direction, and consistency of the associations between physical punishment and child outcomes through the findings of
five meta-analytic reviews: Gershoff (2002; 88 studies),
Paolucci and Violato (2004; 70 studies), Larzelere and
Kuhn (2005; 26 studies), Ferguson (2013; 45 studies), and
Gershoff and Grogan-Kaylor (2016a; 75 studies). There is
some, but not entire, overlap across the meta-analyses in the
studies used; for example, the most recent meta-analysis
(Gershoff & Grogan-Kaylor, 2016a) used 26% of the studies used in Gershoff, 23% of the studies in Paolucci and
Violato, 42% of the studies in Larzelere and Kuhn, and 24%
of the studies in Ferguson. The Gershoff and Grogan-Kaylor
meta-analysis reported a mean effect size of d ⫽ .33 (95%
CI [.29, .38]) that characterized all 111 effect sizes and the
160,927 unique children represented by those effect sizes.
The strength of the association between physical punishment and child outcomes is statistically significant and
moderate in strength.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
Cindy L. MillerPerrin
When the direction of the associations is considered,
spanking has been significantly correlated with the following child outcomes, almost all in a detrimental direction:
lower long-term compliance (Gershoff, 2002; Gershoff &
Grogan-Kaylor, 2016a); more aggression (Gershoff, 2002;
Gershoff & Grogan-Kaylor, 2016a; Larzelere & Kuhn,
2005); more behavior problems (Ferguson, 2013; Gershoff,
2002; Gershoff & Grogan-Kaylor, 2016a), but not in Larzelere and Kuhn (2005); more mental health problems (Ferguson, 2013; Gershoff, 2002; Gershoff & Grogan-Kaylor,
2016a); lower cognitive performance (Ferguson, 2013; Gershoff & Grogan-Kaylor, 2016a); lower parent– child relationship quality (Gershoff, 2002; Gershoff & GroganKaylor, 2016a); and higher risk for physical injury or abuse
(Gershoff, 2002; Gershoff & Grogan-Kaylor, 2016a). Research on immediate compliance has been equivocal. Physical punishment was associated with higher rates of immediate compliance in two meta-analyses (Gershoff, 2002;
Larzelere & Kuhn, 2005); however, when Gershoff and
Grogan-Kaylor (2016a) reanalyzed these studies to take into
account differences in initial rates of compliance between
the spanked and nonspanked groups, the association was no
longer statistically significant.
It is one thing to know that physical punishment is, on
average, associated with detrimental child outcomes; it is
another to know that this association is consistent across
dozens or hundreds of examinations of it. Hill (1965) encouraged researchers to consider the question, “Has it been
repeatedly observed by different persons, in different
places, circumstances and times?” (p. 296). In other words,
the strength and direction of the finding must be replicated
629
across studies but also across a range of populations, locations, and cohorts.
The research on physical punishment demonstrates both
aspects of consistency. Physical punishment has been linked
with detrimental child outcomes consistently across studies.
In the most recent meta-analysis, a total of 79 of all the
effect sizes (71%) were statistically significant, and 99%
(all but one) were in the direction of an unfavorable outcome (Gershoff & Grogan-Kaylor, 2016a). In other words,
just under three quarters of all effect sizes were statistically
significant and indicated physical punishment was associated with detrimental child outcomes. Links between physical punishment and problematic child outcomes have also
shown consistency across samples. Using meta-regression,
Gershoff and Grogan-Kaylor (2016a) found that spanking
was linked with detrimental outcomes to the same extent in
the United States and in countries other than the United
States (13 countries, constituting 30% of studies). In a group
of studies that included cohorts of children from the 1950s
through the 2000s, they also found that the mean effect size
was robust to the age of the children and to the operationalizations of spanking (e.g., ever; in last week). The literature on physical punishment is thus characterized by a
remarkable level of consistency across studies, populations,
settings, and cohorts.
Temporal Precedence
The best way to ensure that physical punishment precedes
child outcomes in a statistical model is through a longitudinal design in which spanking is used to predict subsequent
child outcomes. Although longitudinal designs meet the
temporal precedence criterion, they do not necessarily address selection bias. A particularly important selection factor, as mentioned earlier, is a child effect of children’s initial
behavior: Children’s difficult behaviors could elicit physical
punishment from their caregivers. If the “true” association
between physical punishment and child outcomes is either
entirely or partially a child effect, an association between
physical punishment and child outcomes could be mistakenly viewed as a parent effect. With this selection concern
in mind, nine longitudinal studies have controlled for children’s initial behavior problems. All of these studies determined that parents’ use of physical punishment predicted
increases in children’s externalizing behavior problems over
time, even after race, gender, and family socioeconomic
status had been statistically controlled (e.g., Grogan-Kaylor,
2005; Larzelere, Cox, & Smith, 2010; Olson, Lopez-Duran,
Lunkenheimer, Chang, & Sameroff, 2011).
Even stronger are research designs that model both the
longitudinal parent effect (physical punishment to child
behavior) and the longitudinal child effect (child behavior
eliciting more physical punishment) simultaneously, such
that each statistically controls for the other; these are known
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
630
GERSHOFF ET AL.
George W.
Holden
as cross-lagged panel designs. Using this design, spanking
at an initial time point predicted increases in behavior
problems from Ages 1 to 3 (Berlin et al., 2009), 3 to 5
(Maguire-Jack, Gromoske, & Berger, 2012), 5 to 8 (Gershoff et al., 2012), and 12 to 14 as well as 14 to 16 (Wang
& Kenny, 2014). Three of these four studies used large and
nationally representative samples (N ⫽ 2,573: Berlin et al.,
2009; N ⫽ 11,044: Gershoff et al., 2012; N ⫽ 3,870:
Maguire-Jack et al., 2012).
These three approaches to determining temporal order
effects generally find that spanking predicts a worsening of
child behavior over time, with only a few exceptions that
find no relation between spanking and changes in child
behavior (e.g., Lansford, Wager, Bates, Pettit, & Dodge,
2012). It is important to note that in none of the cross-lagged
studies did physical punishment predict a decrease in child
problem behaviors over time while controlling for the child
elicitation effect. If selection entirely explained the links
between physical punishment and children’s behavior,
physical punishment would have significantly predicted improvements in child behavior. Longitudinal studies have
overwhelmingly found that physical punishment predicts
deteriorations in child behavior over time. That said, it is
worth noting that in the Gershoff and Grogan-Kaylor
(2016a) meta-analysis, the mean effect size for longitudinal
studies was not significantly different from the mean effect
size for correlational studies, indicating that both methods
reached the same conclusion.
Gradient
A gradient in the association between two variables is
observed when a one-unit change in the predictor leads to a
regular change in the outcome variable that is observed
along the full range of the predictor variable. Hill (1965)
noted that a gradient does not prove causality, but it does
strongly suggest researchers should look for a cause. In
medicine, a gradient is often referred to as a dose–response
relationship, and it can be linear (which is then also called
a step function) or curvilinear. For example, a gradient
could be illustrated by an increase in children’s externalizing scores by 2 points for every additional spank per month.
Whether there is a gradient in the associations between
physical punishment and child outcomes, and what form
that gradient takes, is an issue that underlies one of the
common arguments about physical punishment. Defenders
of physical punishment (Larzelere et al., 2017) argue that
the association is curvilinear, namely, that infrequent or
light spankings are not thought to be harmful to children and
that physical punishment is only linked with harm to children when it is frequent, harsh, or injurious. A few studies
have examined the shape of the association. A study of
parents in Chile (Ma, Han, Grogan-Kaylor, Delva, & Castillo, 2012) found that, compared with adolescents who
were never physically punished, externalizing behavior
problems were higher among adolescents who were even
infrequently subjected to physical punishment. Moreover,
compared with those adolescents who were subjected to
infrequent physical punishment, externalizing behavior
problems were significantly higher among adolescents subjected to frequent physical punishment. In this crosssectional study, the association between frequency of physical punishment and externalizing behaviors followed a step
function gradient. A second study used a causally stronger
fixed effects regression method with longitudinal data to
examine links between three levels of physical punishment
(“never,” “once,” “more than once in the past week”) and
children’s behavior problems (Grogan-Kaylor, 2004). An
increase in the frequency of parents’ physical punishment
between Wave 1 and Wave 2 from “never” to “once in the
past week” predicted the same level of increase in children’s
behavior problems, as did an increase from “once” to “more
than once in the past week.” That the association between
physical punishment and children’s behavior problems followed a linear gradient in both studies is consistent with a
causal link, such that exposure to more physical punishment
leads to more behavior problems.
Ruling Out Spurious Factors
Longitudinal designs used to examine spanking can rule
out the influence of selection factors measured in the study.
However, there could be unmeasured factors that account
for significant associations between physical punishment
and child outcomes. The ideal method for ruling out both
selection factors is the randomized experiment (Duncan et
al., 2004). Very few studies have examined physical pun-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
Yo Jackson
ishment with experimental designs. However, there are also
a few experimental evaluations of interventions and quasiexperiments that bear on the question of whether physical
punishment harms children.
Experiments. The extent of experimental studies is
limited to a series of experiments with small samples conducted by Mark Roberts and his colleagues to examine the
utility of spanking used as a “back up” when children with
behavior problems leave “time out” prematurely. He concluded that spanking was effective compared with other
disciplinary techniques (e.g., Roberts, 1988; Roberts &
Powers, 1990). However, the spank and no-spank groups
had significant differences in their levels of compliance at
the start of the experiments. Once initial differences were
controlled by comparing pre–post differences rather than
just postexperiment differences (see Gershoff & GroganKaylor, 2016a), effect sizes were nonsignificant. There is
thus no experimental evidence that spanking is more effective than other methods at securing children’s immediate
compliance.
There are no other traditional experiments examining
spanking’s direct effect on children’s behaviors. However, dozens of parenting interventions have been found
to reduce parents’ use of physical punishment (Gershoff,
Lee, & Durrant, 2017). A few studies have examined
whether parenting interventions that target a reduction in
physical punishment predict change in child outcomes. A
randomized controlled trial (RCT) evaluation of the Incredible Years intervention for young children with behavior problems found that treatment effects were significantly mediated through a reduction in parents’ use of
spanking (Beauchaine, Webster-Stratton, & Reid, 2005).
631
An analysis of data from a national RCT of the federal
Head Start program showed that parents of children randomly assigned to participate in the Head Start program
decreased their use of spanking more than parents in the
control group, and that this reduction in spanking, in turn,
predicted declines in children’s aggression (Gershoff,
Ansari, Purtell, & Sexton, 2016). An RCT of the Chicago
Parent Program, part of which focused on African American and Latino/a parents and their preschool children,
found that the intervention group reduced parents’ use of
physical punishment significantly more than the control
group and that their children had fewer behavior problems over time (Breitenstein et al., 2012). These experimental program evaluations provide evidence that interventions can reduce child problem behavior by reducing
parents’ use of spanking, and in doing so provide evidence for a causal link between spanking and children’s
problem behavior.
Quasi-experiments. When experiments are difficult to
conduct, statistical methods can be employed to rule out
alternative explanations and to enhance our confidence in
making causal inferences. One such method is fixed effects
regression, which uses difference scores to control for timeinvariant unobserved characteristics that may account for
observed relations between physical punishment and child
outcomes, such as children’s initial levels of problem behavior. Using fixed effects regressions, Grogan-Kaylor
(2005) found that parents’ increased use of spanking predicted increases in children’s subsequent externalizing behaviors.
A rigorous statistical method from econometrics known
as propensity score matching has recently been applied to
the issue of physical punishment with samples in the United
States and Japan. Gershoff, Sattler, and Ansari (2018) used
propensity score matching to create two groups of children
who were matched on 38 demographic characteristics so
that they varied only in whether parents reported they had
ever spanked their children. The spank and no-spank groups
were matched so that they were not significantly different
on characteristics at the levels of the child (e.g., demographic characteristics and initial behavior problems as
rated by teachers), the parent (e.g., demographic and mental
health characteristics), and the family (e.g., income, language, regional location). In the models with the matched
samples, children in the spanked group (n ⫽ 9,479) were
rated by teachers as having significantly more externalizing
behavior problems 1 and 2 years later than children who had
never been spanked (n ⫽ 2,479); there were no differences
for internalizing problems. Similarly, using a sample of
more than 29,000 children, Okuzono, Fujiwara, Kato, and
Kawachi (2017) found that spanked children exhibited more
behavior problems after matching children on 28 variables.
By ruling out many of the plausible third-variable explanations, these results give considerable confidence to the con-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
632
GERSHOFF ET AL.
Alan E. Kazdin
clusion that physical punishment predicts increases in children’s problem behaviors over time.
Summary
The conclusion from this combination of meta-analyses,
experiments, and quasi-experiments is that the preponderance of evidence links physical punishment with detrimental child outcomes. There is no evidence that physical punishment is effective at improving child behavior or at
reducing other negative outcomes for children. The research
linking physical punishment with harm to children is, with
only a few exceptions, consistent and unidirectional, and it
has been replicated across a range of study designs and
methods, thereby increasing the validity of causal inference.
Do Harsh or Abusive Methods Drive the Links
Between Physical Punishment and Detrimental
Child Outcomes?
A persistent criticism of the physical punishment research
is that the associations with detrimental outcomes are driven
by the inclusion of harsh and potentially abusive methods
along with acceptable methods such as spanking (Larzelere
et al., 2017). The underlying argument is that physical
punishment may be effective and not harmful up to a certain
threshold of either severity or frequency, after which it
becomes harmful. If this argument were to be supported,
then socially acceptable manifestations of physical punishment, such as spanking, should be linked with either benefits to children or neutral effects on children, whereas un-
acceptable forms, such as beating children to injury, should
be associated with harm.
One way to address this criticism would be to compare
children who are physically punished but not abused with
children who are abused. Such a comparison would be
problematic for reasons of selection bias— children who are
abused may be different from nonabused children on a
range of individual and family characteristics that may drive
any differences in their outcomes. Children who are physically abused are also likely to be physically punished (Lee,
Grogan-Kaylor, & Berger, 2014), and thus such betweensubjects comparative studies cannot isolate impacts of physical punishment from impacts of physical abuse.
The solution to this problem is to use within-subjects
designs in which both physical punishment and physical
abuse are measured for the same children, thereby eliminating any selection bias. Gershoff and Grogan-Kaylor (2016a)
identified seven studies with 10 different outcomes that
reported effect sizes for links between spanking and child
outcomes as well as for physically abusive methods and the
same outcomes. Both spanking and physical abuse were
linked with more problematic outcomes for children. The
mean effect size for the link between spanking and detrimental outcomes was two thirds the magnitude of the effect
size for the link between physical abuse and the same
detrimental outcomes (Gershoff & Grogan-Kaylor, 2016a).
A similar conclusion was reached in an analysis of data
from the Adverse Childhood Experiences Survey (Afifi et
al., 2017). Adults who reported having ever been spanked in
childhood were more likely to have attempted suicide, to
have used street drugs, and to have had a drinking problem
than adults who had not been spanked, over and above
whether the adults reported having been physically or emotionally abused as children.
These findings suggest that the distinction between
acceptable physical punishment and unacceptable physical abuse is largely semantic; they are linked with the
same detrimental outcomes for children, just to varying
degrees. Euphemisms such as “spanking” have obscured
the fact that both physical punishment and physical abuse
involve hitting and hurting children emotionally and
physically (Brown, Holden, & Ashraf, 2018; Gershoff,
2013).
Are the Detrimental Outcomes Linked With
Physical Punishment Generalizable
Across Contexts?
The third and final major concern about the research on
physical punishment is that it is not generalizable across
contexts (Deater-Deckard & Dodge, 1997). Some researchers have argued that physical punishment may benefit children growing up in some contexts more than children growing up in other contexts. The three main contexts that have
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
been investigated as with regard to physical punishment are
culture, parenting style, and neighborhoods.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Culture
Within the United States, examinations of “culture” and
physical punishment have tended to focus on a single cultural marker: race or ethnicity. Race refers to categories of
individuals based on physical characteristics, whereas ethnicity implies the acceptance of a set of beliefs and values
from a culture of origin (American Psychological Association, 2002). We acknowledge that there is wide variability
both in cultures and in parenting practices within racial
groups, but because race has been theoretically and empirically considered a cultural moderator of discipline effects
on children, we will consider race and ethnicity to be
markers of culture as well.
The discipline used in African American families has
often been contrasted with that found in European American
families. African Americans tend to more strongly endorse
physical punishment as an appropriate and effective discipline strategy than do European Americans (Mosby, Rawls,
Meehan, Mays, & Pettinari, 1999) and to spank their children more frequently than European American or Hispanic/
Latino parents (e.g., Barkin, Scheindlin, Ip, Richardson, &
Finch, 2007; Deater-Deckard, Dodge, Bates, & Pettit, 1996;
Gershoff et al., 2012; Grogan-Kaylor & Otis, 2007). Such
findings led to the development of the cultural normativeness hypothesis (Deater-Deckard & Dodge, 1997), which
states that physical punishment may have less negative
impact, and a potentially positive impact, when it is culturally normative because children expect and accept it when
they misbehave. A few studies have indeed found that
African American children who are physically punished
have levels of behavior problems that are lower (Gunnoe &
Mariner, 1997) or the same as (Polaha, Shapiro, Larzelere,
& Pettit, 2004) European American children who are physically punished.
In contrast, other studies using data from longitudinal,
nationally representative samples have not found race or
ethnicity to moderate the association of physical punishment with later child behavior (Grogan-Kaylor, 2004, 2005;
Lau, Litrownik, Newton, Black, & Everson, 2006; McLoyd
& Smith, 2002; Mulvaney & Mebert, 2007). Three methodologically strong studies that utilized cross-lagged longitudinal models controlling for children’s initial behavior
along with empirical tests of moderation each found no
moderation by race or ethnicity (Berlin et al., 2009; Gershoff et al., 2012; Wang & Kenny, 2014).
In addition, a meta-analysis of five studies that reported
bivariate associations between spanking and child outcomes
separately for White and Black children determined that the
effect sizes were nearly identical for both groups (d ⫽ .274
for Whites [n ⫽ 11,814]; d ⫽ .279 for Blacks [n ⫽ 3,065])
633
and both effect sizes were statistically different from zero
(Gershoff & Grogan-Kaylor, 2016b). Taken together, these
studies support a conclusion that there are few, if any, racial
or ethnic group differences in the extent to which physical
punishment is linked with worse child behavior.
Parenting Style
Physical punishment is but one of many disciplinary
techniques parents use and it occurs in the context of an
overall parenting style. Parenting style is thought to be more
influential than individual parenting behaviors and to serve
as a moderator of them (Darling & Steinberg, 1993). Some
researchers have thus argued that a warm or positive parenting style may buffer against any negative effects of
physical punishment (Benjet & Kazdin, 2003).
This possibility has been examined in a few studies and
there is some evidence that parental warmth buffers associations between physical punishment and child outcomes.
Several studies have found that the link between physical
punishment and negative child outcomes is stronger when
parents are low in warmth or high in negativity (e.g., Aucoin, Frick, & Bodin, 2006; Deater-Deckard, Ivy, & Petrill,
2006; McKee et al., 2007). Yet other cross-sectional studies
have not found maternal warmth to buffer the crosssectional association of spanking with children’s behavior
problems in the United States (Stacks, Oshio, Gerard, &
Roe, 2009) or in 20 different countries (Fréchette, Zoratti, &
Romano, 2015).
Four studies have examined this question using longitudinal data. One study of over 11,000 children found
that parental warmth did not moderate the link between
spanking at Age 5 and lower math scores at Age 10
(Bodovski & Youn, 2010). An investigation using a
cross-lagged panel design with 2,500 children at Ages 1,
2, and 3 found maternal warmth did not buffer the associations of spanking with increased behavior problems
and lower cognitive ability (Berlin et al., 2009). In another study using a cross-lagged design with 3,200 children, greater use of spanking at Age 3 predicted an
increase in aggression from Age 3 to Age 5 regardless of
their mothers’ level of warmth (Lee, Altschul, & Gershoff, 2013). A fourth study also used a cross-lagged
design and failed to detect any evidence of moderation of
parental warmth on relations between maternal or paternal physical punishment and children’s behavior problems and depressive symptoms (Wang & Kenny, 2014).
The fact that warmth has not been consistently found to
buffer the links between physical punishment and child
behavior in cross-sectional studies, and not at all in
longitudinal studies, indicates that parenting warmth is
not a reliable buffer against the negative outcomes associated with physical punishment.
GERSHOFF ET AL.
634
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Neighborhood Quality
There has been speculation that neighborhood contexts
moderate the link between physical punishment and children’s behavior problems. For families living in neighborhoods with high levels of violence, crime, and illicit activity, the consequences of child misbehavior may be serious
(e.g., injury or death). Thus one possibility is that parents’
use of physical punishment may be more successful at
promoting positive child behavior when families reside in
dangerous neighborhoods (Garbarino, Kostelny, & Barry,
1997).
There is little evidence to support this argument. First, it
is not clear that the rate of physical punishment is indeed
higher in violent or low-income neighborhoods. Neighborhood problems have sometimes been associated with higher
rates of physical punishment (Molnar, Buka, Brennan, Holton, & Earls, 2003) but sometimes not (Grogan-Kaylor &
Otis, 2007). Second, neighborhood context has not been
found to be a predictable moderator of associations between
physical punishment and child behavior. Although one
study found that physical punishment was linked with lower
child behavior problems when families lived in communities with higher levels of neighborhood problems (Eamon,
2002), this finding was refuted when reanalyzed with more
sophisticated fixed effects regression methods that found
physical punishment was equally deleterious for children in
all neighborhoods (Grogan-Kaylor, 2005). Similarly, a large
two-state study using a multilevel design found that community deviance did not moderate the association between
physical punishment and children’s conduct problems (Simons et al., 2002).
Implications of the Research on
Physical Punishment
The extensive body of empirical research on physical
punishment, including some experimental and quasiexperimental studies, is quite consistent and affords a clear
conclusion: Physical punishment increases the risk that children will experience detrimental outcomes and this risk is
experienced equally across cultural groups, families, and
neighborhoods. The consistency and strength of these findings suggest several implications for parents, psychologists,
and policymakers.
Implications for Parents
Parents and the general public should be educated about
the body of research indicating physical punishment is
ineffective and may be harmful to children. Professional
associations such as the American Academy of Pediatrics
(1998, 2014), the National Association of Pediatric Nurse
Practitioners (2011), the American Academy of Child and
Adolescent Psychiatry (2012), and the American Profes-
sional Society on the Abuse of Children (2016) have all
issued policy statements both urging parents to avoid using
physical punishment and directing professionals who work
with parents to advise them to discipline their children with
nonphysical techniques.
Education about alternatives to physical punishment can
occur in a range of contexts and at universal, selective, or
indicated levels of intervention (Gershoff et al., 2017).
Interventions in pediatric health care settings have successfully reduced attitudes about or intent to use physical punishment (e.g., Canfield et al., 2015). Home visiting programs that focus on teaching nonviolent parental discipline
have been successful at reducing physical punishment and
preventing physical abuse (Bugental et al., 2002). Groupbased interventions can also be effective at teaching alternatives to physical punishment; the Adults and Children
Together against Violence educational program (http://
actagainstviolence.apa.org/) created by APA’s Violence
Prevention Office teaches social problem-solving skills and
nonviolent discipline and has been shown to reduce spanking and to increase positive discipline in the United States
and around the world (e.g., Howe et al., 2017; Portwood,
Lambert, Abrams, & Nelson, 2011).
Interventions to educate parents about physical punishment need not be intensive to be successful. New mothers in
an RCT who received baby books that informed them about
alternatives to physical punishment reported significantly
less favorable views toward physical punishment than did
mothers who received noneducational baby books or no
books (Reich, Penner, Duncan, & Auger, 2012). A randomized controlled study effectively used motivational interviewing coupled with psychoeducation to change the attitudes and behavior intentions of mothers of young children
(Holland & Holden, 2016). Merely presenting parents with
summaries of the research on physical punishment has also
been found to reduce approval of physical punishment
(Holden, Brown, Baldwin, & Croft Caderao, 2014).
Implications for Psychologists
The work of psychologists is guided by APA’s (2010)
ethics code as well as by resolutions passed by APA related
to research and practice. Principle A in the ethics code
concerns safeguarding the welfare and rights of others.
Because physical punishment poses a harm to children’s
welfare, psychologists are obligated to caution parents
against using physical punishment. APA was a leader in
opposing physical punishment in schools when it became
the first professional organization in the United States to
pass a resolution opposing school paddling (APA, 1975).
That resolution was largely based on research on parents’
use of physical punishment because there was no research
then on the effects of school physical punishment on children. Out of over 800 APA members in a recent survey,
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
59% were in favor of an APA resolution opposing any use
of spanking or physical punishment by parents or caregivers
(Miller-Perrin & Rush, 2018). APA should take a stand
against physical punishment and work to educate its members and the public about the harms of physical punishment
and about the benefits of alternative positive forms of discipline.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Implications for Policymakers
Children are the only group of American citizens who can
be legally subjected to physical punishment. Physical punishment of adults was banned from all prisons and military
training facilities in the United States decades ago (Block,
1997). Physical punishment has been banned from public
schools in 31 states (Gershoff & Font, 2016) and from child
care centers, residential treatment facilities, and juvenile
detention facilities in most states (Bitensky, 2006). At the
federal level, the notion that parents have a right to discipline their children as they see fit reflects a historical
perspective that children are the property of their parents
(Pollard, 2003). Whether parents have a fundamental constitutional right to use physical punishment, or children
have the right to be protected from it, has not been brought
before the United States Supreme Court (Pollard, 2003).
The committee that monitors compliance with the United
Nations Convention on the Rights of the Child (1989) has
declared that physical punishment is a form of “legalized
violence against children” (United Nations Committee on
the Rights of the Child, 2007, para. 18). The Committee
stressed that all physical punishment of children, including
that in homes, should be eliminated through “legislative,
administrative, social and educational measures” (United
Nations Committee on the Rights of the Child, 2007, para.
18). As of 2018, 53 countries, including most of Europe and
South America, have banned all physical punishment of
children (Global Initiative to End All Corporal Punishment
of Children, 2018).
The United States is the only country in the world that has
not ratified the Convention on the Rights of the Child. The
United States’ failure to do so rests in part on some critics’
objection to what they consider to be governmental interference in parenting (Wilkins, Becker, Harris, & Thayer,
2003). A legal ban would be a clear example of government
regulating family life, and thus the prospects of a ban in the
United States are unlikely at present. In a survey of adults in
the State of New York 10 years ago, a majority (64%) were
opposed to legislation that would ban parents from using
physical punishment (SurveyUSA, 2007). Yet there are
many policy strategies short of a legal ban that could be
used to reduce parents’ use of physical punishment in the
United States. One prime example would be educational
efforts that target societal attitudes about physical punishment as a means of reducing its use. In their review of
635
primary prevention strategies to prevent child abuse, Klevens and Whitaker (2007) importantly recognized that “social norms regarding physical discipline may be the most
prevalent risk factor for child abuse in the United States” (p.
371). Changing norms about physical punishment can lead
to reductions in risk at the family level that will accumulate
to improve overall public health (D. A. Cohen, Scribner, &
Farley, 2000).
Conclusion
This article summarized three key conclusions about
spanking and physical punishment that bear on psychologists’ views about it: (a) Research on physical punishment
has met the requirements for causal conclusions, despite
there being few true experiments; (b) Research on spanking
and physical punishment is highly consistent in showing
links to detrimental child outcomes for children; (c) Spanking and physical punishment are linked with the same harms
to children as physical abuse; (d) Spanking and physical
punishment have been linked with harm to children across
multiple contexts; (e) Spanking and physical punishment
are disavowed by a number of professional organizations
outside psychology; and (e) Human rights organizations and
53 countries agree that spanking and physical punishment
are forms of violence that infringe on children’s human
rights. The message to parents, psychologists, and policymakers is clear—it is time to end the debate about physical
punishment and to end this outdated parenting practice.
References
Afifi, T. O., Ford, D., Gershoff, E. T., Merrick, M., Ports, K., GroganKaylor, A., . . . Bennett, R. P. (2017). Spanking and adult mental health
impairment: The case for the designation of spanking as an adverse
childhood experience. Child Abuse & Neglect, 71, 24 –31. http://dx.doi
.org/10.1016/j.chiabu.2017.01.014
American Academy of Child and Adolescent Psychiatry. (2012, July 30).
Policy statement on corporal punishment. Retrieved from http://www
.aacap.org/aacap/policy_statements/2012/Policy_Statement_on_Corporal_
Punishment.aspx
American Academy of Pediatrics. (1998). Guidance for effective discipline. Pediatrics, 101, 723–728.
American Academy of Pediatrics. (2014). AAP publications reaffirmed or
retired. Pediatrics, 134, e1520. http://dx.doi.org/10.1542/peds.20142679
American Professional Society on the Abuse of Children. (2016). APSAC
position statement on corporal punishment of children. Retrieved from
https://docs.wixstatic.com/ugd/4700a8_7fefe2fdfba249d1949d536
ca3789d78.pdf
American Psychological Association. (1975). Corporal punishment. APA
Council Policy Manual. Retrieved from http://www.apa.org/about/policy/
corporal-punishment.aspx
American Psychological Association (APA). (1999). Resolution on male
violence against women. APA Council Policy Manual. Retrieved from
http://www.apa.org/about/policy/chap.-12b.aspx#male-violence
American Psychological Association (APA). (2002). Guidelines on multicultural education, training, research, practice, and organizational
change for psychologists. Retrieved from http://www.apa.org/pi/oema/
resources/policy/multicultural-guidelines.aspx
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
636
GERSHOFF ET AL.
American Psychological Association (APA). (2004). APA resolution on
bullying among children and youth. APA Council Policy Manual. Retrieved from http://www.apa.org/about/policy/chap.-12b.aspx#maleviolence
American Psychological Association (APA). (2010). Ethical principles of
psychologists and code of conduct. Washington, DC: Author. Retrieved
from http://www.apa.org/ethics/code/principles.pdf
Aucoin, K. J., Frick, P. J., & Bodin, S. D. (2006). Corporal punishment and
child adjustment. Journal of Applied Developmental Psychology, 27,
527–541. http://dx.doi.org/10.1016/j.appdev.2006.08.001
Barkin, S., Scheindlin, B., Ip, E. H., Richardson, I., & Finch, S. (2007).
Determinants of parental discipline practices: A national sample from
primary care practices. Clinical Pediatrics, 46, 64 – 69. http://dx.doi.org/
10.1177/0009922806292644
Beauchaine, T. P., Webster-Stratton, C., & Reid, M. J. (2005). Mediators,
moderators, and predictors of 1-year outcomes among children treated
for early-onset conduct problems: A latent growth curve analysis. Journal of Consulting and Clinical Psychology, 73, 371–388. http://dx.doi
.org/10.1037/0022-006X.73.3.371
Benjet, C., & Kazdin, A. E. (2003). Spanking children: The controversies,
findings, and new directions. Clinical Psychology Review, 23, 197–224.
http://dx.doi.org/10.1016/S0272-7358(02)00206-4
Berlin, L. J., Ispa, J. M., Fine, M. A., Malone, P. S., Brooks-Gunn, J.,
Brady-Smith, C., . . . Bai, Y. (2009). Correlates and consequences of
spanking and verbal punishment for low-income white, African American, and Mexican American toddlers. Child Development, 80, 1403–
1420. http://dx.doi.org/10.1111/j.1467-8624.2009.01341.x
Bitensky, S. H. (2006). Corporal punishment of children: A human rights
violation. Ardsley, NY: Transnational. http://dx.doi.org/10.1163/ej
.9781571053657.i-398
Block, N. (1997). Would you hit Grandma? Children’s Legal Rights
Journal, 17, 3– 4.
Bodovski, K., & Youn, M. (2010). Love, discipline and elementary school
achievement: The role of family emotional climate. Social Science
Research, 39, 585–595. http://dx.doi.org/10.1016/j.ssresearch.2010.03
.008
Breitenstein, S. M., Gross, D., Fogg, L., Ridge, A., Garvey, C., Julion, W.,
& Tucker, S. (2012). The Chicago Parent Program: Comparing 1-year
outcomes for African American and Latino parents of young children.
Research in Nursing & Health, 35, 475– 489. http://dx.doi.org/10.1002/
nur.21489
Brown, A. S., Holden, G. W., & Ashraf, R. (2018). Spank, slap, or hit?
How labels alter perceptions of child discipline. Psychology of Violence,
8, 1–9. http://dx.doi.org/10.1037/vio0000080
Bugental, D. B., Ellerson, P. C., Lin, E. K., Rainey, B., Kokotovic, A., &
O’Hara, N. (2002). A cognitive approach to child abuse prevention.
Journal of Family Psychology, 16, 243–258. http://dx.doi.org/10.1037/
0893-3200.16.3.243
Canfield, C. F., Weisleder, A., Cates, C. B., Huberman, H. S., Dreyer,
B. P., Legano, L. A., . . . Mendelsohn, A. L. (2015). Primary care
parenting intervention and its effects on the use of physical punishment
among low-income parents of toddlers. Journal of Developmental and
Behavioral Pediatrics, 36, 586 –593. http://dx.doi.org/10.1097/DBP
.0000000000000206
Cohen, D. A., Scribner, R. A., & Farley, T. A. (2000). A structural model
of health behavior: A pragmatic approach to explain and influence health
behaviors at the population level. Preventive Medicine: An International
Journal Devoted to Practice and Theory, 30, 146 –154. http://dx.doi.org/
10.1006/pmed.1999.0609
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
Darling, N., & Steinberg, L. (1993). Parenting style as a context: An
integrative model. Psychological Bulletin, 113, 487– 496. http://dx.doi
.org/10.1037/0033-2909.113.3.487
Deater-Deckard, K., & Dodge, K. A. (1997). Externalizing behavior problems and discipline revisited: Nonlinear effects and variation by culture,
context, and gender. Psychological Inquiry, 8, 161–175. http://dx.doi
.org/10.1207/s15327965pli0803_1
Deater-Deckard, K., Dodge, K. A., Bates, J. E., & Pettit, G. S. (1996).
Physical discipline among African American and European American
mothers: Links to children’s externalizing behaviors. Developmental
Psychology, 32, 1065–1072. http://dx.doi.org/10.1037/0012-1649.32.6
.1065
Deater-Deckard, K., Ivy, L., & Petrill, S. A. (2006). Maternal warmth
moderates the link between physical punishment and child externalizing
problems: A parent-offspring behavioral genetic analysis. Parenting:
Science and Practice, 6, 59 –78. http://dx.doi.org/10.1207/
s15327922par0601_3
Duncan, G. J., Magnuson, K. A., & Ludwig, J. (2004). The endogeneity
problem in developmental studies. Research in Human Development, 1,
59 – 80. http://dx.doi.org/10.1207/s15427617rhd0101&2_5
Eamon, M. K. (2002). Poverty, parenting, peer, and neighborhood influences on young adolescent antisocial behavior. Journal of Social Service
Research, 28, 1–23. http://dx.doi.org/10.1300/J079v28n01_01
Fedak, K. M., Bernal, A., Capshaw, Z. A., & Gross, S. (2015). Applying
the Bradford Hill criteria in the 21st century: How data integration has
changed causal inference in molecular epidemiology. Emerging Themes
in Epidemiology, 12, 14. http://dx.doi.org/10.1186/s12982-015-0037-4
Ferguson, C. J. (2013). Spanking, corporal punishment and negative longterm outcomes: A meta-analytic review of longitudinal studies. Clinical
Psychology Review, 33, 196 –208. http://dx.doi.org/10.1016/j.cpr.2012
.11.002
Fréchette, S., Zoratti, M., & Romano, E. (2015). What is the link between
corporal punishment and child physical abuse? Journal of Family Violence, 30, 135–148. http://dx.doi.org/10.1007/s10896-014-9663-9
Garbarino, J., Kostelny, K., & Barry, F. (1997). Value transmission in an
ecological context: The high-risk neighborhood. In J. E. Grusec & L.
Kuczynski (Eds.), Parenting and children’s internalization of values: A
handbook of contemporary theory (pp. 307–332). New York, NY: Wiley.
Gershoff, E. T. (2002). Corporal punishment by parents and associated
child behaviors and experiences: A meta-analytic and theoretical review.
Psychological Bulletin, 128, 539 –579. http://dx.doi.org/10.1037/00332909.128.4.539
Gershoff, E. T. (2013). Spanking and child development: We know enough
now to stop hitting our children. Child Development Perspectives, 7,
133–137. http://dx.doi.org/10.1111/cdep.12038
Gershoff, E. T., Ansari, A., Purtell, K. M., & Sexton, H. R. (2016).
Changes in parents’ spanking and reading as mechanisms for Head Start
impacts on children. Journal of Family Psychology, 30, 480 – 491. http://
dx.doi.org/10.1037/fam0000172
Gershoff, E. T., & Font, S. A. (2016). Corporal punishment in U.S. public
schools: Prevalence, disparities in use, and status in state and federal
policy. Social Policy Report, 30, 1–25.
Gershoff, E. T., & Grogan-Kaylor, A. (2016a). Spanking and child outcomes: Old controversies and new meta-analyses. Journal of Family
Psychology, 30, 453– 469. http://dx.doi.org/10.1037/fam0000191
Gershoff, E. T., & Grogan-Kaylor, A. (2016b). Race as a moderator of
associations between spanking and child outcomes. Family Relations:
An Interdisciplinary Journal of Applied Family Studies, 65, 490 –501.
http://dx.doi.org/10.1111/fare.12205
Gershoff, E. T., Lansford, J. E., Sexton, H. R., Davis-Kean, P., & Sameroff, A. J. (2012). Longitudinal links between spanking and children’s
externalizing behaviors in a national sample of White, Black, Hispanic,
and Asian American families. Child Development, 83, 838 – 843. http://
dx.doi.org/10.1111/j.1467-8624.2011.01732.x
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
THE EVIDENCE AGAINST PHYSICAL PUNISHMENT
Gershoff, E. T., Lee, S. J., & Durrant, J. E. (2017). Promising intervention
strategies to reduce parents’ use of physical punishment. Child Abuse &
Neglect, 71, 9 –23. http://dx.doi.org/10.1016/j.chiabu.2017.01.017
Gershoff, E. T., Sattler, K. M. P., & Ansari, A. (2018). Strengthening
causal estimates for links between spanking and children’s externalizing
behavior problems. Psychological Science, 29, 110 –120. http://dx.doi
.org/10.1177/0956797617729816
Global Initiative to End All Corporal Punishment of Children. (2017).
States which have prohibited all corporal punishment. Retrieved from
http://www.endcorporalpunishment.org/progress/prohibiting-states/
Grogan-Kaylor, A. (2004). The effect of corporal punishment on antisocial
behavior in children. Social Work Research, 28, 153–162. http://dx.doi
.org/10.1093/swr/28.3.153
Grogan-Kaylor, A. (2005). Corporal punishment and the growth trajectory
of children’s antisocial behavior. Child Maltreatment, 10, 283–292.
http://dx.doi.org/10.1177/1077559505277803
Grogan-Kaylor, A., & Otis, M. D. (2007). The predictors of parental use of
corporal punishment. Family Relations: An Interdisciplinary Journal of
Applied Family Studies, 56, 80 –91. http://dx.doi.org/10.1111/j.17413729.2007.00441.x
Gunnoe, M. L., & Mariner, C. L. (1997). Toward a developmentalcontextual model of the effects of parental spanking on children’s
aggression. Archives of Pediatrics & Adolescent Medicine, 151, 768 –
775. http://dx.doi.org/10.1001/archpedi.1997.02170450018003
Hill, A. B. (1965). The environment and disease: Association or causation?
Proceedings of the Royal Society of Medicine, 58, 295–300.
Holden, G. W., Brown, A. S., Baldwin, A. S., & Croft Caderao, K. (2014).
Research findings can change attitudes about corporal punishment. Child
Abuse & Neglect, 38, 902–908. http://dx.doi.org/10.1016/j.chiabu.2013
.10.013
Holden, G. W., Miller, P. C., & Harris, S. D. (1999). The instrumental side
of corporal punishment: Parents’ reported practices and outcome expectancies. Journal of Marriage and Family, 61, 908 –919. http://dx.doi.org/
10.2307/354012
Holland, G. W. O., & Holden, G. W. (2016). Changing orientations to
corporal punishment: A randomized control trial of the efficacy of a
motivational approach to psycho-education. Psychology of Violence, 6,
233–242. http://dx.doi.org/10.1037/a0039606
Howe, T. R., Knox, M., Altafim, E. R. P., Linhares, M. B. M., Nishizawa,
N., Fu, T. J., . . . Pereira, A. I. (2017). International child abuse
prevention: Insights from ACT Raising Safe Kids. Child and Adolescent
Mental Health, 22, 194 –200. http://dx.doi.org/10.1111/camh.12238
Klevens, J., & Whitaker, D. J. (2007). Primary prevention of child physical
abuse and neglect: Gaps and promising directions. Child Maltreatment,
12, 364 –377. http://dx.doi.org/10.1177/1077559507305995
Lansford, J. E., Wager, L. B., Bates, J. E., Pettit, G. S., & Dodge, K. A.
(2012). Forms of spanking and children’s externalizing behaviors. Family Relations: An Interdisciplinary Journal of Applied Family Studies,
61, 224 –236. http://dx.doi.org/10.1111/j.1741-3729.2011.00700.x
Larzelere, R. E., Cox, R. B., & Smith, G. L. (2010). Do nonphysical
punishments reduce antisocial behavior more than spanking? A comparison using the strongest previous causal evidence against spanking. BMC
Pediatrics, 10, 10. http://dx.doi.org/10.1186/1471-2431-10-10
Larzelere, R. E., Gunnoe, M. L., Roberts, M. W., & Ferguson, C. J. (2017).
Children and parents deserve better parental discipline research: Critiquing the evidence for exclusively “positive” parenting. Marriage & Family Review, 53, 24 –35. http://dx.doi.org/10.1080/01494929.2016
.1145613
Larzelere, R. E., & Kuhn, B. R. (2005). Comparing child outcomes of
physical punishment and alternative disciplinary tactics: A metaanalysis. Clinical Child and Family Psychology Review, 8, 1–37. http://
dx.doi.org/10.1007/s10567-005-2340-z
Larzelere, R. E., Kuhn, B. R., & Johnson, B. (2004). The intervention
selection bias: An underrecognized confound in intervention research.
637
Psychological Bulletin, 130, 289 –303. http://dx.doi.org/10.1037/00332909.130.2.289
Lau, A. S., Litrownik, A. J., Newton, R. R., Black, M. M., & Everson,
M. D. (2006). Factors affecting the link between physical discipline and
child externalizing problems in Black and White families. Journal of
Community Psychology, 34, 89 –103. http://dx.doi.org/10.1002/jcop
.20085
Lee, S. J., Altschul, I., & Gershoff, E. T. (2013). Does warmth moderate
longitudinal associations between maternal spanking and child aggression in early childhood? Developmental Psychology, 49, 2017–2028.
http://dx.doi.org/10.1037/a0031630
Lee, S. J., Grogan-Kaylor, A., & Berger, L. M. (2014). Parental spanking
of 1-year-old children and subsequent child protective services involvement. Child Abuse & Neglect, 38, 875– 883. http://dx.doi.org/10.1016/j
.chiabu.2014.01.018
Ma, J., Han, Y., Grogan-Kaylor, A., Delva, J., & Castillo, M. (2012).
Corporal punishment and youth externalizing behavior in Santiago,
Chile. Child Abuse & Neglect, 36, 481– 490. http://dx.doi.org/10.1016/
j.chiabu.2012.03.006
Maguire-Jack, K., Gromoske, A. N., & Berger, L. M. (2012). Spanking and
child development during the first 5 years of life. Child Development,
83, 1960 –1977. http://dx.doi.org/10.1111/j.1467-8624.2012.01820.x
McKee, L., Roland, E., Coffelt, N., Olson, A. L., Forehand, R., Massari, C.,
. . . Zens, M. S. (2007). Harsh discipline and child problem behaviors:
The roles of positive parenting and gender. Journal of Family Violence,
22, 187–196. http://dx.doi.org/10.1007/s10896-007-9070-6
McLoyd, V. C., & Smith, J. (2002). Physical discipline and behavior
problems in African American, European American, and Hispanic children: Emotional support as a moderator. Journal of Marriage and
Family, 64, 40 –53. http://dx.doi.org/10.1111/j.1741-3737.2002.00040.x
Miller-Perrin, C., & Rush, R. (2018). Attitudes, knowledge, practices, and
ethical beliefs of psychologists related to spanking: A survey of American Psychological Association division members. Manuscript submitted
for publication.
Molnar, B. E., Buka, S. L., Brennan, R. T., Holton, J. K., & Earls, F.
(2003). A multilevel study of neighborhoods and parent-to-child physical aggression: Results from the project on human development in
Chicago neighborhoods. Child Maltreatment, 8, 84 –97. http://dx.doi
.org/10.1177/1077559502250822
Mosby, L., Rawls, A. W., Meehan, A. J., Mays, E., & Pettinari, C. J.
(1999). Troubles in interracial talk about discipline: An examination of
African American child rearing narratives. Journal of Comparative
Family Studies, 30, 489 –521.
Moyer, M. W. (2016, May 3). What science says—And doesn’t—About
spanking. Scientific American. Retrieved from https://www.scientific
american.com/article/what-science-says-and-doesn-t-about-spanking/
Mulvaney, M. K., & Mebert, C. J. (2007). Parental corporal punishment
predicts behavior problems in early childhood. Journal of Family Psychology, 21, 389 –397. http://dx.doi.org/10.1037/0893-3200.21.3.389
National Association of Pediatric Nurse Practitioners. (2011). NAPNAP
position statement on corporal punishment. Journal of Pediatric Health
Care, 25, e31– e32. http://dx.doi.org/10.1016/j.pedhc.2011.07.003
Okuzono, S., Fujiwara, T., Kato, T., & Kawachi, I. (2017). Spanking and
subsequent behavioral problems in toddlers: A propensity scorematched, prospective study in Japan. Child Abuse & Neglect, 69, 62–71.
http://dx.doi.org/10.1016/j.chiabu.2017.04.002
Olson, S. L., Lopez-Duran, N., Lunkenheimer, E. S., Chang, H., & Sameroff, A. J. (2011). Individual differences in the development of early peer
aggression: Integrating contributions of self-regulation, theory of mind,
and parenting. Development and Psychopathology, 23, 253–266. http://
dx.doi.org/10.1017/S0954579410000775
Paolucci, E. O., & Violato, C. (2004). A meta-analysis of the published
research on the affective, cognitive, and behavioral effects of corporal
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
638
GERSHOFF ET AL.
punishment. The Journal of Psychology: Interdisciplinary and Applied,
138, 197–222. http://dx.doi.org/10.3200/JRLP.138.3.197-222
Polaha, J., Shapiro, S. K., Larzelere, R., & Pettit, G. S. (2004). Physical
discipline and child behavior problems: A study of ethnic group differences. Parenting: Science and Practice, 4, 339 –360. http://dx.doi.org/
10.1207/s15327922par0404_6
Pollard, D. A. (2003). Banning child corporal punishment. Tulane Law
Review, 77, 575– 657.
Portwood, S. G., Lambert, R. G., Abrams, L. P., & Nelson, E. B. (2011).
An evaluation of the adults and children together (ACT) against violence
parents raising safe kids program. The Journal of Primary Prevention,
32, 147–160. http://dx.doi.org/10.1007/s10935-011-0249-5
Reich, S. M., Penner, E. K., Duncan, G. J., & Auger, A. (2012). Using baby
books to change new mothers’ attitudes about corporal punishment.
Child Abuse & Neglect, 36, 108 –117. http://dx.doi.org/10.1016/j.chiabu
.2011.09.017
Roberts, M. W. (1988). Enforcing chair timeouts with room timeouts.
Behavior Modification, 12, 353–370. http://dx.doi.org/10.1177/
01454455880123003
Roberts, M. W., & Powers, S. W. (1990). Adjusting chair timeout enforcement procedures for oppositional children. Behavior Therapy, 21, 257–
271. http://dx.doi.org/10.1016/S0005-7894(05)80329-6
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2001). Experimental and
quasi-experimental designs for generalized causal inference. New York,
NY: Houghton Mifflin.
Simons, R. L., Lin, K., Gordon, L. C., Brody, G. H., Murry, V., & Conger,
R. D. (2002). Community differences in the association between parenting practices and child conduct problems. Journal of Marriage and
Family, 64, 331–345. http://dx.doi.org/10.1111/j.1741-3737.2002
.00331.x
Stacks, A. M., Oshio, T., Gerard, J., & Roe, J. (2009). The moderating
effect of parental warmth on the association between spanking and child
aggression: A longitudinal approach. Infant and Child Development, 18,
178 –194. http://dx.doi.org/10.1002/icd.596
SurveyUSA. (2007, November 28). News poll (#11013). Retrieved from
http://www.surveyusa.com/client/PollReportEmail.aspx?g⫽edef1e4354b6-4ec2-b2bc-a916b0fc6740
United Nations Committee on the Rights of the Child (CRC). (2007, March
2). CRC General Comment No. 8 (2006): The Right of the Child to
Protection from Corporal Punishment and Other Cruel or Degrading
forms of Punishment (U.N. CRC/C/GC/8). Retrieved from http://www
.refworld.org/docid/460bc7772.html
United Nations Convention on the Rights of the Child. G. A. Res. 44/25,
U. N. GAOR, 44th Sess., at 3, U. N. Doc. A/RES/44/25. (1989, November 20). Retrieved from http://www.unicef.org/crc/
Wang, M. T., & Kenny, S. (2014). Parental physical punishment and
adolescent adjustment: Bidirectionality and the moderation effects of
child ethnicity and parental warmth. Journal of Abnormal Child Psychology, 42, 717–730. http://dx.doi.org/10.1007/s10802-013-9827-8
Wilkins, R. G., Becker, A., Harris, J., & Thayer, D. (2003). United States
and its participation in the Convention on the Rights of the Child: Why
the United States should not ratify the Convention on the Rights of the
Child. Saint Louis University Public Law Review, 22, 411– 434.
Received November 29, 2017
Revision received March 1, 2018
Accepted March 8, 2018 䡲
645072
research-article2016
PSSXXX10.1177/0956797616645072Psychological Science
Corrigendum
Corrigendum: Low Childhood
Socioeconomic Status Promotes Eating
in the Absence of Energy Need
Psychological Science
2016, Vol. 27(6) 932
© The Author(s) 2016
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797616645072
pss.sagepub.com
Hill, S. E., Prokosch, M. L., DelPriore, D. J., Griskevicius, V., & Kramer, A. (2016). Low childhood socioeconomic status
promotes eating in the absence of energy need. Psychological Science, 27, 354–364. (Original DOI: 10.1177/
0956797615621901)
Because of an error in mean-centering of the energyneed predictor used in Study 1 of this article, the inferential statistics reported for that study were incorrect. The
corrected inferential statistics are reported here, and the
Supplemental Material has also been updated to reflect
the corrected results. Note that the pattern of results and
their interpretation do not change with this correction.
The Results section for Study 1, beginning with the
second paragraph, should read as follows:
Total calories consumed. In our target analysis, we
examined the impact of childhood SES and energy need
on the total number of calories participants consumed during their laboratory session. Results revealed a significant
interaction between the two predictors, b = 36.94, SE =
13.21, t(24) = 2.80, p = .01, semipartial r 2 = .13. We probed
this interaction by conducting regions-of-significance tests
in which we examined the impact of each predictor on
food intake at 1 standard deviation above (high) and
below (low) the mean of the other predictor. First, we
probed this interaction by examining the impact of childhood SES on calorie consumption at different levels of
energy need. When energy need was high, there were no
differences between participants from high- and low-SES
environments in the number of calories consumed,
b = 17.66, p = .26. However, when energy need was low,
there was a negative relationship between childhood SES
and food intake, with individuals from low-SES childhood
environments consuming a significantly greater number
of calories than those from high-SES environments,
b = −46.51, SE = 16.36, t(24) = −2.84, p = .009, semipartial
r 2 = .13 (see Fig. 1).
We next examined the impact of energy need on calories consumed at different levels of childhood SES. For
individuals reared in high-SES environments, calorie consumption was greater when self-reported energy need
was high than when it was low, b = 125.11, SE = 30.64,
t(24) = 4.08, p < .001, semipartial r 2 = .28 (see Fig. 1). We
did not observe an effect of energy need on the number
of calories consumed by participants reared in low-SES
environments, however, b = 11.36, p = .67.
Cookie and pretzel intake. We next examined separately the impact of childhood SES and energy need on
the number of grams of each snack type (calorically
dense cookies vs. relatively low-calorie pretzels) consumed by participants. All covariates were the same as in
the target analysis, except that instead of controlling for
liking of both food items, we controlled only for the
hedonic response to the food examined in each analysis
(e.g., we controlled for liking of the cookies but not the
pretzels in our analysis of cookie intake).
We first looked at the impact of childhood SES and
energy need on the number of grams of cookies that participants consumed. Results revealed a significant interaction
between childhood SES and energy need, b = 6.81, SE =
1.99, t(25) = 3.42, p = .002, semipartial r 2 = .18. We probed
this interaction by examining the impact of childhood SES
on cookie intake at different levels of energy need. When
energy need was high (1 SD above the mean), there were
no differences between participants from high- and low-SES
environments (i.e., 1 SD above and 1 SD below the mean
SES, respectively) in the number of grams of cookies consumed, b = 3.05, p = .17. However, when energy need was
low (1 SD below the mean), there was a negative relationship between childhood SES and cookie intake, with individuals from low-SES childhood environments consuming a
significantly greater number of grams of cookies than those
from high-SES environments, b = −8.78, SE = 2.54, t(25) =
−3.46, p = .002, semipartial r 2 = .18.
We next examined the impact of energy need on
cookie intake at different levels of childhood SES. For
individuals reared in high-SES environments, high selfreported energy need, compared with low energy need,
predicted greater cookie consumption, b = 20.76, SE =
4.54, t(25) = 4.58, p < .001, semipartial r 2 = .32. We did
not observe an effect of energy need on the number of
grams of cookies consumed by participants reared in
low-SES environments, however, b = −0.21, p = .96.
Finally, we looked at the impact of childhood SES and
energy need on the number of grams of pretzels that participants consumed. Our analysis revealed no interaction
between childhood SES and energy need, b = 1.51, p = .19.
621901
research-article2016
PSSXXX10.1177/0956797615621901Hill et al.Hunger, Childhood Socioeconomic Status, and Food
Research Article
Low Childhood Socioeconomic Status
Promotes Eating in the Absence of
Energy Need
Psychological Science
2016, Vol. 27(3) 354–364
© The Author(s) 2016
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797615621901
pss.sagepub.com
Sarah E. Hill1, Marjorie L. Prokosch1, Danielle J. DelPriore1,2,
Vladas Griskevicius3, and Andrew Kramer1
1
Department of Psychology, Texas Christian University; 2Norton School of Family and Consumer Sciences,
University of Arizona; and 3Carlson School of Management, University of Minnesota
Abstract
Life-history theory predicts that exposure to conditions typical of low socioeconomic status (SES) during childhood
will calibrate development in ways that promote survival in harsh and unpredictable ecologies. Guided by this insight,
the current research tested the hypothesis that low childhood SES will predict eating in the absence of energy need.
Across three studies, we measured (Study 1) or manipulated (Studies 2 and 3) participants’ energy need and gave them
the opportunity to eat provided snacks. Participants also reported their SES during childhood and their current SES.
Results revealed that people who grew up in high-SES environments regulated their food intake on the basis of their
immediate energy need; they ate more when their need was high than when their need was low. This relationship was
not observed among people who grew up in low-SES environments. These individuals consumed comparably high
amounts of food when their current energy need was high and when it was low. Childhood SES may have a lasting
impact on food regulation.
Keywords
life-history theory, childhood socioeconomic status, energy regulation, evolutionary-developmental psychology,
thrifty phenotype, eating behavior, open data, open materials
Received 3/12/15; Revision accepted 11/19/15
Obesity is a growing problem in the United States and
around the world (Flegal, Carroll, Kit, & Ogden, 2012).
An important factor that contributes to obesity risk is
childhood socioeconomic status (SES). Several studies
have found low childhood SES to be a major predictor of
obesity and insulin resistance in adulthood (Poulton
et al., 2002; Wells, Evans, Beavis, & Ong, 2010), even
among individuals who are able to improve their conditions later in life (Lawlor, Ebrahim, & Smith, 2002; Power,
Manor, & Matthews, 2003).
Despite growing evidence that low childhood SES
may increase obesity risk, little is known about the
mechanisms that drive this association (Laitinen, Power,
& Jarvelin, 2001). The proposed explanations typically
focus on the environmental conditions of poverty that
promote weight gain in childhood, such as lack of
access to healthy foods and safe places to play (Baltrus,
Everson-Rose, Lynch, Raghunathan, & Kaplan, 2007;
Laitinen et al., 2001). Although these factors undoubtedly
contribute to the link between low childhood SES and
obesity, we propose that exposure to harsh and unpredictable early-life conditions may also become biologically embedded in one’s energy-regulation mechanisms
in ways that promote survival in environments that are
scarce in resources, but promote obesity in those with a
rich food supply.
A person’s childhood environment provides a blueprint for the types of environments likely to be encountered in adulthood. Accordingly, life-history theory
predicts that organisms calibrate their development in
ways that promote survival and reproduction in their
expected adult environments (Belsky, Steinberg, &
Corresponding Author:
Sarah E. Hill, Department of Psychology, TCU, Fort Worth, TX 76129
E-mail: s.e.hill@tcu.edu
Hunger, Childhood Socioeconomic Status, and Food
Draper, 1991; Ellis, Figueredo, Brumbach, & Schlomer,
2009; Kaplan & Gangestad, 2005). Researchers have
therefore hypothesized that exposure to harsh and unpredictable early-life environments should promote the
development of an adult phenotype that is well adapted
to survive in such conditions (Gluckman et al., 2007;
Hales & Barker, 1992; Kuzawa, McDade, Adair, & Lee,
2010). Consistent with this hypothesis, research has
found that exposure to resource scarcity in utero and
during early childhood encourages the development of a
thrifty phenotype, characterized by a small body, slow
metabolism, and reduced level of behavioral activity (see,
e.g., Barker, 1997; Bateson et al., 2004; Bateson & Martin,
1999; Gluckman et al., 2007).
Here, we build on these insights, examining whether
early-life environments may also have a lasting impact
on the mechanisms that guide food intake. Mechanisms
of energy regulation typically develop such that current
energy need plays an important role in regulating food
intake (Havel, 1999; Woods, Seeley, Porte, & Schwartz,
1998). Indeed, people eat more when hungry than when
full. However, in low-SES environments—where there is
diminished access to resources that have historically provided a buffer from food shortages (Gurven & Kaplan,
2007)—it makes adaptive sense to eat when food
is available, even if current energy need is low. Exposure to the conditions typical of low SES during development may therefore undermine the role that bodily
signals of hunger and satiety play in guiding food regulation, promoting eating in the absence of bodily need.
Although eating in the absence of bodily need is associated with obesity in contemporary food-rich environments (Fisher & Birch, 2002; Herman & Polivy, 1984), it
would help promote survival in environments that are
resource scarce.
The Current Research
Here, we present the results of three studies testing the
hypothesis that low childhood SES predicts eating in the
absence of energy need. In each of our studies, we either
measured (Study 1) or manipulated (Studies 2 and 3)
participants’ energy need and gave them the opportunity
to eat provided snacks. We predicted that people who
grew up in high-SES environments would regulate their
food intake on the basis of their immediate physiological
energy need, consuming more calories when need was
high than when need was low. In contrast, we predicted
that physiological energy need would have a negligible
impact on food intake among people who grew up in
low-SES environments. Specifically, we predicted that
their food intake would be comparably high when
bodily energy need was high and when it was low.
355
Study 1
In Study 1, we assessed individuals’ current energy need
by measuring the length of time since their last meal and
their current level of hunger. We then provided participants with an opportunity to eat snack foods (cookies
and pretzels) and measured how many calories they consumed. We predicted that participants who grew up in
high-SES environments would eat more when their
energy need was high than when it was low. However,
we predicted that participants who grew up in low-SES
environments would eat comparably high numbers of
calories when their current energy need was high and
when it was low.
Method
Participants. Thirty-one1 female students at a North
American university (mean age = 19.21 years, SD = 1.26,
range = 18–22) participated in exchange for partial course
credit. Because people from low-SES environments, compared with those from high-SES environments, are more
likely to be obese (e.g., Gonzalez et al., 2012), and
because obesity impairs energy regulation (see, e.g.,
Galic, Oakhill, & Steinberg, 2010), we screened potential
participants in advance and included only those who
were not obese (body mass index < 30); we also excluded
participants who had food allergies or diabetes.
Procedure and materials. Participants came to the
laboratory individually and were told that they would be
participating in a consumer research study. First, they
filled out a survey that contained two key items: the number of hours since they had last eaten and how hungry
they felt (rated on a scale from 1, very full, to 7, very hungry). Responses to these two items were highly correlated
(r = .52) and were therefore transformed into z scores and
averaged to form a measure of current energy need.
Next, participants were told that they would be evaluating some food products as part of a consumer taste-test
study. They were presented with a 3-oz bag of chocolatechip cookies (Famous Amos) and a 0.9-oz bag of pretzels
(Snyder’s). Each snack was presented to participants in a
white Styrofoam bowl. Participants were instructed to
sample each item and to evaluate its flavor by answering
the question: “How much did you like this product?”
(7-point rating scale from 1, dislike extremely, to 7, like
extremely). We included this item both to buttress the
cover story and to provide a control for liking of the food
in our data analysis, as liking is a strong predictor of food
intake (e.g., Spiegel, Shrager, & Stellar, 1989; Yeomans,
Gray, Mitchell, & True, 1997). After tasting and evaluating
each product, participants were told that it would take a
Hill et al.
356
few moments to set up the next part of the study and that
they could eat as much of the remaining food as they
would like while waiting and while completing the
remainder of the study.
After a 2-min waiting period, participants were directed
to complete a survey that asked questions about their
age, height, weight, and childhood SES. We used an
established measure of relative childhood SES (Griskevicius, Delton, Robertson, & Tybur, 2011; Hill, Rodeheffer,
DelPriore, & Butterfield, 2013) as a proxy measure of
exposure to harshness and unpredictability in childhood.
This proxy was chosen because research indicates that
individuals who grow up in lower-SES environments
experience higher levels of morbidity and mortality and
have less stability in their day-to-day life (e.g., more chaotic and unpredictable home environments; Evans, Gonnella, Marcynyszyn, Gentile, & Salpekar, 2005; Jensen,
James, Boyce, & Hartnett, 1983; Miller, Chen, & Parker,
2011). Participants were therefore asked to think about
their childhood before age 12 and rate their agreement or
disagreement with the following statements on 7-point
rating scales (from 1, strongly disagree, to 7, strongly
agree): “My family had enough money for things growing
up,” “I grew up in a relatively wealthy neighborhood,”
and “I felt relatively wealthy compared to others my age.”
Responses to these three items were aggregated to form
an index of childhood SES (α = .87), with higher scores
reflecting higher childhood SES. Although it is possible
that such retrospective accounts are prone to error, past
studies have documented a strong link between adults’
retrospectively reported and actual SES in childhood (S.
Cohen, Janicki-Deverts, Chen, & Matthews, 2010; Duncan, Ziol-Guest, & Kalil, 2010).
The dependent measure consisted of the total number
of calories consumed by each participant. At the end of
each testing session, the uneaten cookies and pretzels
were weighed separately, and the amount of food consumed was calculated by subtracting the remaining
amount of each product from its starting weight. We then
used the information provided on each product’s nutrition label to calculate the total number of calories consumed during the laboratory session.
differences in food intake based on participants’ hedonic
responses to the food items (Spiegel et al., 1989;
Yeomans et al., 1997).
Total calories consumed. In our target analysis, we
examined the impact of childhood SES and energy need
on the total number of calories participants consumed during their laboratory session. Results revealed a significant
interaction between the two predictors, b = 36.94, SE =
13.21, t(24) = 2.80, p = .01, semipartial r2 = .13. We probed
this interaction by conducting regions-of-significance tests
in which we examined the impact of each predictor on
food intake at 1 standard deviation above (high) and
below (low) the mean of the other predictor. First, we
probed this interaction by examining the impact of childhood SES on calorie consumption at different levels of
energy need. When energy need was high, there were no
differences between participants from high- and low-SES
environments in the number of calories consumed,
b = 17.66, p = .26. However, when energy need was low,
there was a negative relationship between childhood SES
and food intake, with individuals from low-SES childhood
environments consuming a significantly greater number
of calories than those from high-SES environments,
b = −46.51, SE = 16.36, t(24) = −2.84, p = .009, semipartial
r2 = .13 (see Fig. 1).
We next examined the impact of energy need on calories consumed at different levels of childhood SES. For
individuals reared in high-SES environments, calorie consumption was greater when self-reported energy need
was high than when it was low, b = 125.11, SE = 30.64,
t(24) = 4.08, p < .001, semipartial r2 = .28 (see Fig. 1). We
did not observe an effect of energy need on the number
of calories consumed by participants reared in low-SES
environments, however, b = 11.36, p = .67.
Cookie and pretzel intake. We next examined separately the impact of childhood SES and energy need on
the number of grams of each snack type (calorically dense
cookies vs. relatively low-calorie pretzels) consumed by
Table 1. Descriptive Statistics for Study 1
Results
Variable
Table 1 presents descriptive statistics for this study. We
used multiple regression to test our predictions. In each
of our analyses, calorie consumption was regressed on
childhood SES and current energy need (both centered)
in the first step and on the interaction between these
variables in the second step. Participants’ body weight
and ratings of how much they liked the food (both centered)2 were also entered in the first step, to control for
differences in energy need based on body weight and
Age (years)
Weight (pounds)
Childhood socioeconomic
status (1–7)
Hours since eating
Hunger (1–7)
Liking of the cookies (1–7)
Liking of the pretzels (1–7)
Total calories consumed
M
SD
Range
19.21
140.61
5.24
1.26
22.98
1.54
18–22
102–211
1.33–7.00
3.97
4.23
5.58
5.45
199.59
3.84
0–10
1.28
2–6
1.29
2–7
0.89
4–5
119.62 42.69–396.51
Hunger, Childhood Socioeconomic Status, and Food
High Energy Need
Low Energy Need
p = .01
p < .001
Total Calories Consumed
350
300
p = .67
250
357
SE = 4.54, t(25) = 4.58, p < .001, semipartial r2 = .32. We
did not observe an effect of energy need on the number
of grams of cookies consumed by participants reared in
low-SES environments, however, b = −0.21, p = .96.
Finally, we looked at the impact of childhood SES and
energy need on the number of grams of pretzels that
participants consumed. Our analysis revealed no interaction between childhood SES and energy need, b = 1.51,
p = .19.
200
Study 2
150
100
50
0
Low Childhood SES
High Childhood SES
Fig. 1. Results from Study 1: total number of calories that participants
consumed as a function of energy need and childhood socioeconomic
status (SES). For both energy need and SES, high refers to values 1
standard deviation above the mean, and low refers to values 1 standard
deviation below the mean. Error bars represent ±1 SEM.
participants. All covariates were the same as in the target
analysis, except that instead of controlling for liking of
both food items, we controlled only for the hedonic
response to the food examined in each analysis (e.g., we
controlled for liking of the cookies but not the pretzels in
our analysis of cookie intake).
We first looked at the impact of childhood SES and
energy need on the number of grams of cookies that
participants consumed. Results revealed a significant
interaction between childhood SES and energy need, b =
6.81, SE = 1.99, t(25) = 3.42, p = .002, semipartial r2 = .18.
We probed this interaction by examining the impact of
childhood SES on cookie intake at different levels of
energy need. When energy need was high (1 SD above
the mean), there were no differences between participants from high- and low-SES environments (i.e., 1 SD
above and 1 SD below the mean SES, respectively) in the
number of grams of cookies consumed, b = 3.05, p = .17.
However, when energy need was low (1 SD below the
mean), there was a negative relationship between childhood SES and cookie intake, with individuals from lowSES childhood environments consuming a significantly
greater number of grams of cookies than those from
high-SES environments, b = −8.78, SE = 2.54, t(25) =
−3.46, p = .002, semipartial r2 = .18.
We next examined the impact of energy need
on cookie intake at different levels of childhood SES.
For individuals reared in high-SES environments, high
self-reported energy need, compared with low energy
need, predicted greater cookie consumption, b = 20.76,
Study 2 was designed to conceptually replicate and extend
the findings of Study 1 using an experimental procedure to
manipulate participants’ energy need. After abstaining
from eating or drinking for at least 5 hr prior to the study,
participants were randomly assigned to drink either a beverage containing calories (Sprite) or a beverage devoid of
calories (sparkling water). We then provided all participants with an opportunity to eat snacks and asked them to
report on their childhood and current SES.
We predicted that participants who drank the water
would consume comparably high numbers of calories
across levels of childhood SES. However, we predicted
that participants’ childhood SES would moderate the
impact of drinking Sprite on subsequent calorie consumption. Specifically, we predicted that among individuals from high-SES childhood environments, those who
consumed the sugar-sweetened drink would eat less than
those who consumed water. In contrast, we predicted
that among individuals reared in low-SES environments,
individuals who had received a glucose boost from the
sugar-sweetened drink and those who had consumed
water would eat comparably large amounts of food.
Additionally, we predicted that the moderating effect of
SES would be specific to childhood SES and would not
emerge for current SES.
Method
Participants. Sixty female students at a North American university (mean age = 19.33 years, SD = 1.55,
range = 18–24; 31 in the Sprite condition and 29 in the
water condition) participated in exchange for partial
course credit. We screened potential participants in
advance and included only those who were not obese
(body mass index < 30); we also excluded individuals
who had food allergies or diabetes. Participants were
instructed to avoid eating or drinking anything other than
water for at least 5 hr prior to their study session. Five
participants (4 in the Sprite condition, 1 in the water condition) were removed from analyses because they did
not comply with the fasting procedure prior to their session or failed to finish their drink.
Hill et al.
358
Procedure. Participants came to the laboratory individually and were told that they would be consuming a beverage as part of a consumer research study on taste
preferences. Upon being seated, they were asked to indicate how many hours it had been since they had eaten
and how hungry they felt (same 7-point rating scale as in
Study 1). After responding to these initial questions, participants were given an unmarked, red plastic cup containing 12 oz of either a sucrose-sweetened soda (Sprite)
or an unsweetened sparkling mineral water (La Croix).
They were given 2 min to drink the beverage and were
then asked follow-up questions about their enjoyment of
it (to buttress the cover story). They then completed a
10-min filler task (listing consumer brand names) that
allowed time for changes in blood glucose level to occur
(Aarøe & Petersen, 2013; Wang & Dvorak, 2010).
Following the filler task, participants were informed
that they would next be asked to evaluate a food item.
They were presented with cookies (the contents of a
1-oz bag of mini Oreo cookies), which were served in a
white bowl. After tasting and evaluating the cookies
(“How much do you like this product?”; rating scale
from 1, dislike extremely, to 7, like extremely), participants were told that that they could eat as much of the
remaining food as they liked while waiting for the next
survey to load on the computer and while they completed the survey questions. Participants were then
asked to report their body weight, their childhood SES,
and their current SES. Childhood SES was measured by
the three-item measure of relative childhood SES used
in Study 1, as well as by a more objective single-item
measure: “Based on your best estimate, what was your
family’s socioeconomic status during your early childhood (age 12 and earlier)?” Current SES was measured
with an analogous item, “Based on your best estimate,
what is your family’s socioeconomic status currently?”
Participants responded to these additional SES items on
a 7-point scale (1 = very poor, 7 = very wealthy). We
included these two additional SES measures to test
whether our pattern of results would be replicated using
a more objective measure of childhood SES and to
determine whether only childhood SES, and not adult
SES, moderates the effect of energy need on calorie
consumption, as our theory predicts. The study ended
with a question probing participants’ compliance with
the fasting procedure. Participants read, “Please note
that your response will not affect your ability to receive
credit for participation in this study. How many hours
has it really been since you last ate or drank anything
other than water before today’s study began?”
The dependent measure in Study 2 was the total number of calories consumed during the laboratory session,
which we calculated by taking the difference between
the starting weight of the cookies and the weight of the
uneaten cookies and using the calorie information provided on the nutrition label.
Results
Table 2 presents descriptive statistics for this study. We
used multiple regression to test our predictions. In our
analysis, number of calories consumed was regressed on
drink condition (dummy-coded) and SES (centered) in
the first step and on the interaction between these variables in the second step. As in Study 1, participants’ body
weight and ratings of how much they liked the food
(centered) were also entered in the first step to control
for differences in consumption based on energy need
and hedonic response. We also included the number of
hours it had been since participants had last eaten as a
covariate, as the length of the presession fasting period
would also influence the degree to which the fixed number of calories administered (via the soft drink) affected
participants’ energy needs.
We first examined the effect of energy need and childhood SES on calorie consumption using the single-item
measure of childhood SES. This analysis revealed a significant interaction between drink condition and childh...
Purchase answer to see full
attachment