Introduction

There is a tremendous amount of overlap between types of aggressive and violent behavior within families. Child maltreatment is more common in homes where intimate partner aggression and violence (IPAV) occur.1 – 4 And data are evolving to suggest that corporal punishment (CP), a strong risk factor for child maltreatment,5 , 6 co-occurs with IPAV as well.7 , 8 The link between CP and IPAV also is intergenerational as experiencing CP in childhood raises risk for subsequent aggression5 , 9 and both CP and conduct disorder increase risk for later IPAV perpetration.10 , 11 Hence, reducing the use of CP might assist in curbing the cycle of family and community violence.

US prevalence estimates of CP use are high, ranging anywhere from 35% to 90% depending upon moderating factors such as the age or gender of the child and the type of punishment specified.12 Approval of CP use also is high with nearly three quarters of US adults thinking it is okay and sometimes necessary to spank a child;13 , 14 however, such findings tend to vary demographically with approval being highest in the South15 and among Blacks,16 Conservative Protestants,17 persons with lower socioeconomic status (SES) and education18, and persons who experienced CP as a child.19 , 20 Such a link, however, has been more equivocal among adults that were physically or psychologically abused as children.16 , 19 , 21

Primary prevention of violence requires a careful assessment of potential root causes that are both significant and malleable. Having a positive attitude toward the use of CP is consistently one of the strongest predictors of CP use22 – 27 and, based on declines of CP approval in other countries, it is clear that such attitudes are malleable.28 , 29 Yet, little attention has been paid to understanding specific modifiable predictors of these attitudes, particularly those set in the social environment.

The Theory of Planned Behavior (TPB)30 suggests that perceived social norms regarding CP use might be such a predictor. Social norms can be injunctive (indicating approval of a behavior, or what persons ought to do) or descriptive (indicating prevalence of a behavior, or what most persons actually do).31 Empirical studies focused on behaviors such as speeding,32 recycling,31 smoking,33 and classroom aggression34 suggest that both types of perceived norms might have roles in shaping beliefs and behaviors. Importantly, such subjective assessments of the social environment might be just as or even more important than objective assessments in shaping parenting.35 Thus, while objective, community-level measures of the social environment have been linked with harsh or abusive parenting (e.g., neighborhood poverty,36 concentrated disadvantage, and murder rate statistics),37 subjective measures of related social contexts, such as parents’ perceptions of neighborhood danger, lack of adequate public services,36 or their normative environment regarding CP, might also be important in shaping parenting attitudes and choices. The latter will be the focus of the present study.

TPB and prior evidence also suggest that parents’ expectations for outcomes of CP use and for their children are linked with attitudes toward and use of CP. Persons less inclined to think that CP could lead to physical harm are likely to have more positive attitudes toward CP use.19 And mothers that expect CP to result in positive child outcomes, such as immediate compliance and long-term learning, but not in negative child outcomes, such as child distress, spank more frequently.38 Unfortunately, many adults have expectations for their children that are inappropriate to the child’s age,39 and this can impact parents’ disciplinary choices. For example, one study found that parental attributions about their children’s cognitive and behavioral competence and responsibility for negative actions were associated with parents having more approval of power-assertive child discipline and less approval of reasoning and explaining.40 Thus, parents that expect positive outcomes for CP and that have unrealistically high expectations for their children might be inclined toward approval and use of CP.

The current study aims to add to the literature by using the TPB and a social ecological framework to uncover modifiable predictors of parents’ positive attitudes toward CP, with a particular focus on examining the importance of perceived social norms regarding CP. A social ecological framework41 suggests that multiple levels of the social environment are likely to influence parenting and childhood risk.42 Prior research on social norms and CP has focused on the potential modifying effects of descriptive norms regarding CP43 , 44 and perceived neighborhood social cohesion45 on the link between CP use or harsh parenting and poor outcomes, such as aggression, in children. The present study, however, examines parents’ perceived injunctive and descriptive norms regarding CP use as possible predictors of parents’ own attitudes toward CP use. In particular, we asked parents about 2 types of injunctive norms regarding CP: those that they perceived from professionals they sought advice from about child discipline, and those that they perceived from close family and friends. We also asked about the descriptive norms they perceived about CP within their social network of parents.

The primary aim of this study then was to examine associations between these perceived injunctive and descriptive norms regarding CP use, along with parental outcome expectancies of CP use and knowledge of child development, and having positive attitudes toward CP. As a secondary aim, we conducted these analyses controlling for a multitude of important demographic constructs known to be linked with attitudes toward and use of CP, including family structure; SES; religiosity; aggression in the family of origin; and child characteristics, particularly age and gender. This survey was conducted among an urban, population-based sample of parents with the intent of informing community interventions designed to shift social contexts associated with approval and use of CP.

Methods

Study Sample

A stratified random-digit-dial (RDD) telephone survey was conducted in a Southern metropolitan area with a population just under 400,000. Potential participants were considered eligible for the study if they were at least 18 years of age, the parent or legal guardian of a child in their household under the age of 16, fluent in English, and self-identified as Black or White. Although other racial and ethnic groups are present in the city, their numbers would have been too small to analyze statistically. If a person was eligible and interested in participating, the interviewer either proceeded immediately with the interview or scheduled a call back for a later time. All questions referred to an Index Child, identified as the parent’s child closest to age 4. Interviews took about 25 minutes to complete and were conducted between December 2008 and February 2009. The Tulane University Institutional Review Board approved the study procedure.

The response rate was 33.4%. Sample demographics are presented in Table 1. The sample (n = 500) was stratified by parent gender and race in order to prevent over-representation of particular groups. The race strata were designed to reflect the city as a whole, thus a Black to White quota ratio of 60:40 was established. Census data showed that the adult female-to-male ratio in the city was about 53:47.46 However, data also showed that 91% of families with children in the area contained a female adult while only 51% contained a male adult. Therefore, in order to assure that female parents were adequately represented, a 70:30 female to male quota ratio was established.

Table 1 Sample demographics
Full size table

Dependent Variable

Positive Attitudes toward CP Use

Items from the Attitudes Toward Spanking (ATS) questionnaire47 were used to measure each parent’s personal attitudes toward CP. The 4 items used for this study (“Spanking is a normal part of my parenting,” “Sometimes the only way to get my child to behave is with a spank,” “When all is said and done, spanking is harmful for my child,” and “Overall, I believe spanking is a bad disciplinary technique”) were rated on a 5-point Likert scale (1 = strongly agree, 5 = strongly disagree). Item coding was adjusted for analysis so that a higher score always meant having more positive attitudes toward CP. Item scores were summed and then averaged so that final scores ranged from 1 to 5 (α = 0.79).

Predictor Variables

The primary aim of our study was to examine associations between the following predictors and the dependent variable.

Perceived Norms

Perceived Injunctive Norms of Professionals Regarding CP Use

First, parents were asked, “When it comes to seeking advice from a professional source about how best to discipline your child, are you more likely to seek advice from …(1) your child’s doctor, (2) a religious leader such as your pastor, minister, or rabbi, or (3) some other professional.” Parents’ most common responses were child’s doctor, religious leader, or mental health professional.48 Next, the parents were asked, “Do you think your [response to prior question] would strongly agree, agree, neither agree nor disagree, disagree, or strongly disagree with each of the following statements about spanking?” The same ATS items were used and adjusted slightly to be about spanking in general (e.g., “Spanking is a normal part of parenting” and “When all is said and done, spanking is harmful for children”). Item coding was adjusted for analysis so that a higher score meant perceiving an injunctive norm indicating the professional’s approval of CP use. Item scores were summed and then averaged so that final scores ranged from 1 to 5 (α = 0.82).

Perceived Injunctive Norms of Close Family and Friends Regarding CP Use

Parents were asked: “Now I would like you to think about your friends and family members whose opinions mean the most to you. Do you think that most of these friends and family members would strongly agree, agree, neither agree nor disagree, disagree, or strongly disagree with each of the following statements about spanking?” This was followed by the same adjusted ATS items used for the prior question. Item coding was adjusted for analysis so that a higher score meant perceiving an injunctive norm indicating family and friends’ approval of CP use. Item scores were summed and then averaged so that final scores ranged from 1 to 5 (α = 0.83).

Perceived Descriptive Norms

Three questions measured parents’ perceived prevalence of CP use within their social network of fellow parents. First, participants indicated their agreement or disagreement with the following statements using a 5-point Likert scale (1 = strongly agree, 5 = strongly disagree): “Most parents who are important to me do not use spanking or swatting as a regular way to discipline their child,” and “The parents of children whose opinions mean the most to me spank or swat their children at least once per month.” Next, participants were asked to think about other parents they know with children about the same age as their Index Child, and then asked how often they thought these parents used CP. Response options were: almost every day, often, sometimes, seldom, or never (1–5). Item coding was adjusted for analysis so that a higher score meant perceiving a descriptive norm indicating common use of CP among family and friends. Item scores were summed and then averaged so that final scores ranged from 1 to 5 (α = 0.68).

Parental Expectations and Knowledge

Expected Outcomes of Using CP

Eight items measured parents’ expectations for using CP. The question stem and 6 of the 8 items came directly from the Outcomes of Physical Punishment Scale.49 Parents were asked, “How often do you think that physical discipline, such as spanking, of a child leads to: (1) obedience of parents, (2) respect for parents, (3) physical injury to the child, (4) long-term emotional upset in the child, (5) learning of acceptable behavior, and (6) increased child aggression.” Two additional items were added to this list: (7) healthy family relationships later in life, and (8) a better sense of self-control. Participants rated each item on a 5-point Likert scale (1 = never, 5 = always). The scale was divided into 2 subscales, one consisting of the 5 items (1, 2, 5, 7, and 8) indicating positive expectations for CP use (α = 0.80), and the other consisting of the 3 items (3, 4, and 6) indicating negative expectations for CP use (α = 0.84). Item scores for each subscale were summed and then averaged so that final scores ranged from 1 to 5.

Expectations for and Knowledge of Child Development

Nine items were used from a questionnaire developed by Reich.50 Seven of these items assessed knowledge regarding certain infant safety issues (e.g., “Once a baby turns 6 months, the car seat should be front-facing,” “Hard foods like popcorn or carrots are dangerous for babies,” “Babies are safer if they sleep on their backs”). Two items addressed cognitive development (e.g., “Most 18-month old children are able to sit quietly at a dinner table for an hour while everyone else eats,” “A one-year-old child will sometimes do things that are mean on purpose”). Items were scored as correct or incorrect; the scale ranged from 0% to 100%, with a higher score indicating greater knowledge and appropriate expectations for child development (mean score = 72%).

Control Variables

A secondary aim of our study was to control for 5 key demographic constructs known to be associated with attitudes toward and use of CP. The 5 demographic constructs, followed by the variables we assessed to represent each of them, included: (1) family structure: marital status, current living situation, number of children in household, and primary disciplinarian; (2) SES: education, household money to meet needs, and employment status (Although income was measured, this data point was missing for 8.2% of respondents); (3) religiosity: religion, attendance at religious services, and importance of religion in daily life; (4) aggression in the family of origin: witnessed IPAV, experienced verbal or physical (non-spanking) aggression, and spanked in childhood; and (5) index child variables: gender, age, and relationship to respondent.

Data Analysis

Pairwise correlations of the 6 main predictor variables were conducted and significance levels were Bonferroni-adjusted (Table 2). One-way ANOVAs were conducted for all demographics with “attitudes toward CP” (Table 3). Six simple (bivariate) ordinary least squares (OLS) models regressing “attitudes toward CP” separately on each of the 6 main predictor variables were conducted (Table 4).

Table 2 Correlations matrix of main predictors: perceived norms and parental expectations and knowledge
Full size table
Table 3 Demographic variations in positive attitudes toward CP use based on 1-way ANOVAs
Full size table
Table 4 Simple ordinary least squares (OLS) models regressing positive attitudes toward CP use on parents’ perceived norms, expectations, and knowledge
Full size table

In order to achieve the most parsimonious final regression models, and because we had a number of control variables addressing the same constructs, we chose to remove redundant control variables empirically. To do this, we ran 5 preliminary multivariable OLS models regressing “attitudes toward CP” on each of the 5 demographic control construct blocks listed above. Each regression was followed by omnibus tests; the following 4 variables remained statistically significant (p < 0.05) in each block, and so were included in the final regression models: education and household money (block 2), importance of religion in daily life (block 3), and spanked in childhood (block 4). (No variables from blocks 1 or 5 remained significant.)

Two final multivariable OLS models were run (Table 5). Model 1 regressed “positive attitudes toward CP use” on the four demographic variables found significant in the aforementioned block analyses, along with participant gender and race; the latter 2 variables were included due to the sampling procedures used. Model 2 was the same as model 1, with the addition of the 6 predictors of interest—perceived norms (3 variables) and parental expectations and knowledge (3 variables). Multicollinearity was within acceptable limits (mean variance inflation factor = 1.51, highest = 2.04).

Table 5 Multivariate ordinary least squares (OLS) models regressing positive attitudes toward CP use on demographics and parents’ perceived norms, expectations, and knowledge
Full size table

Results

Frequency distributions for all demographics are presented in Table 1. Most of the predictors of interest were strongly correlated (Table 2). The 3 types of perceived norms were the most highly correlated variables (r = 0.57–0.60). Expecting positive outcomes from CP use was also highly correlated with all 3 types of norms (r = 0.46–0.50). Expecting negative outcomes from CP use was negatively correlated with all the 3 types of norms (r = −0.29 to –0.43) and with expecting positive outcomes (r = −0.40). Knowledge of child development was negatively associated with perceiving injunctive norms of professionals (r = −0.22) and descriptive norms (r = −0.16).

Table 3 shows how positive attitudes toward CP varied by demographic characteristics; only those significant at p < 0.05 are included in this table. Four types of demographic variables seemed to matter most: race, SES indicators, family history of being spanked, and religion/religiosity. In particular, Blacks and those with less education, without enough money to meet needs, and who were spanked more frequently as children all had more positive attitudes toward CP than their counterparts. Additionally, Christians (non-Catholics), those with more frequent attendance at religious services, and those who stated that religion was “very important” in daily life also were associated with more positive attitudes toward CP.

Table 4 shows simple, bivariate linear (OLS) regression results for associations between each of the main predictors of interest and having positive attitudes toward CP. Perceived approval of CP by professionals, family and friends, perceived descriptive norms of CP, and positive outcome expectancies for CP all were associated positively with positive attitudes toward CP. Expecting negative outcomes from CP and having greater knowledge of child development were associated negatively with positive attitudes toward CP.

Table 5 shows results from the 2 multivariate linear regression models predicting positive attitudes toward use of CP. Model 1 shows that only 3 demographic control variables independently predicted having positive attitudes toward CP: race (Black), religion being very important, and having been spanked in childhood. Once the predictor variables of interest were added (Model 2), only the latter variable remained significant.

Model 2 shows that 5 of the 6 predictors of interest (all 3 types of perceived norms and both positive and negative expected outcomes) remained strongly associated with having positive attitudes toward CP use, even after adjusting for key demographics and the other main predictors. In this final model, the perceived norms were the strongest predictors of having positive attitudes toward CP in the following order: (1) perceived approval of CP by professionals (β = 0.30), (2) perceived descriptive norms of CP use (β = 0.22), and (3) perceived approval of CP by family and friends (β = 0.19). Both positive (β = 0.13) and negative (β = −0.13) expected outcomes for CP use also were strong predictors of attitudes.

Discussion

Perceived injunctive and descriptive norms regarding CP, as well as expected outcomes of CP use, were strong predictors of parents’ attitudes toward CP use. The strongest predictor of parents’ positive attitudes toward CP use was their perception that the professional they were most likely to turn to for advice about child discipline approved of CP. Within this sample, we reported elsewhere48 that 48% of these parents selected their pediatrician as the professional they were most likely to turn to for such advice; next most common were religious leaders (21%) and mental health professionals (18%). Similar results were found in a mostly White sample from Minnesota: spankers were more likely to perceive that their sources of advice about discipline recommend the use of CP, whereas non-spankers perceived that their sources opposed its use.51

Unfortunately, it might be particularly common for pediatricians to provide ineffective responses (e.g., ignoring, denying, or doubting the problem, or being too authoritative or seeming helpless in addressing the problem) to parents who raise the issue of CP,52 suggesting a need for better training on this topic in pediatric residency programs. A population-based study of primary care doctors in Ohio found that 59% of pediatricians and 70% of family practitioners supported the use of CP.53 In a smaller convenience sample of academic pediatricians specializing in child abuse, only 39% thought that CP was sometimes appropriate, but only 29% taught their residents how to handle situations when parents used CP during a medical visit.54 Both of these studies are more than a decade old, however, and it is likely that physicians’ attitudes toward CP have changed to some extent in that time. Despite the fact that the American Academy of Pediatrics does not endorse spanking55 and has summarized methods to guide physicians’ discussions with patients,56 substantially more work is needed to understand pediatricians’ attitudes and practices regarding this issue.

Religious leaders and mental health professionals are also important to consider given their potential normative influence on parents’ attitudes about CP. A recent nationwide survey of Presbyterian clergy indicated that CP was not wholly endorsed by most clergy. Clergy that were male, non-White, of lower SES, Republican, and had conservative political leanings had an increased likelihood of supporting CP.57 Regarding mental health professionals, a survey of 300 child clinical psychologists published in 2000 found that 70% would never endorse spanking and 30% would do so only in rare circumstances; just over half (52%) thought that the use of spanking was “ethical” in rare circumstances.58

The normative influence of family and friends also was an important predictor of CP attitudes. Given the widespread approval of CP in the United States, this finding suggests a need for universal efforts to educate the general public about the risks associated with CP use and the importance of providing children with positive, non-physical discipline to correct behavior. Our study also mirrors the findings of many other studies that have shown that certain segments of the population are particularly prone to having positive attitudes toward CP, namely Blacks, those of lower SES, persons spanked as children, and those who attach high importance to religion; on the latter point, a recent study demonstrated that having higher ratings in extrinsic religiosity and biblical literalism was associated with increased child abuse potential.59 Targeted efforts might be needed to change beliefs and attitudes supportive of CP within such communities.

Finally, the association between expected outcomes of CP use and attitudes is particularly promising as it indicates a possible mechanism for shifting attitudes among both parents and those who influence them. Other studies have reported similar findings. Professionals less likely to endorse the use of CP include psychologists who believed that the research literature shows negative outcomes for CP,58 as well as clergy who knew someone in their congregation who had experienced abuse.57 Moreover, adults who do not anticipate negative physical consequences from spanking are more likely to endorse it.19 Hence, intervention efforts that focus on shifting knowledge and outcome expectations regarding CP, targeting both parents and key influencers, might be a promising approach for shifting attitudes about CP.

Limitations

This study has some important limitations. First, the survey was cross-sectional and therefore temporal relationships between variables cannot be established. Second, the response rate was low (33.4%), which is not unusual for non-incentivized RDD telephone surveys.60 Given the mean income ($66,000) and the high education level for this sample, it is likely that lower SES parents were under-represented. Also, findings might not be generalizable; however, similar findings have emerged from prior research with very different samples.38 , 51 Third, the “knowledge of child development” scale was not ideal for this study as many of the items referred to knowledge about infant development and safety, yet the average age of the index child in this study was 7 years old. This might explain the weak correlation of this variable with other predictors and the variable’s non-significance in the final model, rather than the construct’s true lack of importance in this model.

Conclusions and Implications

The most common community-level efforts to date to shift attitudes and reduce CP use have involved national or schoolwide bans on CP use.61 Given the strong link between perceived injunctive norms (of professionals/religious leaders and close family/friends) and descriptive norms of fellow parents, our findings suggest that efforts targeting multiple and specific social contexts could assist in changing attitudes about CP. In particular, targeted efforts are needed to both assess and shift the attitudes and practices of professionals who influence parents, especially pediatricians, mental health professionals, and religious leaders. The social context of “professional provider visits,” especially with the medical community, is particularly important to many young parents. Fortunately, some efforts already have been made with pediatricians to improve their ability to provide effective counseling regarding violence prevention and positive parenting, albeit with mixed or untested results.62 – 64 The social context and support of religious communities and leaders also is very important to many parents. Although religious organizations are recognized globally as potential partners for promoting social change and justice 65 , 66 as well as family and community health in urban and high-risk populations,(e.g., 67 , 68 ) targeted efforts are needed to engage religious leaders around the issue of parenting education and child discipline specifically. Finally, universal- or population-based efforts, such as public education campaigns, have been utilized successfully in public health efforts regarding issues such as alcohol policy,69 tobacco use,70 and getting help for abused children.71 Similar efforts are needed to educate parents and the general public about the high risk/benefit ratio for using CP and the effectiveness of non-physical forms of child discipline.