Neglect,Sexual Abuse,and Witnessing Intimate Partner Violence During Childhood Predicts Later Life Violent Attitudes Against Children Among Kenyan Women: Evidence of Intergenerational Risk Transmission From Cross-Sectional Data

Abstract

Violence against children, including corporal punishment, remains a global concern. Understanding sources of support for corporal punishment within cultures, and the potential for intergenerational transmission of child maltreatment, is essential for policy-development and community engagement to protect children. In this study, we use data from a cross-section of women in Meru County, Kenya (n = 1,974) to profile attitudes toward violence against children using the Velicer Attitudes Towards Violence–Child subscale. We find reported histories of sexual abuse, emotional and physical neglect, and witnessing interpersonal violence during childhood predict more violent attitudes toward children in adulthood. The pathway between these forms of child maltreatment and violent attitudes is significantly mediated by family function, perceived stress, and attitudes toward violence against women. Interventions to prevent sexual abuse, intimate partner violence, and promote attachments between parents and children may benefit future generations in this population. Furthermore, secondary prevention of the effects of these childhood adversities may require development of social support, improving family function and challenging violent attitudes against women.

Keywords

intergenerational transmission of trauma domestic violence cultural contexts prevention of child abuse child abuse neglect children exposed to domestic violence

Background

Child abuse continues to be an international public health problem (Carter et al., 2006). Despite studies showing a high value of children and desire for parenthood across cultures (Dyer, 2007), the frequency of child abuse remains high. Recent analyses of data from 28 low- and middle-income countries (LMICs) show that the median proportion of children in Africa who have experienced psychological, moderate, and severe physical abuse is 83%, 64% and 43%, respectively, compared with median proportions of 56%, 46%, and 9% in other transitional countries (Akmatov, 2010). A survey conducted in Kenya in 2010 found a high prevalence of child abuse, with 32% of females reporting childhood experiences of sexual violence, and 66% of females reporting childhood experiences of physical violence (Centers for Disease Control and Prevention, United Nations Children’s Fund, Kenya National Bureau of Statistics, 2010).

Physical abuse and corporal punishment of children carry with a preponderance of negative consequences for the child, including many that continue to affect the individual throughout his or her adult life. A meta-analysis of data from high-income countries show that individuals who were physically abused as a child are 54% more likely to develop depressive disorders and 92% more likely to use drugs; individuals who experienced emotional abuse or neglect as a child have even higher probabilities of developing depressive disorders (Norman et al., 2012). Empirical evidence supports the view that early life “toxic stress” explains some of the negative lifelong consequences that arise from childhood abuse (Johnson, Riley, Granger, & Riis, 2013). Toxic stress occurs when a person undergoes significant amount of stress during formative years such that it alters the formation of the brain structure, endocrine system, and immune system (Johnson et al., 2013).

A recent study of corporal punishment in Kenyan elementary schools reveals areas of concern and further research. While the Kenyan government outlawed the use of corporal punishment in schools in 2001, the behavior persists. Focus groups with Kenyan teachers revealed that parents commonly support the practice (Mweru, 2010). As scientific evidence from the past 50 years, covering multiple cultural and economic contexts, uniformly indicates that spanking has a negative effect on the short and long-term development of children (Gershoff, 2010; Gershoff & Grogan-Kaylor, 2016), it is worthwhile to consider support for the behavior across cultural and economic contexts.

Evidence over the past two decades has revealed intergenerational risks of child abuse and neglect in high-income (Pears & Capaldi, 2001) and low-income settings (Crombach & Bambonyé, 2015). Parents who were more exposed to abuse during childhood were more likely to perpetrate abuse against their children in a longitudinal sample from the United States (Pears & Capaldi, 2001). In a recent cross-sectional study from Burundi, adults who recalled experiencing more physical abuse during childhood were more likely to report experiencing intimate partner violence and to perpetrate violence against children than were adults with fewer experiences of physical abuse during childhood (Crombach & Bambonyé, 2015). These findings were consistent for both genders. These patterns tend to be replicated across income strata in high-income countries (Thornberry & Henry, 2013), though not without methodological problems (Thornberry & Henry, 2013; Widom, Czaja, & DuMont, 2015).

One of the methodological issues in extant literature on child maltreatment is a lack of consistent definition for child maltreatment (Widom et al., 2015), confusing efforts to target specific forms of adverse childhood experiences (ACEs). Various forms of family dysfunction, child abuse, child neglect, and other hazardous exposures in the social environment often co-occur (Dong et al., 2004; Edleson, 2001). In part to assist with the equivocation between definitions of child maltreatment, and to foster global surveillance of ACEs, the World Health Organization (WHO) recently released an instrument to assess the presence and frequency of 13 childhood adversities: physical abuse, emotional abuse, sexual abuse, physical neglect, emotional neglect, exposure to peer violence, exposure to community violence, exposure to collective violence, exposure to violence between parents or caregivers, experiencing loss of a parent through divorce or death, living with a person who abuses chemical substances, living with household members who were mentally ill, depressed or suicidal, and living with household members who were imprisoned (WHO, 2012). Investigation into the hypothesis of intergenerational transmission of child abuse must consider the potential for unobserved influence by other forms of childhood adversities when these other forms are not included in statistical models.

Further clarity is required to understand the potential mechanisms of intergenerational transmission of child maltreatment. There are many current theories on the causes of family violence (Hyde-Nolan & Juliao, 2012), many of which may be implicated in intergenerational transmission of child maltreatment. Attachment theory has long been utilized to explain associations between childhood trauma and maltreatment, family functioning in adulthood, and coercive or violent attitudes toward children in high-income countries (Banyard, 1997; Rothbaum, Rosen, Ujiie, & Uchida, 2002; Sheridan, 1995; Strahan, 1991). Attachment theory in this context can be summarized as follows: children who experience abuse or neglect develop poorer sense of selves and less intimate adult relationships during adulthood. This reduction in interpersonal security and intimacy yields higher stress, poorer social relationships, and more controlling or violent attitudes toward children (Serbin & Karp, 2004).

Higher perceived stress and lower family functioning were recently found to be associated with more childhood adversities in a Kenyan population of women (Goodman et al., 2016). Higher perceived stress has previously been found to predict corporal punishment attitudes among Kenyan grandmothers, supporting the hypothesis that stress mediates the association between childhood adversities and the spanking of children in this population (Oburu & Palmérus, 2003). Consistent with attachment theory, prior research suggests that social support may moderate the relationship between stress and child-directed violent attitudes (Chan, 1994; Rodriguez & Richardson, 2007). Social support can provide a personal affirmation, which may reduce the need to control one’s environment according to attachment theory.

Social learning theory, in contrast to attachment theory, posits that observing and experiencing violence during childhood teaches children to be aggressive toward others during adulthood through operant conditioning (Renner & Slack, 2006). Social learning theory may explain the cyclical relationship between and co-occurrence of intimate partner violence and child-directed violence (Edleson, 2001; Renner & Slack, 2006). Attitudes justifying intimate partner violence have been found higher among Kenyan women with more childhood adversities (Goodman et al., 2017). It is not known whether attitudes justifying violence against women predict attitudes supporting violence against children in Kenya, nor whether this association mediates any risk of intergenerational transmission of child maltreatment.

The present study utilizes the definitions of childhood adversities provided by the WHO’s Adverse Childhood Experiences International Questionnaire (ACE-IQ) instrument to determine whether, and which of, these exposures predict later life attitudes supporting corporal punishment of children among Kenyan mothers. We assess whether perceived stress, attitudes toward intimate partner violence, and family functioning mediate observed associations between childhood adversities and later life violent attitudes. We hypothesized that social support moderates the relationship between stress and violent attitudes toward children.

Method

Study Design

Data were collected using a cross-sectional design with systematic random sampling in 23 purposely selected townships across the North Igembe District of Meru County, Kenya over a 5-week period during May and June 2015. The survey instrument was a closed questionnaire requiring approximately 45 min of respondent time. The study was conducted in joint partnership between the staff of Sodzo International and the Maua Methodist Hospital Community Health Department, in an effort to understand characteristics perceived to be relevant to the health status of children in the community.

Sample Selection

Inclusion criteria for the study required that a primary caregiving woman be available and consent to be interviewed. Households without children under the age of 18 years of age were excluded. Households were selected using a skipping pattern that included every other household, and followed a random-number generated pattern through streets and pathways. Six more rural areas were selected for geographic variation, and these areas were serviced through community health clinics by the study-administrating hospital. Of the 2,129 houses visited and found eligible, 51 houses refused (2.4%) and 104 women were not at home (4.8%). A total of 1,974 interviews were completed and included in the present analysis.

Survey Administration

The questionnaire was administered in Kimeru, the local language, by trained interviewers from Maua Methodist Hospital. The survey was translated from English to Kimeru and back translated to English for comparison and refinement. Interviewers conducted interviews in households and requested study participants to identify a private and quiet location to conduct the interview.

Outcome Measure

The Velicer Attitudes Towards Violence Scale–Child subscale (VATVS-cs) measures attitudes supporting corporal punishment against children through an 8-item 5-point Likert-type scale (Anderson, Benjamin, Wood, & Bonacci, 2006). Items include statements like “children deserve to be spanked for temper tantrums” and “a teacher hitting a child when he or she does something bad on purpose teaches the child a good lesson.” The subscale showed good reliability in the current population (α = .72). The outcome measure was generated by averaging item responses with mean-imputation for missing items. Higher scores on the VATVS-cs indicate more supportive attitudes of violence against children.

Antecedent Variables

The ACE-IQ index was used to measure ACEs reported by the respondent (WHO, 2012). The index measures three types of abuse (physical, sexual, and emotional), two types of neglect (physical and emotional), five types of household dysfunction (living with someone with a drug/alcohol addiction, living with someone who was imprisoned, living with someone struggling with suicide/depression, witnessing intimate partner violence between household adults, and the loss of a parent through divorce or death), and three types of violence (peer/bullying, community, and collective). Each category of exposure reported by the respondent was added to generate a range of 0 to 13, with a unit increase for each exposure reported.

Mediating Variables

Family functioning was measured using the general subscale of the McMaster Family Assessment Device. The measure evaluates the overall functioning of a family—that is, the ability to communicate about personal matters, and make collective decisions in a context of mutual respect and understanding. The 12-item scale includes statements such as “in times of crisis, we can turn to each other for support” and “individuals are accepted for who they are” that are assessed on a 4-point Likert-type scale. Higher scores indicate greater facility to coordinate activity within a context of respect and acceptance. The whole subscale showed adequate reliability (α = .65).

Perceived stress was measured using the 10-item version of the scale from Cohen et al. (1983). The measure asks respondents about their ability to manage unexpected events, their emotional response to uncontrolled events, and general sense of powerlessness over the past month using a 5-point Likert-type scale with items such as “in the past month, how often have you been unable to control the important things in your life” and “in the past month, how often have you felt that you were on top of things.” Higher scale scores indicated higher perceived stress from the past month. The scale showed good reliability in the present sample (α = .81).

Attitudes supporting intimate partner violence were assessed using five items from the WHO’s questionnaire on violence against women (WHO, 2005). These items assessed acceptance of reasons why a man may hit his wife, including her unsatisfactory completion of housework, her disobedience, her refusal to have sex, her suspicion that he has been sexually unfaithful, and him finding out that she has been unfaithful. These five binary response items are combined, and in the present population had acceptable internal consistency (KR20 = 0.81).

Moderating Variable

Social support was measured using the Multidimensional Scale of Perceived Social Support (MSPSS; Zimet, Dahlem, Zimet, & Farley, 1988; α = .94). The MSPSS measures social support through three different sources of support—familial, friendship, and a special/romantic partner. The scale includes 12 items measured on a 7-point, Likert-type scale. Average scores were calculated and used in analysis, and item-level missing values were mean imputed.

Control Variable

Education, wealth status, and age were included as control variables in all statistical models. Education was assessed through a single item asking respondents to indicate the number of years of school they completed successfully. Wealth was measured as the number of household assets owned including land, electricity, radio, television, telephone, refrigerator, bicycle, motorbike, automobile, and the number of rooms used for sleeping. These items showed a strong single factor solution in principal factor analysis. Age was included in each model as a control variable. Age was assessed as years of life completed at last birthday.

Data Analysis

The outcome variable, VATVS-cs, was modeled as a continuous variable. The first round of modeling determined which specific ACEs were significantly associated with higher VATVS-cs scores using multivariable linear regression, checking variance inflation factors to ensure multicollinearity was not a concern. Precision estimates were calculated using 1,000 bootstrap replicates; unstandardized and fully standardized coefficients are reported.

The exposures that were found to predict significantly higher VATVS-cs scores were then summed and used as the continuous antecedent variable for the second round of modeling—conditional process analysis. Potential mediators in the analysis included family functioning, perceived stress, and justifying beliefs toward intimate partner violence. Social support was included as a potential moderator of the perceived stress-VATVS-cs association. Analyses were conducted using structural equation modeling, and confirmed using the PROCESS command in SPSS to assess path significance (Hayes, 2012). PROCESS utilizes the bootstrapping method (1,000 replicates) to generate confidence intervals to make empirical statements about path significance.

STATA v.14 was used to conduct all data analyses, with the exception of the assessment of path significance performed with the PROCESS macro in SPSS. EpiData v.3.1 was used to double enter data from the original paper-based questionnaires.

Ethical Considerations

All respondents were informed that participation was voluntary and they would not receive any direct benefit for participating. All respondents provided informed consent prior to beginning the interview. An ethics committee hosted by the sponsoring hospital, comprising regional governmental and community leaders, provided ethical approval prior to study initiation. The Institutional Review Board of the University of Texas Medical Branch provided permission for the publication of collected data.

Results

Description of data is provided in Table 1. The most highly endorsed reason to hit a child was the statement “a teacher hitting a child when he/she does something bad on purpose teaches the child a good lesson.” The least highly endorsed reason to hit a child was the statement “giving mischievous children a quick slap is the best way to quickly end trouble.” The most commonly experienced adversity was exposure to violence in the community, and the least commonly experienced adversity was exposure to the loss of a parent to death or divorce. The average respondent reported 7.6 adversities. The mean respondent age was 38.2 years and the mean number of completed school years was 6.

Table 1.

Respondent Characteristics, by Variable Type.

	Item-Level n	M (%)	95%CI
Outcome variable
Velicer Attitudes Towards Violence Scale–Child subscale, range 8-40	1,950	28.74	28.44	29.03
Children should be spanked for temper tantrums.	1,965	3.224	3.149	3.299
Punishing a child physically when she or he deserves it will make him or her a responsible and mature adult.	1,968	3.765	3.705	3.824
Giving mischievous children a quick slap is the best way to quickly end trouble.	1,966	2.923	2.848	2.999
A parent hitting a child when he or she does something bad teaches the child a good lesson.	1,970	3.881	3.825	3.938
A child’s habitual disobedience should be punished physically.	1,966	3.808	3.75	3.866
An adult should beat a child with a strap or stick for being expelled from school.	1,970	2.98	2.907	3.053
Young children who refuse to obey should be whipped.	1,970	3.97	3.917	4.023
A teacher hitting a child when he or she does something bad on purpose teaches the child a good lesson.	1,968	4.176	4.125	4.227
Antecedent variable
ACE-IQ, range 0-13	1,930	7.64	7.51	7.77
Abuse, emotional	1,974	72.46%	70.29%	74.62%
Abuse, physical	1,934	57.15%	54.73%	59.58%
Abuse, sexual	1,956	39.69%	37.04%	42.35%
Neglect, emotional	1,974	57.65%	55.25%	60.04%
Neglect, physical	1,951	63.41%	61.06%	65.76%
Violent exposure, bullying	1,974	65.08%	62.74%	67.42%
Violent exposure, collective	1,931	53.94%	51.51%	56.38%
Violent exposure, community	1,944	92.30%	91.00%	93.60%
Household dysfunction, alcohol/drug abuse	1,956	57.02%	54.62%	59.43%
Household dysfunction, mental illness/depression/suicide	1,964	31.66%	29.40%	33.92%
Household dysfunction, imprisoned family member	1,966	35.95%	33.62%	38.28%
Household dysfunction, loss of parent (death/divorce)	1,974	20.56%	18.60%	22.52%
Household dysfunction, witness intimate partner violence	1,966	78.03%	76.02%	80.04%
Mediating variables
Perceived stress, range 0-4	1,973	2.187	2.166	2.208
Family functioning, range 1-4	1,974	2.659	2.638	2.681
Justification of intimate partner violence, range 0-5	1,971	2.244	2.153	2.334
Moderating variable
Social support, range 1-7	1,973	5.424	5.364	5.484
Control variables
Wealth index, range 0-14	1,974	4.555	4.43	4.679
Age, range 18-89	1,974	38.21	37.61	38.82
Years of completed school, range 0-14	1,974	5.96	5.78	6.14

Note. Means and 95% confidence intervals provided for each variable used in statistical modeling. ACE-IQ uses binary method. Range for each scale provided. ACE-IQ = Adverse Childhood Experiences International Questionnaire; CI = confidence intervals.

Table 2 shows the results of multivariable linear regression. Model 1 shows the initial model with all potential exposures included, along with control variables. As seen, the childhood adversities that predict significantly higher VATVS-cs scores are emotional neglect (coef.: 1.44, 95%CI = [0.72, 2.15]), physical neglect (coef: 1.46, [0.69, 2.23]), sexual abuse (coef: 0.87, [0.15, 1.6]), and witnessing intimate partner violence (coef: 1.51, [0.6, 2.43]). Living with someone suffering from alcohol/drug abuse, or with a mental illness/depression/suicide predicted lower VATVS-cs scores. Experiencing physical or emotional abuse during childhood was not significantly associated with VATVS-cs scores. A composite measure of adversities directly predicting VATVS-cs scores was created, combining sexual abuse, emotional neglect, physical neglect, and witnessing intimate partner violence during childhood.

Table 2.

Associations Between Specific Childhood Adversities and Later Life Attitudes Towards Violence Against Children Among Kenyan Mothers.

	Coef	95% CI		β
Abuse, emotional	0.39	−0.52	1.31	.03
Abuse, physical	−0.65	−1.47	0.17	−.05
Abuse, sexual	0.87*	0.15	1.60	.07
Neglect, emotional	1.44***	0.72	2.15	.12
Neglect, physical	1.46***	0.69	2.23	.12
Violent exposure, bullying	0.4	−0.40	1.20	.03
Violent exposure, collective	0.03	−0.70	0.77	.00
Violent exposure, community	−0.35	−1.60	0.89	−.02
Household dysfunction, alcohol/drug abuse	−0.79*	−1.52	−0.05	−.06
Household dysfunction, mental illness/depression/suicide	−1.05**	−1.88	−0.22	−.08
Household dysfunction, imprisoned family member	−0.36	−1.14	0.41	−.03
Household dysfunction, loss of parent (death/divorce)	−0.19	−1.09	0.71	−.01
Household dysfunction, witness intimate partner violence	1.51***	0.60	2.43	.10
Age (years)	−0.01	−0.05	0.02	−.03
Wealth index (0-14)	−0.17*	−0.34	−0.01	−.07
Years of completed school	−0.09	−0.21	0.03	−.05

Note. “Coef” indicates unstandardized coefficient from linear regression model with 1,000 bootstrap replicates. β indicates fully standardized coefficient. ACE score measured using WHO ACE-IQ binary method (2015). CI = confidence intervals; ACE-IQ = Adverse Childhood Experiences International Questionnaire.

p < .05. **p < .01. ***p < .001.

Details on the composite measure of selected ACEs can be seen in Table 3. The selected ACEs each separately predicted higher VATVS-cs scores. The average number of selected ACEs experienced was 2.4 (95%CI = [2.34, 2.46]). Less than 8% of respondents experienced none of the selected ACEs. Nearly 20% of respondents experienced all four selected ACEs (95%CI = [17.4, 21.5%]).

Table 3.

Selected ACEs, Based on Significant Association With Higher VATVS-cs Scores.

	M (%)	95%CI
Selected ACEs, 0-4	2.4	2.34	2.46
Number of selected ACEs
0	7.48%	6.15%	8.82%
1	14.56%	12.77%	16.35%
2	27.59%	25.32%	29.86%
3	30.93%	28.58%	33.27%
4	19.44%	17.43%	21.45%

Note. Selected ACEs variable created by combining experience of separate ACEs significantly associated with VATVS-cs. This includes sexual abuse, emotional neglect, physical neglect, and witnessing intimate partner violence. ACE = adverse childhood experiences; VATVS = Velicer Attitudes Towards Violence Scale; CI = confidence intervals.

The antecedent and final models from the conditional process analysis are shown in Table 4. The antecedent model shows that a standard deviation increase in the selected ACEs is associated with a .23 increase in VATVS-cs score. When including the mediating variables, this association decreases to 0.14—a 39% reduction. Each pathway depicted in the conceptual model was statistically significant using the bootstrap method (Figure 1).

Table 4.

Conditional Process Analysis of Attitudes Towards Violence Against Children and Associated Adverse Childhood Experiences of Kenyan Mothers.

	Model 1 (Antecedent Only)				Model 2 (Final Model)
	Coef	95% CI		β	Coef	95% CI		β
Exposure variables
Selected ACEs (0-4)	1.1***	0.84	1.36	0.23	0.7***	0.45	0.95	0.13
Conditional process variables
Family function					−1.11**	−1.88	−0.34	−0.06
Justification of IPV					0.64***	0.47	0.81	0.18
Social support					1.87**	0.51	3.23	0.34
Perceived stress					6.49***	3.28	9.71	0.38
Social support × Perceived stress					−0.57*	−1.1	0	−0.26
Control variables
Age (years)	−0.01	−0.04	0.02	−0.02	−0.02	−0	0.01	−0.04
Wealth index (0-14)	−0.18*	−0.33	−0.03	−0.07	−0.19**	−0.3	−0.1	−0.05
Years of completed school	−0.06	−0.16	0.04	−0.04	0.01	−0.1	0.11	−0.02

Note. “Coef” indicates unstandardized coefficient from linear regression model with 1000 bootstrap replicates. β indicates fully standardized coefficient. Selected ACEs determined by statistically significant association with violent attitudes toward children (Table 2), and include physical neglect, emotional neglect, sexual abuse and witnessing intimate partner violence during childhood. CI = confidence intervals; ACE = adverse childhood experiences; IPV = intimate partner violence.

p < .05. **p < .01. ***p < .001.

Figure 1.

Conditional process analysis, selected ACEs and VATVS-cs score, conceptual model (n = 1974).

Experiencing more of the selected ACEs predicts higher perceived stress, lower family function, and more justifying beliefs about intimate partner violence. In turn, more stress, lower family functioning, and more justifying beliefs about intimate partner violence predict more support for violence against children. The perceived stress-VATVS-cs pathway was significantly moderated by social support. Women with more stress were less likely to support violence against children if they had more social support.

Discussion

ACEs were common in the present population, with more than half of respondents from a community-based sample of caregiving women (n = 1,974) indicating childhood experiences of physical abuse (72.5%), emotional abuse (57.2%), emotional neglect (57.7%), physical neglect (63.4%), and witnessing intimate partner violence (78%). Nearly 40% of women reported some experience of sexual abuse during childhood (39.7%). We assessed the empirical relationship between ACEs and later life violent attitudes toward children, expecting experiences of physical and emotional abuse to significantly predict later life violent attitudes toward children. We were surprised that physical and emotional abuse did not predict violent attitudes toward children, while physical and emotional neglect significantly predicted more violent attitudes toward children.

Two theories that have been posited to explain empirical observations of intergenerational transmission of childhood maltreatment—attachment theory and social learning theory—were implicit in our investigation. Attachment theory posits that less emotional attachment with adults during childhood would result in higher stress during adulthood, worse family functioning, and a greater probability of seeking to control behaviors of children in care through use of corporal punishment. Empirical data support this theory. Family functioning and perceived stress significantly mediated the pathway between childhood experiences of neglect, sexual abuse, and witnessing intimate partner violence and later life violent attitudes toward children. Social support moderated the association between perceived stress and violent attitudes toward children, yielding further support to the attachment theory of intergenerational transmission of childhood maltreatment.

Social learning theory posits that behavior performed by role models becomes internalized as appropriate, and replicated through continued reference to that behavior and operant conditioning. Social learning theory forwards the hypothesis that experiences of physical abuse during childhood beget support for physical abuse against children later in life. Our data do not support this hypothesis in this population. Observing intimate partner violence during childhood does predict support of violence against children and women later in life—a finding potentially consistent with social learning theory, but more so with Gelles’ social control theory (Gelles, 1983). The social control theory posits that conflict occurs due to efforts to maintain power and control within relationships. Rigid role expectations may explain the absence of emotional warmth between parents and children (Jaffe, Gullone, & Hughes, 2010), the acceptability of wife battering by women (Ehrensaft, Langhinrichsen-Rohling, Heyman, O’Leary, & Lawrence, 1999), and the use of physical violence to control children (Gershoff, 2002).

It is likely that multiple theories may explain intergenerational transmission of child maltreatment risk, which is only beginning to be empirically explored in sub-Saharan Africa (Crombach & Bambonyé, 2015). This is the first study to our knowledge to assess the potential for intergenerational transmission of child maltreatment in East Africa, and more study is required to explore and expand findings. Longitudinal, preferably multi-generational, study is required to verify associations observed in these data. Further exploration of existing theories would enable policy makers and researchers to develop mental models to guide funding and policy decisions. Investigation and validation of attachment theory in this context would establish relative deficiencies in emotional attachments between neglected children and their social network across the lifespan. Furthermore, validating the utility of attachment theory would require demonstrating that neglected children who had difficulties developing emotional bonds were therefore more likely to have worse emotional attachment with their own progeny. Social learning theory may be expanded beyond specific acts of violence against children and include internalizing role rigidity, which may perpetuate across the lifespan and provide justification for violence against women and children. Social control theory would require the demonstration that role rigid beliefs significantly predict controlling beliefs toward children and women in this context. To be useful, social control theory must also demonstrate that changes in role rigid beliefs can be changed, and that these changes predict less violent attitudes toward women and children.

Data from this population of Kenyan women demonstrate the potential intergenerational transmission of risk for children, as has been found in diverse economic and cultural settings. Social support in this population predicted more violent attitudes toward children, even though it downregulated the association between stress and violent attitudes. In a culture where violent attitudes toward children are common and normative, social support may increase adherence to these attitudes through social network factors. That is, people who share common beliefs are more likely to support each other (McPherson, Smith-Lovin, & Cook, 2001), and when beliefs about violence toward children are more common, support is less available for those who disagree. Common beliefs in the benefits of corporal punishment against children may similarly explain why there was no association between reporting physical abuse during childhood and violent attitudes toward children during adulthood. More than half of respondents reported being spanked, slapped, kicked, punched or beaten up by a parent or guardian at least a few times, meaning that if a respondent did not experience this adversity themselves, they likely knew someone who did. In contexts where such behavior is more isolated, it is possible that the direct experience is more predictive of later life attitudes. Mechanisms of intergenerational transmission of child abuse risk likely vary across settings; in the present setting, we found evidence of the risk for intergenerational transmission of child abuse. The widespread presence of child physical abuse, and supporting attitudes toward corporal punishment, may explain some of the unique mechanisms observed in this context.

While theoretical development of the risk of intergenerational transmission of child maltreatment continues, there are policy implications of the present research. Kenya has several existing laws and policies in place to address child abuse currently. In 1990, Kenya ratified the United Nations Convention on the Rights of the Child, followed by the African Charter on the Rights and Welfare of the Child in 2000 (Mildred & Plummer, 2009), The Children’s Act was passed by the Kenyan parliament in 2002 and offered expanded consideration of the protection of children, including the right to health care, education, and protection from sexual abuse and exploitation (Mildred & Plummer, 2009). To strengthen existing sex offenses legislation and increase rates of reporting, the Sexual Offenses Act was passed in 2006 (Mildred & Plummer, 2009). Most child welfare services are run by voluntary and international agencies, and the majority of sexual abuse and exploitation cases are dealt with informally (Mildred & Plummer, 2009). Although many positive improvements of legislation concerning child abuse have occurred in Kenya, involvement at multiple levels of society is necessary to enhance and effectively carry out legal reform (Landgren, 2005).

At the societal and community level, better access to subsidized child care has been shown to be correlated with lower rates of child maltreatment (Klevens, Barnett, Florence, & Moore, 2015). The current finding that more stress predicts higher attitudes toward violence underscores the need for increased support for maternal caregivers to alleviate sources of stress. More research into sources of perceived stress among maternal caregivers is warranted, and may improve both the mental health of maternal caregivers and the children in their care. Furthermore, child care providers may be sources of emotional attachment and protective environments for children with difficult home environments (McCartney, Dearing, Taylor, & Bub, 2007).

Policies that allow for continuity of child health care may also benefit children in abusive situations. Systems such as the Safe Environment for Every Kid (SEEK) model suggest that participation of health care professionals such as pediatricians may also reduce the rates of child maltreatment (Dubowitz, Feigelman, Lane, & Kim, 2008). Recent conferences held in Kenya have offered instructions for helping to prevent child abuse at the community level, including teaching marketplace women to look after neighborhood children, starting children’s rights clubs in schools, and training soldiers to see child protection as part of their job (Mildred & Plummer, 2009). The UN Committee on the Rights of the Child recommends that professionals who work with children, such as law enforcement officials, health personnel, teachers, and social workers, be provided training and sensitization on children’s rights to create a protective environment for children that goes beyond the legal sphere (United Nations International Children’s Emergency Fund [UNICEF], 1997). Current data showing that increased social support predicts violent attitudes toward children indicate the need to address commonly held beliefs, such that social support can work in favor of protecting children from experiencing violence. Legal action, media campaigns, and directed education on the rights of children should be developed as envisioned by current Kenyan policy.

The results of the present study indicate that family functioning is significantly associated with the potential for child abuse. Having a strong relationship with a caring adult may be the most influential factor in a child’s positive psychosocial development (Gorrese & Ruggieri, 2012). This emphasizes the incredible importance of reinforcing positive parenting and family care practices and discouraging the continuation of harmful ones (Landgren, 2005), all of which may be more likely to occur in better functioning families. Analysis of Parent–Child Interaction Therapy and the Triple P-Positive Parenting Program has revealed positive effects, including a reduction in child-behavior and parenting problems (Thomas, 2007). In addition, analysis of parenting intervention in LMICs shows the potential of positive effects and emphasizes the need for further studies that support parenting intervention as a possible strategy for the prevention of child maltreatment in LMIC (Knerr, Gardner, & Cluver, 2013). Improving the capacity of families (and the community) to care for, support, and offer positive role models to children is imperative in the creation of an environment protective against abuse (Landgren, 2005). Addressing cultural sources of respect for women in families and reducing violent attitudes toward women may further benefit children at risk of spanking and physical abuse.

At the level of the individual, strategies such as life skills training at an early age offers children the opportunity to become an agent of their own protection (Landgren, 2005). Teaching children skills in decision making, problem solving, and critical and creative thinking as well as information about their rights can make them less vulnerable and more resourceful (Landgren, 2005). Helping children to develop positive self-esteem, self-confidence, and coping self-efficacy may increase individual resilience and therefore allow for the handling of adverse childhood events in a healthier way and with less perceived stress (Betancourt et al., 2011).

Limitations

Data analysis is limited by a number of factors. Selection bias may be present as some potential respondents refused. As the response rate was high (97.6%), we believe findings represent the majority of potential respondents in our survey area, but cannot know baseline differences between those who consented and those who refused participation. Recall bias and other forms of response bias may have systematic altered associations. For example, respondents may have misremembered the number of childhood adversities they experienced. Previous study of ACE scores found a tendency to underreport childhood adversities (Anda et al., 2006), though this assertion requires consideration of the cultural context. Respondents may have been driven by social desirability biases, over or underreporting each variable. Such a risk is inherent in survey-based research. There is no way to determine whether these biases are differential or nondifferential forms of misclassification without more rigorous study methods. Assuming they are nondifferential, observed associations would be conservative and tend toward null associations. The limitations of this study suggest the potential benefit of longitudinal study of the connection between childhood adversities of the maternal caregiver and subsequent attitudes toward violence and abuse potential in the next generation. A study of this nature would require more funding than is currently devoted to such research.

Conclusion

We assessed whether violent attitudes toward children among Kenyan women caregivers were predicted by the woman’s own experience of childhood adversity (ACEs), after controlling for wealth, education and age. Furthermore, we explored which ACEs predict later life violent attitudes toward children and some of the potential mediators of this association. Surprisingly, the experience of emotional and physical abuse during childhood did not predict later life violent attitudes toward children, but sexual abuse, emotional neglect, physical neglect and witnessing intimate partner violence among household adults did predict significantly higher violent attitudes toward children. Family functioning, perceived stress, and affirming justifications for violence against women mediated nearly 30% of the observed association between the selected ACEs and violent attitudes toward children. Current government practice in Kenya to protect children from abuse involves largely punitive measures for child abusers. Data here suggest the need for promoting positive attachments between children and parents, and the need for emotional responsiveness throughout childhood. Improved emotional attachment and responsiveness may improve lifelong perceptions of stress, family functioning, and reduced justification of violence against women and children. In addition, protection of intimate partners may benefit children in households with domestic conflict and reduce lifelong perceptions of the acceptability of violence against women and children. Further empirical and theoretical research into intergenerational transmission of adversity is required in Kenya and across sub-Saharan Africa, though present data provide more support to the attachment theory than social learning theory.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Michael L. Goodman, MDiv, MPH, DrPH, is a postdoctoral fellow in the Center for Tropical Diseases and a research fellow at the Center to Eliminate Health Disparities at University of Texas Medical Branch (UTMB). He completed his training at the University of Texas School of Public Health where he studied community health practice, epidemiology, and global health. His research interests focus on determinants of health among children, especially vulnerable children globally. In addition, he is interested in the lifelong consequences of childhood adversities, and mechanisms of interrupting intergenerational transmission of adversities. He leads projects in Kenya helping street-involved children find ways back into their community, and strengthening capacity of families to care for vulnerable children. He enjoys mentoring and teaching medical students.

Andrea Hindman is a medical student at the University of Texas Medical Branch in Galveston, TX. She is passionate about children and women’s health, special needs populations, and mental health.

Philip H. Keiser has dedicated his medical career to the treatment and prevention of HIV, and recently has begun developing programs to benefit vulnerable children and communities in Kenya. He is the director of the international HIV clinical program and professor of medicine at the University of Texas Medical Branch in Galveston.

Stanley Gitari is the director of community and strategic partnerships at Maua Methodist Hospital in Meru County, Kenya. He completed his bachelor’s in nursing in the United States, and his master’s in public health in Ghana. He has served the community of Meru County for nearly 30 years, beginning numerous community programs to benefit vulnerable children and families.

Katherine Ackerman Porter completed her master’s degree in public health at the University of Texas Health Science Center in Houston, TX. She served for two years in Malawi through the Peace Corps, and is passionate about the welfare of children and families in Eastern and Southern Africa. She currently works for UNICEF USA.

Ben G. Raimer began his career as a pediatrician before moving into administrative roles at the University of Texas Medical Branch in Galveston, TX. He currently a Professor in the Department of Pediatrics and serves on numerous gubernatorial committees related to health and child welfare in Texas, and is the Senior Vice President of Health Policy and Legislative Affairs at the University of Texas Medical Branch. Throughout his career, he has won numerous state- and national-awards for the prevention of child abuse.

References

Akmatov

M. K.

(2010). Child abuse in 28 developing and transitional countries—Results from the Multiple Indicator Cluster Surveys. International Journal of Epidemiology, 40, 219-227.

Anda

R. F.

Felitti

V. J.

Bremner

J. D.

Walker

J. D.

Whitfield

C. H.

Perry

B. D.

Giles

W. H.

(2006). The enduring effects of abuse and related adverse experiences in childhood. European Archives of Psychiatry and Clinical Neuroscience, 256, 174-186.

Anderson

C. A.

Benjamin

A. J.

Wood

P. K.

Bonacci

A. M.

(2006). Development and testing of the Velicer Attitudes Toward Violence Scale: Evidence for a four-factor model. Aggressive Behavior, 32, 122-136.

Banyard

V. L.

(1997). The impact of childhood sexual abuse and family functioning on four dimensions of women’s later parenting. Child Abuse & Neglect, 21, 1095-1107.

Betancourt

T. S.

Meyers-Ohki

Stulac

S. N.

Barrera

A. E.

Mushashi

Beardslee

W. R

. (2011). Nothing can defeat combined hands (Abashize hamwe ntakibananira): Protective processes and resilience in Rwandan children and families affected by HIV/AIDS. Social Science & Medicine, 73, 693-701.

Carter

Y. H.

Bannon

M. J.

Limbert

C. L. A. I. R. E.

Docherty

Barlow

(2006). Improving child protection: A systematic review of training and procedural interventions. Archives of Disease in Childhood, 91, 740-743.

Centers for Disease Control and Prevention, United Nations Children’s Fund, Kenya National Bureau of Statistics. (2010). Violence Against Children in Kenya. Retrieved from http://www.unicef.org/esaro/VAC_in_Kenya.pdf

Chan

Y. C.

(1994). Parenting stress and social support of mothers who physically abuse their children in Hong Kong. Child Abuse & Neglect, 18, 261-269.

Cohen

Kamarck

Mermelstein

(1983). A global measure of perceived stress. Journal of Health and Social Behavior, 385-396.

10.

Crombach

Bambonyé

(2015). Intergenerational violence in Burundi: Experienced childhood maltreatment increases the risk of abusive child rearing and intimate partner violence. European Journal of Psychotraumatology, 6, Article 26995.

11.

Dong

Anda

R. F.

Felitti

V. J.

Dube

S. R.

Williamson

D. F.

Thompson

T. J.

Giles

W. H.

(2004). The interrelatedness of multiple forms of childhood abuse, neglect, and household dysfunction. Child Abuse & Neglect, 28, 771-784.

12.

Dubowitz

Feigelman

Lane

Kim

(2009). Pediatric primary care to help prevent child maltreatment: The Safe Environment for Every Kid (SEEK) model. Pediatrics, 123, 858-864.

13.

Edleson

J. L.

(2001). Studying the co-occurrence of child maltreatment and domestic violence in families. In Graham-Bermann

S. A.

Edleson

J. L.

(Eds.), Domestic violence in the lives of children: The future of research, intervention, and social policy (pp. 91-110). Washington, DC: American Psychological Association.

14.

Ehrensaft

M. K.

Langhinrichsen-Rohling

Heyman

R. E.

O’Leary

K. D.

Lawrence

(1999). Feeling controlled in marriage: A phenomenon specific to physically aggressive couples? Journal of Family Psychology, 13, 20-32.

15.

Gelles

R. J.

(1983). An exchange/social control theory. In The dark side of families: Current family violence research (pp. 151-165). Beverly Hills, CA: Sage.

16.

Gershoff

E. T.

(2002). Corporal punishment by parents and associated child behaviors and experiences: A meta-analytic and theoretical review. Psychological Bulletin, 128, 539-579.

17.

Gershoff

E. T.

(2010). More harm than good: A summary of scientific research on the intended and unintended effects of corporal punishment on children. Law and Contemporary Problems, 73, 31-56.

18.

Gershoff

E. T.

Grogan-Kaylor

(2016). Spanking and child outcomes: Old controversies and new meta-analyses. Journal of Family Psychology, 30, 453-469.

19.

Goodman

M. L.

Gutarra

Billingsley

K. M.

Keiser

P. H.

Gitari

(2016). Childhood exposure to emotional abuse and later life stress among Kenyan women: a mediation analysis of cross-sectional data. Anxiety, Stress, & Coping, 1-28.

20.

Goodman

M. L.

Patel

Tutak

Keiser

P. H.

Gitari

Raimer

B. G.

(2017). Adverse childhood experiences predict greater intimate partner violence justification in Kenyan Women. Unpublished manuscript.

21.

Gorrese

Ruggieri

(2012). Peer attachment: A meta-analytic review of gender and age differences and associations with parent attachment. Journal of Youth and Adolescence, 41(5), 650-672.

22.

Hayes

A. F.

(2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/public/process2012.pdf

23.

Hyde-Nolan

M. E.

Juliao

(2012). Theoretical basis for family violence. In Family violence: What health care providers need to know (pp. 5-16). Sudbury, MA: Jones & Bartlett Learning.

24.

Jaffe

Gullone

Hughes

E. K.

(2010). The roles of temperamental dispositions and perceived parenting behaviours in the use of two emotion regulation strategies in late childhood. Journal of Applied Developmental Psychology, 31(1), 47-59.

25.

Johnson

S. B.

Riley

A. W.

Granger

D. A.

Riis

(2013). The science of early life toxic stress for pediatric practice and advocacy. Pediatrics, 131, 319-327.

26.

Klevens

Barnett

S. L.

Florence

Moore

(2015). Exploring policies for the reduction of child physical abuse and neglect. Child Abuse & Neglect, 40, 1-11.

27.

Knerr

Gardner

Cluver

(2013). Improving positive parenting skills and reducing harsh and abusive parenting in low- and middle-income countries: A systematic review. Preventive Science, 14, 352-363.

28.

Landgren

(2005). The protective environment: Development support for child protection. Human Rights Quarterly, 27, 214-248.

29.

McCartney

Dearing

Taylor

B. A.

Bub

K. L.

(2007). Quality child care supports the achievement of low-income children: Direct and indirect pathways through caregiving and the home environment. Journal of Applied Developmental Psychology, 28, 411-426.

30.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415-444.

31.

Mildred

Plummer

C. A.

(2009). Responding to child sexual abuse in the United States and Kenya: Child protection and children’s rights. Children and Youth Services Review, 31(6), 601-608.

32.

Mweru

(2010). Why are Kenyan teachers still using corporal punishment eight years after a ban on corporal punishment? Child Abuse Review, 19(4), 248-258.

33.

Norman

R. E.

Byambaa

Butchart

Scott

Vos

(2012). The long-term health consequences of child physical abuse, emotional abuse, and neglect: A systematic review and meta-analysis. PLoS Medicine, 9(11), Article e1001349. doi:10.1371/journal.pmed.1001349

34.

Oburu

P. O.

Palmérus

(2003). Parenting stress and self-reported discipline strategies of Kenyan caregiving grandmothers. International Journal of Behavioral Development, 27, 505-512.

35.

Pears

K. C.

Capaldi

D. M.

(2001). Intergenerational transmission of abuse: A two-generational prospective study of an at-risk sample. Child Abuse & Neglect, 25, 1439-1461.

36.

Renner

L. M.

Slack

K. S.

(2006). Intimate partner violence and child maltreatment: Understanding intra- and intergenerational connections. Child Abuse & Neglect, 30, 599-617.

37.

Rodriguez

C. M.

Richardson

M. J.

(2007). Stress and anger as contextual factors and preexisting cognitive schemas: Predicting parental child maltreatment risk. Child Maltreatment, 12, 325-337.

38.

Rothbaum

Rosen

Ujiie

Uchida

(2002). Family systems theory, attachment theory, and culture. Family Process, 41, 328-350.

39.

Serbin

L. A.

Karp

(2004). The intergenerational transfer of psychosocial risk: Mediators of vulnerability and resilience. Annual Reviews Psychology, 55, 333-363.

40.

Sheridan

M. J.

(1995). A proposed intergenerational model of substance abuse, family functioning, and abuse/neglect. Child Abuse & Neglect, 19, 519-530.

41.

Strahan

B. J.

(1991). Attachment theory and family functioning: Expectations and congruencies. Australian Journal of Marriage and Family, 12, 12-26.

42.

Thomas

Zimmer-Gembeck

M. J.

(2007). Behavioral outcomes of parent-child interaction therapy and Triple P—Positive Parenting Program: A review and meta-analysis. Journal of Abnormal Child Psychology, 35(3), 475-495.

43.

Thornberry

T. P.

Henry

K. L.

(2013). Intergenerational continuity in maltreatment. Journal of Abnormal Child Psychology, 41, 555-569.

44.

Widom

C. S.

Czaja

S. J.

DuMont

K. A.

(2015). Intergenerational transmission of child abuse and neglect: Real or detection bias? Science, 347, 1480-1485.

45.

World Health Organization. (2005). Multi-country study on women’s health and domestic violence against women: Summary report of initial results on prevalence, health outcomes and women’s responses (Summary Report). Retrieved from http://whqlibdoc.who.int/publications/2005/924159358X_eng.pdf

46.

World Health Organization. (2012). Adverse Childhood Experiences International Questionnaire (ACE-IQ). Retrieved from http://www.who.int/violence_injury_prevention/violence/activities/adverse_childhood_experiences/en/index.html

47.

United Nations International Children’s Emergency Fund. (1997). United Nations convention on the rights of the child. Retrieved from http://www.unicef.org/crc/conven.htm

48.

Zimet

G. D.

Dahlem

N. W.

Zimet

S. G.

Farley

G. K.

(1988). The multidimensional scale of perceived social support. Journal of Personality Assessment, 52, 30-41.