Você está na página 1de 12

Family Relations, 56 (January 2007), 80–91. Blackwell Publishing.

Copyright 2007 by the National Council on Family Relations.

The Predictors of Parental Use of Corporal Punishment

Andrew Grogan-Kaylor Melanie D. Otis*

Abstract: Corporal punishment has been the focus of considerable study over the past decade. Some recent research
suggesting that the use of corporal punishment may have significant long-term negative effects on children has
prompted increasing exploration and interest in the issue. We used tobit regression analysis and data from the 2000
National Longitudinal Survey of Youth to examine both the prevalence and the chronicity of spanking in a nation-
ally representative sample of parents. Mother’s characteristics (e.g., age, education) and neighborhood context did
not show a relationship with parental use of corporal punishment. Among parents who used corporal punishment,
being Protestant had a relatively large relationship with its use. Although children’s externalizing behaviors had some
association with parent’s propensity to spank, findings suggest that use of corporal punishment may be better under-
stood as part of a constellation of behaviors relating to a parenting style. Further, findings indicate that it is easier
to predict the incidence of corporal punishment than to predict its frequency of use.

Key Words: corporal punishment, discipline, parent-child relationships, tobit regression.

Debate over appropriate and effective parental disci- that predict parents’ propensity to use corporal pun-
plinary tactics frequently focuses on questions about ishment. A second goal of the analysis was to exam-
corporal punishment. Research suggests that many ine more closely the characteristics of those parents
parents have spanked their children at some point who indicated the use of corporal punishment to
(Gershoff, 2002; Grogan-Kaylor, 2004). It is likely address the question of whether there exist individ-
that many parents who use corporal punishment ual and community differences related to the fre-
consider it to be an elicited response to a child’s quency of use.
problematic behavior. Over the past two decades,
however, an increasingly rigorous body of research Ecological Theory and Corporal
has suggested that corporal punishment contributes
Punishment Utilization
to increases in a broad spectrum of child behavior
problems (Gershoff; Grogan-Kaylor). Indeed, some
important research suggests that measurable out- As previously noted, research examining predictors
comes of parental use of corporal punishment may of corporal punishment has focused primarily on
persist well into adulthood (MacMillan et al., 1999; individual characteristics of parents and children,
Straus, 2000). This emerging picture of the use of and parental expectations about the effectiveness of
corporal punishment as a potential contributor to the tactic in achieving disciplinary goals. Little atten-
the continuation of behavior problems suggests the tion, however, has been given to ways that social
need for better understanding of the individual context and cultural norms may influence both the
and contextual factors that may relate to corporal prevalence and the chronicity of corporal punish-
punishment use. ment. An ecological framework provides both the
The purpose of the current study was twofold. tools and the impetus for considering the influence
Guided by an ecological framework, we first sought of environment on parental and child behavior
to identify those contextual and individual factors (Bronfenbrenner, 1979, 1986). Ecological theory

*Andrew Grogan-Kaylor is an Assistant Professor in the University of Michigan School of Social Work, 1080 S University Avenue, Ann Arbor, MI 48109 (agrogan@
umich.edu). Melanie D. Otis is an Associate Professor in the College of Social Work, University of Kentucky, 651 Patterson Office Tower, Lexington, Kentucky
40506-0027 (melanie.otis@uky.edu).
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 81

suggests tenets that guided our decision to consider the growing number of female households has
both individual and community factors in our further significance because of the relationship
efforts to understand the nature of corporal punish- between stressful economic conditions and fre-
ment utilization. The model underscores the interre- quency of use of corporal punishment (Giles-Sims,
latedness of people and their social and physical Straus, & Sugarman, 1995; McLoyd, 1990; Sariola
worlds. Parenting does not occur in a vacuum. & Uutela, 1992).
Instead, parents’ actions are often viewed and judged The relationship between socioeconomic status
by others around them—neighbors, friends, and (SES) and the use of corporal punishment has been
community members. Reactions in these settings, as less consistent. Although some studies have linked
well as the parents’ own observations of others’ low SES to physical abuse of children (Garbarino,
actions, provide information that may or may not 1977; Straus & Gelles, 1986), others have failed to
support the use of corporal punishment. Through find a significant relationship between SES and the
these reciprocal relationships, parents may come to use of corporal punishment (Hashima & Amato,
view corporal punishment as the preferred method 1994; Straus, 2000; Straus & Donnelly, 1993). A
of discipline, as inappropriate only under certain cir- more fruitful approach to understanding the rela-
cumstances, or as inappropriate regardless of the cir- tionship between individual economic conditions
cumstances of the child’s actions. and the use of corporal punishment may be found
The current study used an ecological framework in considering the relationship between economic
to develop a fuller picture of the nature of corporal factors and stress. To that end, studies have found
punishment with the inclusion of factors that under- significant links between poverty status (i.e., use of
score the dynamic relationship between people and Aid to Families with Dependent Children [AFDC],
their environment. Specifically, parent and child unemployment, percent of poverty level) and the
characteristics, parental expectations, and child behav- likelihood of using corporal punishment (Giles-Sims
ior, as well as community factors such as region of et al., 1995). For example, in their study examining
the country and neighborhood quality, were consid- factors potentially related to the use of corporal pun-
ered collectively. ishment among mothers of 3- to 5-year-olds, Giles-
Sims et al. found that being an AFDC recipient
alone had a significant impact on the frequency of
Literature Review spanking; however, these authors suggested that
their findings supported a stress theory explanation
of the use of corporal punishment by mothers,
Parents’ Characteristics
rather than a specific socioeconomic explanation
Past research has suggested that mothers are likely (Dietz, 2000; Giles-Sims et al.).
to spank their children more often than fathers Past research has also supported a link between
(Nobes, Smith, Upton, & Heverin, 1999; Wolfner certain religious affiliations and the likelihood that
& Gelles, 1993). Frequently, studies have focused a parent will use corporal punishment. In general,
specifically on mothers’ decisions to use corporal these studies have suggested that Conservative Prot-
punishment, arguing that mothers are more likely estants are more likely to use corporal punishment
to be the family disciplinarian by virtue of their as a disciplinary tactic than parents with other reli-
increased level of contact with the child(ren) gious affiliations (Ellison, Bartkowski, & Segal,
(Dietz, 2000; Straus & Donnelly, 1993). How- 1996). Ellison (1996) asserted that these findings
ever, with the growing number of working may be related to ‘‘religious differences in the
mothers, the likelihood that mother, rather than response decision stage’’ where parents assess poten-
father, may have greater contact with the child has tial actions that might be taken to address their
changed somewhat. Thus, a contemporary explana- child’s behavior problem. At this stage, it was
tion for the predominance of female disciplinarians argued that individuals from more conservative
may be the large number of female-headed house- fundamentalist religious backgrounds will be more
holds, which increases the likelihood that mother likely to view corporal punishment not only as a via-
may be the disciplinarian, independent of her ble option but also potentially as the most appropri-
attachment to the workforce. In addition to the ate response to the infraction—the ‘‘spare the rod’’
increased contact between mothers and children, philosophy of childrearing.
82 Family Relations  Volume 56, Number 1  January 2007

Parental Expectations examined less thoroughly. When they are addressed,


Parents who view positive parenting as an effective community size and region of the country are
tool for influencing child behavior may consider among the most frequently examined contextual fac-
corporal punishment as inappropriate and even anti- tors. In general, the presumption associated with the
thetical to achieving the desired results. In contrast, inclusion of these variables has related to cultural
parents who feel their positive efforts will do little to aspects of community that are believed to influence
affect the wanted outcome may be more likely to disciplinary norms. Past research suggests that par-
use punitive childrearing practices (Luster & Kain, ents in rural areas and in the South are more likely
1987). Further, some parents who support the use to use corporal punishment than those in urban
of corporal punishment in specific situations are areas and other regions of the country (Giles-Sims et
aware of the limited effectiveness of the method, yet al., 1995). In addition to cultural factors, these stud-
feel that some instances justify its use (Graziano, ies suggest that mothers in rural areas may have
Hamblen, & Plante, 1996; Holden, Miller, & fewer support resources and less access to/awareness
Harris, 1999). of information on alternatives to spanking than
mothers in urban areas.
Parenting Style
The use of corporal punishment should also be Incidence and Frequency of Spanking:
examined in the larger context of differences in par- Where Are the Differences?
enting styles. To this end, it has been suggested that As previously noted in some of the extant literature,
spanking may be one part of an overall approach to whether a parent spanks and how often he or she
parenting that is devoid of positive input from the spanks may be explained by a variety of factors.
parent. This approach is further characterized by First, use of corporal punishment continues to be
lack of affection and a general climate of aggression, common in the United States, with over 90% of
regardless of whether physical punishment occurs parents of toddlers indicating that they have spanked
(Simons, Johnson, & Conger, 1994; Socolar & their child at least once (Straus, 2000; Straus &
Stein, 1995). Stewart, 1999). For instance, Giles-Sims et al.’s
(1995) examination of data from the 1990 National
Child Characteristics Longitudinal Survey of Youth (NLSY) on maternal
spanking found that approximately 61% of mothers
Research has shown that parents continue to spank
of 3- to 5-year-olds had spanked their child in the
children into their early adolescent years. Although
past week. In fact, the majority of parents in many
most parents of preschool-age children (0 – 4 years)
studies indicated that they have spanked their child
indicate that they have spanked their children, this
at some point in time (Giles-Sims et al.; Straus,
rate declines substantially as children age (Frick,
1983); however, this does not necessarily mean that
Christian, & Wootton, 1999; Giles-Sims et al.,
parents are motivated to spank in response to all dis-
1995; Wauchope & Straus, 1990).
ciplinary concerns.
Gender has also been shown to relate to the like-
In addition to differences in whether or not
lihood that a child will be spanked (Day, Peterson, &
parents engaged in corporal punishment at all,
McCracken, 1998; Lytton & Romney, 1991; Wolfner
there may be important differences in the fre-
& Gelles, 1993). Studies have consistently indicated
quency with which parents employ corporal pun-
that male children are more likely to be spanked.
ishment. For instance, in Giles-Sims et al.’s (1995)
Additionally, Giles-Sims et al. (1995) found that
study, the frequency of spanking ranged from once
regardless of the age of the child, boys are more
(33.1%) to 15 or more times (1.5%), with an
likely to be spanked than girls.
average of three spankings occurring during the
previous week. The wide variation in whether
Neighborhood Context
a parent has spanked their child, and the fre-
Although several studies have examined the relation- quency with which such disciplinary actions are
ship between specific parental and child characteris- taken, suggests that a variety of factors may be at
tics and the likelihood and frequency of the use of work in predicting the likelihood that a parent will
corporal punishment, contextual factors have been use corporal punishment.
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 83

Summary and Hypotheses result in biased and inconsistent parameter estimates


On the basis of our review of the literature, we (Greene, 1997).
examined the following research questions: Indeed, the measurement strategy employed by
the NLSY gives further weight to these concerns.
1) What are the effects of individual and contextual
The NLSY asks parents about the number of times
factors on the latent propensity to use corporal
they have spanked their children in the past week.
punishment?
Therefore, parents who have spanked their children
2) How can such an analysis be extended to under-
in the past month or year, but not in the past week,
stand the effects of individual and contextual fac-
are recorded as nonspankers. Arguably, the measures
tors on the decision of whether or not to use
of corporal punishment in the NLSY may result in a
corporal punishment?
variable that is left censored by the data collection
3) How can such an analysis be extended to under-
process. As discussed in the original formulation by
stand the effects of individual and contextual
Tobin (1958) and in an important expository piece
factors on the frequency of use of corporal punish-
by Roncek (1992), tobit models are admirably suited
ment for those who engage in this practice?
to modeling such a left-censored dependent variable.
The analytic strategy, which has come to be
known as the tobit regression, was originally devel-
oped by Tobin (1958) in a study of the yearly
amount of money that households spent on large
Methods purchases. Recognizing that many households do
not spend any money on large purchases in a particu-
lar year, Tobin developed a regression method that
Analytic Strategy
was appropriate for modeling continuous variables
Unfortunately, standard analytic techniques are not with a large number of zeroes. Over time, Tobin’s
well suited to the simultaneous examination of the model, which is in some ways statistically a general-
incidence and frequency of parental use of corporal ization of the technique known as probit regression,
punishment. For example, ordinary least squares has come to be known as Tobin’s probit or, more
regression can be used to treat the number of spank- simply and more frequently, as the tobit (Greene,
ings in the past week as a continuous variable rang- 1997).
ing from zero upward. However, a well-known Tobit analyses have been used in previous empir-
property of ordinary least squares regression is that ical work to model contributors to violent or antiso-
within the linear regression framework, every incre- cial behavior because of concerns about large
ment of a continuous dependent variable is treated numbers of zero values in the dependent variable, as
as an equivalent step. Thus, modeling parental use well as concerns about censoring mechanisms. For
of corporal punishment in this fashion means that example, Grogan-Kaylor and Otis (2003) used tobit
the transition from no spankings to one spanking is regression to examine the contribution of child
treated exactly the same as the transition from one abuse and neglect to adult arrest, whereas Jacobs and
spanking to each additional spanking, thus conflat- O’Brien (1998) used the tobit model to examine the
ing the incidence of corporal punishment with the use of deadly force by police departments. Although
frequency of corporal punishment. parental use of corporal punishment does not
Additionally, the distribution of corporal punish- approach the level of severity of violence examined
ment in a large sample such as the NLSY raises some in these two research studies, the general principles
concerns about whether corporal punishment as of a dependent variable with a distribution that is
a dependent variable may be appropriately modeled skewed toward zero and the possibility of censoring
using standard ordinary least squares regression. in the dependent variable render tobit modeling
Because there are many parents who do not spank appropriate.
their children on a regular basis, or at all, the value Furthermore, important work by McDonald and
of corporal punishment in the NLSY data has many Moffit (1980) demonstrated a method for decom-
zeroes. Econometricians have long pointed out that posing, or taking apart, the coefficients of a tobit
the use of ordinary least squares regression techni- model into two separate sets of coefficients. The
ques to analyze variables with many zeroes may first set of coefficients from this decomposition
84 Family Relations  Volume 56, Number 1  January 2007

procedure applies to the question of whether or not Table 1. Descriptive Statistics


study participants have a nonzero value for the Variable
dependent variable. When the dependent variable of
interest is corporal punishment, this set of coeffi- Continuous variables M (SD)
cients thus refers to the incidence of parental use of Number of spankings 0.26 (0.86)
corporal punishment. The second set of coefficients in past week
applies to the question of the value of the dependent Child’s age in months 111.29 (27.75)
variable for those study participants with a nonzero Mother’s years 13.66 (2.36)
value of the dependent variable, which in the case of of education
corporal punishment may be thought of as the fre- Mother’s age in years 38.57 (2.18)
quency of spanking for those parents who actually Cognitive stimulation 99.68 (14.39)
spank. Therefore, the method of decomposing tobit Externalizing behavior 100.97 (13.48)
coefficients proposed by McDonald and Moffit Internalizing behavior 100.19 (13.77)
provides a unique opportunity to model simulta- Neighborhood quality 0.12 (0.61)
neously, yet separately, the predictors of parental use Categorical variables Frequency (%)
of corporal punishment and the frequency with Gender of child
which corporal punishment is employed. Male 426 (53.25)
Before engaging in multivariate analyses, a corre- Female 374 (46.75)
lation matrix of all independent variables was exam- Race of child
ined for evidence of multicollinearity or high Latino 119 (14.88)
correlations between independent variables. No cor- Black 156 (19.5)
relations high enough to warrant concerns about White 525 (65.63)
these issues were detected. Poverty status
Not in poverty 677 (84.63)
In poverty 123 (15.38)
Sample
Religious affiliation
The sample used in this study was derived from Protestant 428 (53.5)
the NLSY. Begun in 1979, the NLSY is an Catholic 231 (28.88)
ongoing nationally representative panel study of Other 65 (8.13)
men and women (Center for Human Resource None 69 (8.63)
Research, 2000) and has been used in hundreds of
publications on children and families. Since 1986,
the NLSY has also collected information every 2
Measures
years on the children born to women in the origi-
nal analysis sample (Center for Human Resource Information on parental usage of corporal punish-
Research) and joined records of mothers and their ment was obtained from the NLSY from a question
children were the data source for this analysis. In that asked parents about the number of times they
some cases, multiple children per mother were had spanked a particular child within the past week.
interviewed as part of the NLSY. In such cases, Respondents were allowed to define spanking as
one child from the family was randomly selected they saw appropriate.
for participation so the analysis would not violate In addition to parental use of corporal punish-
assumptions about the independent and identical ment, the NLSY also provides information on the
distribution of error terms. To make use of recent level of cognitive stimulation that parents provide to
information, data from the 2000 cohort of the children in the home. Originally devised by Bradley,
NLSY were used for this analysis. The final sample Caldwell, and Elardo (1979), the scale measuring
size for this analysis was 800. Descriptive statistics cognitive stimulation includes items on the number
for the sample are provided in Table 1. Because of books within the home, the frequency with which
the decompositions of the tobit parameters that the child is taken on educational outings, and the
are reported in this paper rely on the first tobit amount of time that parents spend talking with the
analysis, all analyses in this paper rely on this final child (Center for Human Resource Research, 2000).
sample of 800. The cognitive stimulation scale is normed to have
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 85

a mean of 100 and a standard deviation of 15 in the Table 2. Tobit Regression Model
overall NLSY sample (Center for Human Resource Variable Coefficient SE t p
Research).
Information on children’s demographic charac- Child’s gender (male)
teristics such as age, race, and gender were easily Female 0.080 0.326 0.25 .81
obtained from the NLSY. The NLSY also contains Child’s race (White)
a measure of a family’s poverty status. This measure Latino 0.194 0.581 0.33 .74
is computed by the staff at the Center for Human Black 1.231 0.416 2.96 ,.01
Resources Research at the Ohio State University and Child’s age in months 20.042 0.006 27.00 ,.01
is the result of a comparison of a family’s total net Externalizing behavior 0.107 0.016 6.49 ,.01
income for a given year with the poverty standards problems
provided by the U.S. Census for that year. Internalizing behavior 20.048 0.015 23.16 ,.01
The NLSY also contains questions from the problems
Behavior Problems Index (BPI; Zill & Peterson, Poverty status
1986), an instrument designed to assess the level of (not in poverty)
children’s behavior problems. Drawing largely on the In poverty 0.420 0.456 0.92 .36
Child Behavior Checklist (Achenbach & Edelbrock, Mother’s education 20.060 0.074 20.81 .42
1983), the BPI contains a shorter set of items and is Mother’s age 20.094 0.075 21.24 .21
designed to be administered efficiently within the Cognitive stimulation 20.025 0.012 22.03 .04
context of a larger survey. Research indicates that provided by parents
the BPI factors well into two broad subscales of Neighborhood quality 0.352 0.301 1.17 .24
externalizing and internalizing behavior problems Religious affiliation
(Hofferth, Davis-Kean, Davis, & Finkelstein, 1997). (Protestant)
Externalizing behaviors encompass phenomena such Catholic 21.244 0.469 22.65 .01
as whether children are antisocial or have problems Other 20.870 0.603 21.44 .15
with friends, whereas internalizing behaviors include None 21.095 0.658 21.66 .10
anxiety and depression (Center for Human Resource Region (Northeast)
Research, 2001). North Central 20.609 0.551 21.11 .27
South 0.027 0.510 0.05 .96
West 20.289 0.599 20.48 .63
Results Constant 2.394 3.639 0.66 .51
R 2.714 0.204

Tobit Model Coefficients—Parents’ Propensity to Use Note: Reference categories are in parentheses.
Corporal Punishment
All tobit analyses were conducted using STATA sta- of Latino children were no more likely to use corpo-
tistical software. The tobit model yielded a number ral punishment than parents of White children.
of interesting results, which are provided in Table 2. Children’s age was negatively associated with the use
In interpreting coefficients from a tobit regression of corporal punishment because parents of older
model, it is important to remember that these children were less likely to use corporal punishment.
coefficients are best interpreted as the effects of a set Many family characteristics were not associated
of independent variables on a latent propensity with parental use of corporal punishment. For
(Roncek, 1992)—in this case, the latent propensity example, a mother’s education and age were not
to use corporal punishment. associated with the use of corporal punishment.
Only some of children’s demographic character- However, higher levels of cognitive stimulation pro-
istics were associated with corporal punishment. vided in the home was associated with lower levels
Specifically, a child’s gender was not associated with of corporal punishment. Religious affiliation was
whether or not her or his parents used corporal pun- associated with parental use of corporal punishment
ishment. Also, when compared with parents of and generated the largest effect sizes of any variables
White children, parents of Black children were more in the model. Protestant parents were most likely to
likely to use corporal punishment; however, parents use corporal punishment; in comparison, Catholic
86 Family Relations  Volume 56, Number 1  January 2007

parents were less likely to use corporal punishment. Table 3. Decomposition of Tobit Coefficients: Probability
Parents who reported that they had no religious That Child Was Spanked
affiliation did not differ from Protestant parents in Variable dF/dx (%) SE z p
proclivity to use corporal punishment.
There were interesting relationships between Child’s gender (male)
children’s behavior and the use of corporal punish- Female 0.49 0.020 0.25 .81
ment. Children who displayed greater amounts of Child’s race (White)
externalizing behavior problems were more likely to Latino 1.22 0.035 0.34 .73
be the recipients of corporal punishment, whereas Black 8.95 0.025 3.53 ,.01
children who displayed greater levels of internalizing Child’s age in months 20.25 0.000 27.00 ,.01
behavior problems were less likely to be the recipi- Internalizing behavior 20.29 0.001 23.16 ,.01
ents of corporal punishment. In interpreting the problems
effect sizes of these coefficients, it is worth recalling Externalizing behavior 0.65 0.001 6.49 ,.01
that the measurement of children’s behavior was problems
drawn from the BPI (Zill & Peterson, 1986) in Poverty status
which data were normed so that the standard devia- (not in poverty)
tion of both the internalizing and the externalizing In poverty 2.74 0.028 0.99 .32
behavior problems scales was set at 15. Mother’s education 20.36 0.004 20.81 .42
Some larger community-level variables were also Mother’s age 20.57 0.005 21.24 .21
included in the model. Neither the characteristics of Cognitive stimulation 20.15 0.001 22.03 .04
the neighborhood nor the geographic region in Religion (Protestant)
which the family lived showed a relationship with Catholic 26.68 0.029 22.34 .02
parental use of corporal punishment. Other 24.41 0.037 21.20 .23
None 25.31 0.040 21.33 .19
Region (Northeast)
First Decomposition of Tobit Parameters—Effects North Central 23.47 0.034 21.04 .30
of Independent Variables on Use Versus Nonuse South 0.16 0.031 0.05 .96
of Corporal Punishment West 21.69 0.036 20.46 .64
In the second step of the analysis, the coefficients of Neighborhood quality 2.14 0.018 1.17 .24
the tobit model were decomposed into two more Constant 14.57 0.221 0.66 .51
informative sets of parameters using the procedure
Note: Marginal effects: probability uncensored. Incidence: probability that child was
outlined by McDonald and Moffit (1980). Statisti- spanked at least once. Reference categories are in parentheses. dF/dx ¼ change in out-
cally, this ‘‘decomposition’’ may be seen as a process come associated with change in x.
of disentangling—or taking apart—these estimates
into two sets of estimates that are more intuitively a parent would use corporal punishment. Higher
meaningful. The first set of secondary parameters levels of children’s externalizing behavior problems
indicated the extent to which each independent vari- were associated with an increased probability of
able had a relationship with the probability that using corporal punishment, whereas higher levels of
a particular parent would or would not use corporal children’s internalizing behavior problems were
punishment (see Table 3). Parents of Black children associated with a lower probability of corporal pun-
were almost 9% more likely to use corporal punish- ishment. Higher levels of cognitive stimulation were
ment than were White parents. The probability of associated with lower probabilities that corporal
parents using corporal punishment decreased by punishment would be used. However, neither
a quarter of a percent for every month the child aged mothers’ education nor age was associated with cor-
resulting in a 4% reduction in the probability that poral punishment. The internalizing behavior prob-
corporal punishment would be used for every addi- lems score, externalizing behavior problems score,
tional year in children’s age. However, poverty status and cognitive stimulation score were all normed to
did not have an association with whether or not par- have standard deviations of 15. Multiplying the
ents used corporal punishment. parameter estimates for each of these independent
Both internalizing and externalizing behavior variables by 15 suggested that a standard deviation
problems were associated with the probability that change in these quantities was associated with
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 87

a 4.4% decrease, 9.7% increase, and 2.3% decrease Table 4. Decomposition of Tobit Coefficients: Effect on
in the use of corporal punishment, respectively. Number of Uses of Corporal Punishment for Parents Who
As expected, religion was related to the use of Employed Corporal Punishment
corporal punishment. Catholic parents were almost Variable dF/dx SE z p
7% less likely to use corporal punishment than Prot-
estant parents. But neither geographic region nor Child’s gender (male)
neighborhood quality was associated with corporal Female 0.013 0.054 0.25 .81
punishment usage when accounting for the presence Child’s race (White)
of other variables in the model. Latino 0.032 0.096 0.34 .74
Black 0.220 0.069 3.21 ,.01
Second Decomposition of Tobit Parameters—Effects Child’s age in months 20.007 0.001 27.00 ,.01
of Independent Variables on Frequency of Use Internalizing behavior 20.008 0.003 23.16 ,.01
of Corporal Punishment problems
A second set of parameters was also obtained from Externalizing behavior 0.018 0.003 6.49 ,.01
the decomposition of the tobit coefficients. This sec- problems
ond set of parameters indicated the relationship of Poverty status
each of the independent variables on the number of (not in poverty)
spankings used by parents in the past week, condi- In poverty 0.071 0.075 0.95 .34
tional on those parents having used one or more cor- Mother’s education 20.010 0.012 20.81 .42
poral punishments. Table 4 provides data on this Mother’s age 20.015 0.012 21.24 .21
second set of parameters. Again, children’s gender did Cognitive stimulation 20.004 0.002 22.03 .04
not have an association with the number of times that Religion (Protestant)
corporal punishment was employed, nor did a child’s Catholic 20.195 0.077 22.52 .01
Latino ethnicity. On the other hand, Black children Other 20.133 0.099 21.34 .18
received 0.22 more spankings than White children in None 20.165 0.108 21.52 .13
families where corporal punishment was employed. Region (Northeast)
As expected, every month increase in a child’s age was North Central 20.098 0.091 21.08 .28
associated with a decrease in the number of spankings South 0.004 0.084 0.05 .96
employed, resulting in a 0.12 decrease in the number West 20.047 0.099 20.47 .64
of spankings used for every year that the child aged. Neighborhood quality 0.058 0.050 1.17 .24
In summary, children’s behavior problems were Constant 0.394 0.599 0.66 .51
related to the number of uses of corporal punishment
Note: Marginal effects: conditional on being uncensored. Frequency: effect of inde-
in families that employed corporal punishment. As pendent variables on number of spankings, given that child is spanked. Reference cate-
with the previous analyses, higher levels of children’s gories are in parentheses. dF/dx ¼ change in outcome associated with change in x.
externalizing behaviors were associated with a higher
likelihood of parental use of corporal punishment.
Religion had a relatively large relationship with the
Conversely, children who demonstrated greater levels
frequency of corporal punishment among parents
of internalizing behavior problems were less likely to
employing physical punishment; however, compared
experience corporal punishment. The relationship of
with Protestant parents, Catholic parents spanked
externalizing behavior problems with the use of cor-
much less frequently. As expected, geographic region
poral punishment was greater than the relationship of
and neighborhood quality did not have an association
internalizing behavior problems with the use of cor-
with the number of times parents who used corporal
poral punishment.
punishment actually spanked their children.
As expected, many characteristics of mothers
were not associated with whether or not children
experienced corporal punishment. Mother’s age and
Discussion
mother’s level of education were not associated with
higher levels of use of corporal punishment among
parents who used corporal punishment. Higher levels Concern over the parental use of corporal punish-
of cognitive stimulation were associated with lower ment has been elevated by a growing body of litera-
levels of the use of corporal punishment. ture linking the disciplinary tactic to a number of
88 Family Relations  Volume 56, Number 1  January 2007

negative child welfare outcomes (Gershoff, 2002; only five items that correlate highly with each other
Grogan-Kaylor, 2004; Straus, 2000). In this study, and appear to represent a single dimension of neigh-
we sought to expand our understanding of corporal borhood quality.
punishment by using an ecological framework to Other research, which has relied on data with
examine variation in both the rates and the chronicity measures of multiple dimensions of neighborhood
of corporal punishment among a random sample of quality has suggested that adequately accounting
parents. Our analyses provide only partial support for for the characteristics of a neighborhood may do
our model, with race and ethnicity, religion, child’s a great deal to explain racial and ethnic disparities in
behavior, and parental use of cognitive stimulation child outcomes (Woolley & Grogan-Kaylor, 2006).
being the most consistent predictors of variation in Woolley and Grogan-Kaylor explored the effect of
corporal punishment frequency and chronicity. family and neighborhood characteristics on three
An important limitation of cross-sectional school-related outcomes and found that in some
research of this nature is an inability to ascertain cases the inclusion of neighborhood characteristics
causal direction. There are also some other limita- in a statistical model helped to fully explain racial
tions to this research. First, the NLSY simply asks and ethnic disparities in academic outcomes. This
survey respondents about the number of times that finding calls to mind the assertion in seminal articles
they have spanked their child within the past week. by Massey (1990, 1994) that African American fam-
Thus, it is left up to respondents to define the mean- ilies are often likely to live in very different neigh-
ing of spanking for them. Additionally, although the borhoods from White families. Unfortunately, the
NLSY data contain important information about data set used by Woolley and Grogan-Kaylor did
the incidence and frequency of corporal punish- not contain measures of parental discipline, render-
ment that forms the basis of this analysis, important ing direct tests of these ideas impossible. More
information on the severity with which corporal research on parental discipline that makes use of
punishment is applied is not present in the NLSY data with rich neighborhood indicators is likely to
data. be fruitful.
Bearing these limitations in mind, our analysis is Our findings also suggest that mothers’ decisions
consistent with past research showing that, in com- on whether or not to employ corporal punishment
parison with White families, mothers in African are associated with their child’s behavior. Notably,
American families were more likely to employ corpo- the present research suggests that even when other
ral punishment (Giles-Sims et al., 1995; McLoyd & factors are included in a statistical model, children’s
Smith, 2002). Interestingly, the inclusion of a var- actual behavior is only one of several factors associ-
iable for poverty into the model did not eliminate ated with parental use of corporal punishment.
the statistical significance of race, indicating that fac- Indeed, when the parameter estimates associated
tors other than economic status may play a role in with changes in children’s behavior problems were
the degree to which parents make use of corporal multiplied by 15 to obtain an estimate of the
punishment as a disciplinary strategy. Horn, Cheng, changes in parental corporal punishment related to
and Joseph’s (2004) examination of disciplinary a standard deviation change in behavior problems,
practices among lower SES and middle-to-upper several other factors in the model had comparable
SES African American parents reported similar find- effect sizes. These findings suggest that cultural and
ings, with few SES-related differences noted in their religious factors are likely to play a significant role in
sample of 175 African American parents. affecting whether or not parents employ corporal
Similarly, in this analysis, the inclusion of mea- punishment. Unfortunately, the NLSY data do not
sures of neighborhood quality does not reduce the contain finely grained measures of respondents’ reli-
relationship of family race or ethnicity with parental gious beliefs, rendering further exploration of the
use of corporal punishment. Indeed, the measure of religion measures in this analysis impossible. Further
neighborhood quality is not predictive of parental research of this nature is likely to benefit from the
use of corporal punishment, which stands in con- use of data sets that contain more sophisticated mea-
trast with other research exploring this relationship sures of religion than those present in the NSLY.
(Giles-Sims et al., 1995). To some extent, this lack Similar to other research suggesting that parental
of statistical significance may be a function of the use of corporal punishment may be part of a pattern
fact that the neighborhood scale in the NLSY has of parenting characterized by less use of positive
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 89

parenting tactics (Grogan-Kaylor, 2004), the results parents who spank, a standard deviation increase in
of this research provide evidence that parental use of children’s externalizing behavior problems is associ-
corporal punishment is associated with a lower inci- ated with an additional 0.27 uses of corporal pun-
dence of parental provision of a cognitively stimulat- ishment. Thus, interventions targeted at changing
ing environment for children. Such a finding parents’ use of corporal punishment may have more
suggests that corporal punishment may be part of effect on parents’ decision on whether or not to use
a parenting style with lesser amounts of positive par- corporal punishment, rather than on the frequency
enting practices. Given the effect size of parental use with which parents who already make use of corpo-
of cognitive stimulation on parental use of corporal ral punishment apply this tactic.
punishment, and the fact that increased amounts of
cognitive stimulation are associated with decreased
amounts of children’s behavior problems (Grogan-
Implications for Practice and Policy
Kaylor, 2005a, 2005b), it may be beneficial to focus
on interventions that teach parents to increase the Collectively, these findings suggest the need to con-
amount of intellectual stimulation in the home. sider a complexity of factors when attempting to
Interestingly, community-level variables included address the use of corporal punishment by parents.
in the model had no measurable relationship with First, family and children’s advocates wishing to
parental use of corporal punishment. Family neigh- reduce parental use of corporal punishment would
borhood and the region of the country in which the be well advised to direct attention to the role of
family lived were not associated with parental use of broad cultural factors that have an effect on parents’
corporal punishment. Important research has sug- valuation of the appropriateness and effect of corpo-
gested connections between neighborhood condi- ral punishment. Many parents make a conscious
tions and other forms of family violence (Molnar, decision to use corporal punishment on the basis of
Miller, Azrael, & Buka, 2004). There are also peri- beliefs about its appropriateness and effectiveness as
odic suggestions in the literature on parental disci- a disciplinary tactic. If practitioners presume the
pline that corporal punishment may be adaptive in decision is guided by a desire to be a good parent,
some kinds of neighborhoods (Eamon, 2001), then awareness of more effective alternatives may
although these ideas have been further examined provide the necessary information to encourage
and found wanting (Grogan-Kaylor, 2005b). This alternative disciplinary choices. Efforts to teach par-
research finds no evidence that neighborhood condi- ents how to be more effective disciplinarians will not
tions are likely to play a role in the decision to use only acknowledge their desire to be good parents
corporal punishment. but also include attention to other key aspects of the
Finally, the decomposition of the tobit coeffi- parent-child relationship (American Academy of
cients suggests that it is easier to predict the inci- Pediatrics, 1998).
dence of parental use of corporal punishment than Acknowledging the importance of cultural con-
the frequency with which parents employ corporal text and the dynamic interaction between individu-
punishment. The effect sizes of the parameters asso- als, families, and communities is also critical to
ciated with incidence were much larger than those efforts to affect the utilization of corporal punish-
associated with frequency. For example, the coeffi- ment. Individual parenting classes that teach parents
cient for the effect of externalizing behavior prob- to use nonphysical disciplinary tactics may have little
lems on the use of corporal punishment (Table 3) impact in an environment where corporal punish-
was 0.65%. Given that the standard deviation for ment is the norm and ‘‘spare the rod, spoil the
this measure is approximately 15, an increase of child’’ is a guiding principle of parenting. Family
a standard deviation in children’s externalizing and child advocates must attend to these cultural
behavior is associated with approximately a 9.75% differences while also being cognizant of the need
increase in the likelihood of parents using corporal for cultural sensitivity to provide effective program-
punishment. Conversely, the coefficient for the ming and reduce barriers to client engagement
effect of externalizing behavior problems on parental (Kalyanpur & Harry, 1997; Lieberman, 1998).
use of corporal punishment for those parents who Future research should continue to examine the
do employ corporal punishment was 0.018. Apply- links between individual factors and social and cul-
ing the same multiplier of 15 indicates that for tural factors that are associated with the utilization
90 Family Relations  Volume 56, Number 1  January 2007

of corporal punishment. Although the current study Grogan-Kaylor, A., & Otis, M. (2003). The effect of child maltreatment
on adult criminal behavior: A tobit regression analysis. Child Maltreat-
only partially supports the relevance of environmen- ment, 8(2), 129–137.
tal factors, the findings do suggest the utility of an Hashima, P. Y., & Amato, P. R. (1994). Poverty, social support, and paren-
tal behavior. Child Development, 65, 394–403.
ecological framework for guiding future efforts to Hofferth, S., Davis-Kean, P. E., Davis, J., & Finkelstein, J. (1997). The
clarify these relationships. Thus, future research child development supplement to the panel study of income dynamics:
1997 user guide. Retrieved February 15, 2005, from http://psidonline.
should consider the complex array of cultural norms isr.umich.edu/CDS/usergd.html
and values that influence parenting beliefs and Holden, G. W., Miller, P. C., & Harris, S. D. (1999). The instrumental
behaviors to better understand the interrelationship side of capital punishment: Parents’ reported practices and outcome
expectancies. Journal of Marriage and Family, 61, 908–919.
of these factors. Horn, I. B., Cheng, T. L., & Joseph, J. (2004). Discipline in the African
American community: The impact of socioeconomic status on beliefs
and practices. Pediatrics, 115, 1236–1241.
Jacobs, D., & O’Brien, R. M. (1998). The determinants of deadly force: A
structural analysis of police violence. American Journal of Sociology,
References 103, 837.
Kalyanpur, M., & Harry, B. (1997). A posture of reciprocity: A practical
approach to collaboration between professionals and parents of culturally
diverse backgrounds. Journal of Child and Family Studies, 6, 487–509.
Achenbach, T., & Edelbrock, C. (1983). Manual for the child behavior Lieberman, A. F. (1998). Culturally sensitive intervention with children and
checklist and revised child behavior profile. Burlington: University of Ver- families. Newsletter of the Infant Health Promotion Project, 22, 1–5.
mont, Department of Psychology. Luster, T. & Kain, E. L. (1987). The relation between family context and
American Academy of Pediatrics. (1998). Guidance to effective discipline. perceptions of parental efficacy. Early Childhood Development and Care,
Pediatrics, 101, 723–728. 29(3), 301–311.
Bradley, R. H., Caldwell, B. M., & Elardo, R. (1979). Home environment Lytton, H., & Romney, D. M. (1991). Parents’ differential socialization of
and cognition development in the first two years: A cross-lagged panel boys and girls: A meta-analysis. Psychological Bulletin, 109, 267–296.
analysis. Developmental Psychology, 15, 246–250. MacMillan, H. L., Boyle, M. H., Wong, M. Y., Duku, E. K., Fleming,
Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, J. E., & Walsh, C. A. (1999). Slapping and spanking in childhood and
MA: Harvard University Press. its association with lifetime prevalence of psychiatric disorders in a gen-
Bronfenbrenner, U. (1986). Ecology of the family as a context for human eral population sample. Canadian Medical Association Journal, 161,
development: Research perspectives. Development Psychology, 22, 723–742. 805–810.
Center for Human Resource Research. (2000). NLSY79: 1998 child and Massey, D. S. (1990). American apartheid: Segregation and the making of
young adult data user’s guide (The Ohio State University). Columbus: the underclass. The American Journal of Sociology, 96(2), 329–357.
The Ohio State University, Center for Human Resource Research. Massey, D. S. (1994). America’s apartheid and the urban underclass. The
Center for Human Resource Research. (2001). NLSY79 user’s guide: A guide Social Service Review, 68(4), 471–478.
to the 1979–2000 National Longitudinal Survey of Youth data. Colum- McDonald, J. F., & Moffit, R. A. (1980). The uses of tobit analysis. Review
bus: The Ohio State University, Center for Human Resource Research. of Economics and Statistics, 62, 318–321.
Day, R. D., Peterson, G. W., & McCracken, C. (1998). Predicting spank- McLoyd, V. C. (1990). The impact of economic hardship on Black families
ing of younger and older children by mothers and fathers. Marriage and and children: Psychological distress, parenting, and socioemotional
Family, 60(1), 79–94. development. Child Development 61, 311–346.
Dietz, T. L. (2000). Disciplining children: Characteristics associated with the McLoyd, V. C., & Smith, J. (2002). Physical discipline and behavior prob-
use of corporal punishment. Child Abuse & Neglect, 24, 1529–1542. lems in African American, European American, and Hispanic children:
Eamon, M. K. (2001). Poverty, parenting, peer and neighborhood influen- Emotional support as a moderator. Journal of Marriage and Family,
ces on young adolescent antisocial behavior. Journal of Social Service 64(1), 40–53.
Research, 28(1), 1–23. Molnar, B. E., Miller, M. J., Azrael, D., & Buka, S. L. (2004). Neighbor-
Ellison, C. G., Bartkowski, J. P., & Segal, M. L. (1996). Conservative Prot- hood predictors of concealed firearm carrying among children and ado-
estantism and the parental use of corporal punishment. Social Forces, lescents: Results from the project on human development in Chicago
74, 1003–1028. neighborhoods. Archives of Pediatric and Adolescent Medicine, 158,
Frick, P. J., Christian, R. E., & Wootton, J. M. (1999). Age trends in the 657–664.
association between parenting practices and conduct problems. Behav- Nobes, G., Smith, M., Upton, P., & Heverin, A. (1999). Physical punish-
ior Modification, 23(1), 106–128. ment by mothers and fathers in British homes. Journal of Interpersonal
Garbarino, J. (1977). The human ecology of child maltreatment: A concep- Violence, 14, 887–902.
tual model for research. Journal of Marriage and the Family, 39, 721–732. Roncek, D. W. (1992). Learning more from tobit coefficients: Extending
Gershoff, E. T. (2002). Corporal punishment by parents and associated a comparative analysis of political protest. American Sociological Review,
child behaviors and experiences: A meta-analytic and theoretical review. 57, 503.
Psychological Bulletin, 128, 539–579. Sariola, H., & Uutela, A. (1992). The prevalence and context of family vio-
Giles-Sims, J., Straus, M. A., & Sugarman, D. B. (1995). Child, maternal, lence against children in Finland. Child Abuse & Neglect, 16, 823–832.
and family characteristics associated with spanking. Family Relations, Simons R. L., Johnson, C., & Conger, R. D. (1994). Harsh corporal punish-
44(2), 70–176. ment versus quality of parental involvement as an explanation of adoles-
Graziano, A. M., Hamblen, J., & Plante, W. A. (1996). Subabusive vio- cent maladjustment. Journal of Marriage and the Family, 56, 591–607.
lence in child-rearing in middle class American families. Pediatrics, 98, Socolar, R. R., & Stein, R. E. (1995). Spanking infants and toddlers:
845–848. Maternal belief and practice. Pediatrics, 95(1), 105–111.
Greene, W. H. (1997). Econometric analysis (3rd ed.). Upper Saddle River, Straus, M. A. (1983). Ordinary violence, child abuse, and wife beating:
NJ: Prentice Hall. What do they have in common? In D. Finkelhor, R. Gelles, G. Hotal-
Grogan-Kaylor, A. (2004). The effect of corporal punishment on antisocial ing, & M. Straus (Eds.), The dark side of families: Current family violence
behavior in children. Social Work Research, 28(3), 153–164. research (pp. 213–234). Beverly Hills, CA: Sage.
Grogan-Kaylor, A. (2005a). Corporal punishment and the growth trajectory Straus, M. A. (2000). Beating the devil out of them: Corporal punishment in
of children’s antisocial behavior. Child Maltreatment, 10, 283–292. American families and its effects on children. New Brunswick, NJ: Trans-
Grogan-Kaylor, A. (2005b). Relationship of corporal punishment and anti- action Publications.
social behavior by neighborhood. Archives of Pediatrics & Adolescent Straus, M. A., & Donnelly, A. D. (1993). Corporal punishment of adoles-
Medicine, 159, 938–942. cents by American parents. Youth and Society, 24, 419–420.
The Predictors of Parental Use of Corporal Punishment  Grogan-Kaylor and Otis 91

Straus, M. A., & Gelles, R. J. (1986). Societal change and change in family and occupational class. In M. Straus & R. Gelles (Eds.), Physical vio-
violence from 1975 to 1985 as revealed in two national surveys. Journal lence in American families: Risk factors and adaptations to violence
of Marriage and the Family, 48, 465–479. in 8,145 families (pp. 133–148). New Brunswick, NJ: Transaction
Straus, M. A., & Stewart, J. H. (1999). Corporal punishment by American Publications.
parents: National data on prevalence, chronicity, severity, and duration, Wolfner, G. D., & Gelles, R. J. (1993). A profile of violence toward chil-
in relation to child, and family characteristics. Clinical Child and Family dren: A national study. Child Abuse & Neglect, 17, 197–212.
Psychology Review, 2(2), 55–70. Woolley, M. E., & Grogan-Kaylor, A. (2006). Protective family factors in
Tobin, J. (1958). Estimation of relationships for limited dependent varia- the context of neighborhood: Promoting positive school outcomes.
bles. Econometrica, 26(1), 24–36. Family Relations, 55(1), 93–104.
Wauchope, B., & Straus, M. A. (1990). Physical punishment and Zill, N., & Peterson, J. (1986). Behavior problems index. Washington, DC:
physical abuse of American children: Incidence rates by age, gender, Child Trends.

Você também pode gostar