Factors Associated With Family Member’s Spanking of 3.5-year-old Children in Japan

Background Spanking can cause adverse psychological development and biological functional changes in children. However, spanking is widely used by parents in Japan. This study explored the risk factors for family member’s spanking of 3.5-year-old children using nationwide population data in Japan. Methods Surveys were administered to family members in Japan who had a child in 2001 (first cohort) or in 2010 (second cohort), and the data when their child was 0.5, 1.5, and 3.5 years old were used. We used multivariate binary and ordinal logistic regression analyses to examine the associations between risk factors and spanking children at 3.5 years of age, which was subcategorized into frequencies of never, sometimes, and always spanking, presented with odds ratios (ORs) and 95% confidence intervals (CIs). Results Among 70,450 families, 62.8% and 7.9% sometimes and always spanked their children, respectively. Children in the second cohort were spanked less frequently compared with those in the first cohort, and fathers who responded to the questionnaire spanked children less frequently than mothers who responded. Identified associated factors for spanking were male child, presence of siblings of the child, not living in a two-parent household, not living in a three-generation household, younger parents, parents with lower education, no outside work or unstable work, and lower family income. Conclusions We found a high prevalence of spanking and its associated factors. Approaching those with lower socioeconomic factors and promoting fathers’ involvement in parenting may be important public health strategies for reducing and preventing spanking.


INTRODUCTION
Spanking is the most common form of corporal punishment. 1 More than 50 countries worldwide have banned corporal punishment because, regardless of its severity, it is known to be associated with physical abuse, 2 adverse psychological development, and biological changes in neural functioning. [2][3][4][5] Previous studies have shown behavioral problems among spanked children, including external aggression or antisocial behavior, and impaired mental health problems, such as suicide. 4,[6][7][8][9][10] The United Nations enacted the Convention on the Rights of the Child to protect children from corporal punishment and other cruel or degrading forms of punishment in 1989. 11 However, corporal punishment remains a relatively common disciplinary measure in Japan and other countries worldwide. 12 According to a recent survey among Japanese, approximately 70% of parents experienced being spanked as a child and 60% admitted to spanking their own children. 13 This reflects the fact that corporal punishment by parents had not been banned, whereas corporal punishment by school teachers has been banned in Japan since 1947 under the School Education Act (Act No. 26). 14 Legislative approaches are reported to be a promising strategy in order to reduce physical abuse of children in other countries. 15 In June 2019, the Japanese Diet enacted amendments to the Child Welfare Act (Act No. 164 of 1947) 16 and the Act on the Prevention, etc of Child Abuse (Act No. 82 of 2000), 17 including a ban on corporal punishment of children by parents and other guardians, which will go into effect in April 2020. 18 Therefore, it is of the utmost importance to provide scientific evidence on the risk factors of spanking and corporal punishment among the Japanese population.
Identified risk factors for spanking and corporal punishment consist of parent, child, family, and community=cultural factors. 1,2,9 Reported parental factors from other countries are as follows: very young mothers, lower socioeconomic status (SES), poor maternal physical and mental health, personal experience of physical discipline or abuse, and elevated parenting stress. 1,10,[19][20][21][22] However, as far as we know, there has been no large epidemiological study in Japan that broadly examined the associations between risk factors and parents' spanking. Moreover, there have been no studies to observe the transition in use of parents' spanking in different generations in countries where corporal punishment has not been banned. Therefore, the purpose of this study was to explore the parental risk factors of spanking of 3.5-year-old children using nationwide population data in Japan.

Study participants
Study data were obtained from a Japanese panel survey entitled, "The Longitudinal Survey on Babies Born in the 21 st Century". Based on vital statistics birth record lists, the study sample of the first-generation cohort included all participants whose children were born in Japan between January 10 and 17, 2001 or between July 10 and 17, 2001 (n = 53,575). The second-generation cohort included all participants whose children were born in Japan between May 10 and 24, 2010 (n = 43,767). These selected participants were recruited via mail questionnaires sent to the children's residence when the children were 0.5 years old, which corresponded to the first-wave panel survey. Respondents were considered to have agreed to participate in the study if the questionnaire was returned to the Ministry of Health, Labour and Welfare. There were 47,015 responses (response rate of 87.8%) for the first cohort and 38,554 responses (response rate of 88.1%) for the second cohort, for a total sample of 85,569 in the firstwave panel survey. The surveys were conducted annually and included approximately 20 questions. The variables used in this study were retrieved from data gathered from the first-(0.5 years old), second-(1.5 years old), and fourth-(3.5 years old) wave panel surveys in each cohort. Participants who had not provided information on spanking at the fourth wave (ie, when their children were 3.5 years old) survey were excluded (n = 15,119), which led to a final sample size of 70,450 for this study.

Outcome
Family members were asked in the fourth-wave survey (in 2003-2004 for the first cohort, and in 2013 for the second cohort), when their children were 3.5 years old, "How do you generally react to your child's misbehavior?". The response choices were as follows: 1) explaining why the behavior was wrong, 2) saying "No" without any explanation, 3) spanking, 4) allowing them to recognize their misbehavior by ignoring them, and 5) making them go outside or putting them in a closet. Family members were required to answer each item by selecting the frequency from the three choices of always, sometimes, and never. Spanking was used as the outcome for this study.

Exposure
Respondents in the fourth-wave survey were asked to report their relationship to the child (ie, mother, father, maternal grandmother, maternal grandfather, paternal grandmother, paternal grandfather, or other). Multiple responses were allowed. We recategorized the respondent as mother, father, both parents, and other family members in this study. Other family members were those who did not include either parent as a respondent. For example, if mother and maternal grandmother were chosen, it was re-categorized as "mother", whereas if only the maternal grandmother was chosen, it was re-categorized as "other family members". Caregivers were defined using the question "Who usually takes care of the child?", with the following choices of answers: mother, father, maternal grandmother, maternal grandfather, paternal grandmother, paternal grandfather, nursery teachers, babysitters, kindergarten teachers, and others. Multiple answers were allowed. We re-categorized the responses as mother, father, both parents, and others in this study. Others were those who did not include either parent as a caregiver. For example, if mother and nursery teachers were chosen, it was recategorized as "mother", whereas if only nursery teachers were chosen, it was re-categorized as "others". Information about potential risk factors for spanking was also collected, including the gender of the child, family structure (presence of siblings of the child, living in a two-parent household, and living in a three-generation household), and family socioeconomic factors (parents' age, level of education, working hours, work type, and family income). Information on the parents' level of education was obtained from the second-wave survey, and family income, presence of siblings of the child, living in a two-parent household, living in a three-generation household, and parents' working hours and work types were obtained from the fourth-wave survey.
Gender of the child was categorized as either boy or girl. The presence of siblings of the child was dichotomously recategorized. For the variable "living in a two-parent household", a parent was defined as such irrespective of whether it was the biological parent or a step-parent. Living in a three-generation household was defined as child living with either the mother and=or father and one or more grandparents.
Parents' age was calculated according to the parents' birth date information obtained from the first-wave survey, and categorized into six age groups. Parent's level of education was obtained by asking their highest level of education and was categorized as junior high school, high school, vocational school, junior college, university and higher, and others. Parents' working hours per week was categorized as 0 hours, 1-19 hours, 20-39 hours, and 40 hours and over. Parents' work type was categorized as no outside work (ie, housewife), seeking employment, student, employed full-time, employed part-time, self-employed, domestic side job, and others. For the variables where parents' characteristics were provided and the respondents were "both parents", the mother's variables were used because mothers comprised more than 90% of the participants in this study.
Family income was calculated by summing the mother's and father's incomes during the last year as obtained from the fourthwave survey. If the income for either parent was missing, the other parent's income was considered the family income. Family income was re-categorized into quartiles. Responses of always using forms of discipline other than spanking were used as a covariate.
We obtained permission to use the panel survey data from the Ministry of Health, Labour and Welfare under the Statistics Act in Japan (No. 1020-3). This study was approved by the Ethics Review Board of Osaka University Hospital (No. 16154). generation household, parents' age, parents' level of education, parents' working hours, parents' work type, family income, and cohort generation. The statistical significance of the differences in these proportions was analyzed using the χ 2 test.
Binary logistic regression analyses were used to examine the associations between potential risk factors and spanking. We estimated the risk of sometimes and always spanking, respectively, by obtaining odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) in comparison to never spanking. We also used ordinal logistic regression analyses to estimate the association between potential risk factors and cumulated risk of sometimes or always spanking in an ordered manner. We further applied ordinal logistic regression analyses, stratified by cohort generation and the gender of the child, respectively. All analyses were performed using Statistical Analysis Software version 9.4 (SAS Institute, Inc., Cary, NC, USA).

RESULTS
Of the total 70,450 respondents, 88.5% (n = 62,349) were mothers, 5.9% (n = 4,140) were fathers, 4.7% (n = 3,290) were both parents, and 1.0% (n = 671) were other family members. A total of 45.3% (n = 31,894) of caregivers were mothers, 0.2% (n = 174) were fathers, 46.8% (n = 32,972) were both parents, and 7.7% (n = 5,410) were others (Table 1). Most respondents in the category of other family members were maternal or paternal grandmothers, and most caregivers in the category others were nursery or kindergarten teachers (data not shown). Distributions of spanking and potential risk factors are shown in Table 2. Among all participants, 62.8% sometimes spanked, and 7.9% always spanked to discipline the child at age 3.5 years old. The proportions of always spanking were higher in the following categories: the first cohort; child gender of boy; presence of siblings of the child; not living in a two-parent household or living in a three-generation household; younger parents; parents with a lower level of education or shorter working hours; mother having no outside work, seeking employment, or having a domestic side job; father seeking employment; and lower family income. Table 3 shows the associations between each risk factor and spanking in the binary and ordinal logistic regressions. In terms of respondents, compared with the mother, the father or both parents had lower risk, while others had higher risk of spanking in the ordinal logistic regression model: adjusted ORs were 0.85 (95% CI, 0.79-0.92), 0.92 (95% CI, 0.86-0.99), and 1.34 (95% CI, 1.13-1.59) respectively. Compared with the first cohort, the second cohort showed lower odds for spanking: the adjusted ORs in the binary logistic regression model were 0.67 (95% CI, 0.65-0.70) for sometimes spanking and 0.58 (95% CI, 0.54-0.62) for always spanking, and that obtained from the ordinal logistic regression analyses was 0.54 (95% CI, 0.52-0.56). The gender of the child was associated with spanking: the adjusted ORs for spanking boys in the binary logistic regression were 1.25 (95% CI, 1.21-1.29) for sometimes spanking and 1.45 (95% CI, 1.37-1.54) for always spanking, and the adjusted OR for spanking boys was 1.48 (95% CI, 1.43-1.52) in the ordinal logistic regression model. For family structure factors, the presence of a sibling of the child and not living with both parents were associated with spanking, while living within a three-generation household was inversely associated with spanking. For family socioeconomic factors, dose-response inverse associations of parents' age, parents' level of education, and family income with spanking were observed. Compared with parents who were employed full-time, parents who had no outside work, were employed part-time, were self-employed, or had a domestic side job were more likely to spank their children: the adjusted ORs in the ordinal logistic regression model were 1.21 (95% CI, 1.10-1.33), 1.10 (95% CI, 1.03-1.17), 1.11 (95% CI, 1.02-1.20), 1.19 (95% CI, 1.04-1.37), respectively. In the sensitivity analysis where adjusted ORs were examined when stratified by respondents, respondents did not modify the associations between caregivers and spanking (data not shown). In the same manner, the sensitivity analysis where adjusted ORs were examined when stratified by caregivers, caregivers did not modify the associations between respondents and spanking (data not shown). Table 4 shows the association between risk factors and spanking, stratified by cohort generation. In general, the associations were similar in both cohorts. However, the likelihood of spanking was lower when both parents were the respondents to the survey only in the first cohort. On the contrary, the associations with spanking in the second cohort were more evident when other family members were respondent, for the presence of a sibling of the child, parents' lower level of education, and parents' work types (no outside work, employed part-time, self-employed, domestic side job, and others). Table 5 shows the associations between risk factors and spanking, stratified by gender of the child. The associations were generally similar in both genders. However, the likelihood of spanking was lower in girls when the father was the respondent to the survey.

DISCUSSION
In this study using national longitudinal survey data, we present an overview of family members' spanking of 3.5-year-old children and the associated factors. We found more than 70% of family members spanked their children, which was consistent with previous reports; corresponding rates exceed 70% in some European, Asian, and African countries. 13,21 We found that more children in the second cohort investigated in 2013 were never spanked (38%) compared with those in the first cohort investigated in 2004-2005 (23%). The increase in the prevalence of never spanking among the second cohort might reflect greater social awareness of child abuse. Even though spanking and corporal punishment have not been banned in Japan in the investigated periods, the Child Welfare Act (Act No. 164 of 1947) 15 and the Act on the Prevention, etc of Child Abuse (Act No. 82 of 2000) 16 were amended several times during the interval between the survey waves. In fact, the substantial increase in the annual number of reported cases of suspected child abuse to child Factors for Spanking of 3.5-year-old Children in Japan  23 The cohort generations modified the associations between parents' level of education or parents' work type and spanking. This may reflect changes in the distribution and nature of these variables over time, as more parents shifted to higher education and engaged in full-time employment in the second cohort. These changes over generations also reflected the fact that, if the respondents of the survey were other family members (ie, mostly grandparents), they were more likely to spank the children than in families where mothers were the respondents.
In this study, if the respondents of the survey were fathers or both parents, they were less likely to spank their children than in families where the respondents were mothers. This finding was consistent with the results of previous studies in the United States and Hong Kong. 4,10,24,25 indicating that fathers spanked less frequently than mothers, although this was considered a consequence of mothers typically spending more time with children than fathers. 10,24 However, a previous study using the same longitudinal survey data as the present study showed that the father's involvement in childcare prevented unintentional injuries, 26 implying that the father's involvement in parenting may be beneficial for adverse child outcomes. It may also be because responses to the survey by both parents may reflect a good marital relationship, which is beneficial for refraining from spanking. 27 Consistent with previous studies in the United States, 1,24 being a boy was a risk factor of being spanked for misbehavior. This is probably due to the different types of misbehaviors and parents' reactions by gender. 10 The analysis stratified by the child's gender showed that being a girl was a modifier for the father's spanking reaction to misbehavior.
Regarding family structure, the presence of siblings of the child and not living in a two-parent household were risk factors of being spanked, which was also consistent with previous findings. 1, 10 We also found that living in a three-generation household was a protective factor of spanking, which could be due to informal social support for parents from family members and greater assistance with household chores. 10,28,29 Regarding socioeconomic factors, our results found that a lower level of parental education or family income were risk factors of spanking, which was consistent with the findings of previous Factors for Spanking of 3.5-year-old Children in Japan Adjusted for respondent of the survey, caregivers of the child, cohort generation, gender of the child, presence of siblings of the child, living in a two-parent household, living in a three-generation household, parent's age, education, working hours, work type, family income, and always use of other forms of discipline. Baba S, et al. reports in the United Kingdom and United States. 10,30 We found that, in addition to unstable work types, such as part-time employment, self-employment, and domestic side jobs, no outside work (ie, housewife) was also a risk factor for spanking after adjusting for other socioeconomic factors. This is consistent with a previous study of 1,662 participants in Hong Kong showing the association between the respondent's (mother's or father's) unemployment and corporal punishments, including spanking. 25 The reason for this may be mainly because unemployed parents typically spend more time with their children. However, considering the fact that these results were obtained after adjusting for working hours, other assumptions could be added. First, mothers who are employed full-time could have better moods at home compared with non-working mothers. 31 Second, since full-time employment offers a wider range of social and professional contacts, 32 parents who are employed full-time could have developed more social skills, including anger management, which help them choose other strategies of child discipline apart from corporal punishment. Third, parents who work full-time may feel guilty about leaving their children to go work, resulting in warmer parenting 33,34 to compensate for their absence during working hours compared with non-working parents. We found that more than half of the family members in this study had spanked their children. It is important to promote greater involvement of fathers in parenting and to educate parents in alternative forms of discipline to handle their child's misbehavior or conflicting situations in order to prevent or aOR, adjusted odds ratio; CI, confidence interval. a Adjusted for respondent of the survey, caregivers of the child, cohort generation, gender of the child, presence of siblings of the child, living in a two-parent household, living in a three-generation household, parent's age, education, working hours, work type, family income, and always use of other forms of discipline.
Factors for Spanking of 3.5-year-old Children in Japan reduce the use of spanking. 35 For example, the most prevalent reaction under such situations in Sweden is "to divert the child's attention from the cause of the misbehavior", followed by "discussion with the child". 36 In the United States, "time outs", which physically remove the child from where he=she is misbehaving, are preferred. 35 These alternative forms of discipline do not seem to be commonly used in Japan considering the response items in the questionnaire. Parent training programs are one opportunity to provide information on alternative forms of discipline used in other countries. According to a meta-analysis evaluating effective parent training programs, requiring parents to practice with their child during training sessions showed better parent and child outcomes. 37 Considering our data that more mothers were in the workforce and more fathers were involved in parenting in the second cohort than in the first cohort, providing accessible parenting programs held not only on weekdays, but with flexible participation schedules (eg, on weekends, or as webinars, or even during lunch time at the workplace) could be suggested.
The major strengths of this study include the large populationbased sample that consisted of two generation cohorts. Also, the response choices to the question on how to react to child misbehavior comprised five reactions, and respondents were required to indicate the frequency for each reaction, which could reduce underestimation of the prevalence of spanking compared with previous studies that required answering about the frequency of spanking only. 6 Several limitations should be discussed. First, there might be cultural or ethnic differences regarding the use of spanking. Therefore, it might be difficult to generalize these findings to other populations. Second, some exposure variables were obtained in the same wave survey as the outcome "spanking"; thus, our findings could not confirm a causal association. Third, the outcome "spanking" was not validated or evaluated in an objective manner by referring to the number of times spanking was used during some specific period of time. However, this is difficult in practice because there are no gold standard measurements for the comparison. Fourth, the fathers responded in this survey were limited in number and these subjects may have been biased toward "good fathers". Therefore, further studies which require father's and mother's responses respectively, will be necessary to confirm the protective effect of father's involvement on spanking. Fifth, residual confounding could have occurred from unmeasured confounding variables. For example, the variables related to parents' stress or children's temperaments were not investigated in this study.
In conclusion, our study suggested that spanking is less frequent in more recent generations and the father's involvement in parenting may be protective against spanking. Moreover, the child's gender, family structure, and factors of low socioeconomic status including no outside work were associated with spanking.