Andrea L Pazdera, PhD, School of Behavioral Health Sciences; Lenore M McWey, PhD and Ann Mullis, PhD, Department of Family and Child Sciences, Florida State University; Joyce Carbonell, PhD, Department of Psychology, Florida State University.
Child Sexual Abuse and the Superfluous Association with Negative Parenting Outcomes: The Role of Symptoms as Predictors
Article first published online: 9 FEB 2012
© 2012 American Association for Marriage and Family Therapy
Journal of Marital and Family Therapy
Volume 39, Issue 1, pages 98–111, January 2013
How to Cite
Pazdera, A. L., McWey, L. M., Mullis, A. and Carbonell, J. (2013), Child Sexual Abuse and the Superfluous Association with Negative Parenting Outcomes: The Role of Symptoms as Predictors. Journal of Marital and Family Therapy, 39: 98–111. doi: 10.1111/j.1752-0606.2011.00272.x
- Issue published online: 15 JAN 2013
- Article first published online: 9 FEB 2012
The purpose of this study was to investigate the relationship between child sexual abuse and high-risk maternal parenting indicators and the extent to which maternal depression and self-perceived parenting competence influence that relationship. Using path analysis, results indicate maternal depression and parenting sense of competence mediate the relationship between child sexual abuse and outcome variables. Post hoc analyses indicated that child sexual abuse was significantly associated with decreased parenting sense of competence, controlling for depression. These results highlight that the pathways for increased risk in parenting outcomes for child sexual abuse survivors may be indirect and associated with beliefs of their own sense of competence and depression as opposed to a direct association with sexual abuse itself. Implications are discussed.
Extant research indicates that maternal survivors of child sexual abuse are at increased risk of negative parenting outcomes such as decreased parenting satisfaction and sensitivity, increased anxiety about intimate parenting, increased parent-child role reversal, increased use of physical discipline, and increased child abuse potential (Banyard, 1997; Briere, 1992; DiLillo, Tremblay, & Peterson, 2000; Douglas, 2000; Libby, Orton, Beals, Buchwald, & Manson, 2008; Noll, 2008). However, there is debate about whether the relationship between child sexual abuse (CSA) and parenting outcomes is direct or indirect.
This study contributes to the literature by examining the relationship between CSA and high-risk maternal parenting indicators (parenting stress and maltreatment behaviors) and the extent to which maternal depression and self-perceived parenting competence influence that relationship and identifies the processes through which an experience of child sexual abuse might later lead to parenting difficulty. Identifying the conditions through which maternal survivors of child sexual abuse are at risk of later maltreating their own children is essential for prevention and intervention with survivors of child sexual abuse (Noll, 2008). The theory of symbolic interactionism guided this study.
Child sexual abuse is a serious public health issue, impacting millions of children each year around the world (World Health Organization, 2008). A number of researchers have investigated the immediate and long-term emotional and behavioral impacts of CSA. Noted survivor response patterns to sexual abuse include the following: difficulties in interpersonal relationships, dissociation, marital conflicts, social anxiety, suicidal ideation, posttraumatic stress disorder, substance abuse, and eating disorders (Hetzel-Riggin, Brausch, & Montgomery, 2007; Nelson et al., 2002).
Over the past two decades, researchers have progressively focused on the relationship between CSA on later parenting outcomes. Childhood sexual abuse has been found to impact later parenting outcomes such as increased use of physical discipline and permissive parenting (Banyard, 1997; Ruscio, 2001). One study found that women who had been sexually abused reported less emotional control while interacting with their children (Cole, Woolger, Power, & Smith, 1992). Results from several studies indicate that a maternal history of CSA predicts increased child abuse potential (Caliso & Milner, 1992; Craig & Sprang, 2007; DiLillo et al., 2000; Whissell, Lewko, Carriere, & Radford, 1990).
Yet, the predisposition of hypotheses and research questions to assume a deficiency in mothers with histories of sexual abuse ignores the potential ways in which many mothers with histories of childhood sexual abuse successfully parent their children (Breckenridge, 2006). Minimal research exists examining the influence of mother’s beliefs about themselves and their parenting skills in the link between CSA and negative parenting outcomes. A few recent studies have shown that the association between history of child sexual abuse and the abuse of a child is not a direct path. Indeed, some existing research demonstrates that maternal attitudes and beliefs may mediate the relationship between child sexual abuse and parenting (Banyard, Williams, & Siegal, 2003; Schuetze & Das Eiden, 2005).
Research also suggests that child characteristics such as child age and sex may be associated with child maltreatment. The average age of children who are maltreated is younger than the average age of children in the United States (Gelles, 1997), and younger children are at an increased likelihood of experiencing multiple child maltreatment recurrences than older children (Bae, Soloman, & Gelles, 2009; Fluke, Yuan, & Edwards, 1999). Additionally, research indicates that child age is a significant predictor of parental attitudes toward physical discipline, use of physical discipline, and parental verbal abuse (Jackson et al., 1999). Although both sexes are equally maltreated, sex of child has been associated with proneness to specific forms of abuse. Male children tend to experience physical abuse at higher rates than females (Wolfe, 1987), which may meaningfully impact maltreatment scores. This study used child age and sex as a control variables not as an indicator of a child’s contribution and responsibility in child maltreatment, but rather as an indicator of how parenting beliefs and behaviors are influenced by child characteristics.
The theory of symbolic interactionism was applied to the study of how maternal histories of child sexual abuse influence parenting behaviors. Symbolic interactionism seeks to explain social behavior by understanding how people attach meanings to their experiences with others. The meanings that people have for things are central to the choices that people make and to how they behave in any given situation (Blumer, 1969). Symbolic interactionism theorists also assume that the meaning that a person makes of a particular situation, such as child misbehavior, comes from the social connections that the person has experienced previously and how those previous connections relate to the particular situation being experienced at that time. Importantly, symbolic interactionism asserts that our perception of reality directs behavior and has real consequences, whether or not our perception is “real” (Blumer). The emphasis on constructed meaning is an important component of the fit between symbolic interactionism and the study of mothers who are CSA survivors. Symbolic interaction theoretically explains the indirect association through the concept of meaning and can be used to test meaning as a mediator or moderator in that association. Symbolic interactionism can be used to explain maternal behavior via constructed meanings at the individual, family, and societal levels. Individual level meanings were tested in this study.
Research has consistently indicated that CSA is associated with increased depressive symptomology and decreased feelings of competence in parenting (Banyard, 1997; Fitzgerald, Shipman, Jackson, McMahon, & Hanley, 2000; Mapp, 2006; Schuetze & Das Eiden, 2005). Maternal CSA survivors have reported feeling less confident and satisfied with themselves as parents, more disorganized, inconsistent, and less emotionally controlled (Cohen, 1992; Cole & Woolger, 1989; Cole et al., 1992; Douglas, 2000). Mothers’ beliefs about themselves and their competence may in fact influence their self-reported behavior and the behavior itself. Potential mediating variables including maternal depression and perceptions of parenting competence can help in the understanding of associations between CSA and child maltreatment.
The purpose of this study was to investigate the relationship between child sexual abuse and high-risk maternal parenting indicators and the extent to which maternal depression and self-perceived parenting competence influence that relationship using symbolic interactionism as the theoretical framework. Based on symbolic interactionism and previous literature, the research questions were as follows: (a) Do mothers with a history of CSA have increased parenting stress and maltreatment behaviors than mothers without CSA histories, controlling for child age and sex? (b) Do mothers with a history of CSA have increased levels of depression and decreased self-perceived parenting competence than mothers without CSA histories, controlling for child age and sex? (c) Are increased levels of depression and decreased self-perceived parenting competence associated with increased parenting stress and maltreatment behaviors, controlling for child age and sex? (d) What is the extent to which maternal depression influences the relationship of CSA on parenting stress and maltreatment behaviors, controlling for child age and sex? (e) What is the extent to which sense of competence influences the relationship of CSA on parenting stress and maltreatment behaviors, controlling for child age and sex?
Data for this study were extracted from the Parenting Among Women Sexually Abused in Childhood (PAWSAIC) dataset and expedited human subjects approval from the Florida State University Institutional Review Board (IRB) was obtained on October 23, 2009. The PAWSAIC contains retrospective information on child sexual abuse and up-to-date self-report information on parenting competence, parenting stress, discipline practices, family functioning, allowing for the investigation of potential causal effects of CSA on later development of parenting outcomes. The purpose of PAWSAIC was to explore the relationship between experiences of sexual abuse before the age of 18 and selected parenting beliefs and behaviors (Benedict, 1998). PAWSAIC examined a low-income population of mothers and includes a large sample of African American women allowing for an ethnically diverse sample.
Parenting Among Women Sexually Abused in Childhood was a two-wave longitudinal study in which participants were recruited for baseline interviews from two prenatal clinics located within a University-based hospital. At wave one, women were interviewed during a regular prenatal visit between 28 and 32 weeks gestation to explore the impact of childhood sexual abuse on adult women’s functioning during pregnancy (Benedict, 1998). Participants who completed the baseline interviews were contacted for second interviews when their children were between 2 and 4 years of age (Benedict). History of childhood sexual abuse was measured at the baseline interview. All other study variables were measured only during the second wave of data. All PAWSAIC participants were located through information gathered from medical records (Benedict, 1998). All potential study participants received a letter describing the study, introducing the investigator, describing how names were obtained, and explaining that the participation would entail a 75–90 minute interview for which they would receive $25.00 (Benedict). Interviewers were trained and given a pre-test prior to interviewing to ensure quality of the interview data (Benedict). Trained interviewers conducted 80.9% of the interviews face-to-face and 19.1% were conducted over the telephone because of maternal geographical location. All questionnaires were checked for completeness and accuracy, and interviewers were questioned about any missing items (Benedict).
Participants were limited to the sample of 265 mothers, who participated in a baseline interview when they were primiparous and participated a second interview when their first child was 2–4 years old (Benedict, 1998). African American women comprised 73% of the sample and Caucasian women comprised the remainder of the sample (Benedict). Eighty-one percent of the sample reported having a high school diploma or GED. Nearly 18% reported being in school and 52% reported working at the time of second interview (Benedict). Approximately, 15% of the women had intimate partners, and 88% of those partners were working (Benedict). Over 50% of the sample had never been married. Over 40% of the sample reported an income less than $15,000 annually, and 33% reported incomes over $30,000 per year (Benedict).
Child Sexual Abuse. An adapted version of Russell’s (1983) questionnaire was used to assess for a history of child sexual abuse during the first interview. Women who reported at least one contact or non-contact episode before the age of 18 were included in the index group. Generally the abuse perpetrator had to be 5 years older than the victim except where the women reported that force was used in the assault (Benedict, 1998). In those situations, women were included as in the index group regardless of age difference (Benedict). Although Russell’s questionnaire assesses for intrafamilial and extrafamilial abuse, the PAWSAIC dataset provides no information regarding the relationship between the CSA survivors and their abuse perpetrators, and abuse was categorized as a dichotomous variable with (0) representing the control group and (1) representing the index group.
Parenting Stress. The Parenting Stress Index (PSI) was administered in the second wave of the study (Abidin, 1995). The PSI has 101-items measuring two domains: child characteristics and mother characteristics. Higher scores equal greater levels of parenting stress. The PSI is predictive of physical abuse and neglect, as well as poor parenting (Abidin) and has been used in studies with both abusive and non-abusive parents and with parents identified as “at-risk” for parenting problems (DiLauro, 2004; Huebner, 2002). The scale has reliability coefficients ranging from .89 to .93 for child and parent domains, and between .65 and .96 for the total scale (Abidin). For this sample, an analysis of internal consistency yielded an alpha coefficient of .82.
Maltreatment Behavior. A scale was derived from a number of questions designed to assess respondents’ attitudes and behaviors regarding verbal and physical discipline. This scale is comprised of 30-items included in the interview that were designed to measure the participants’ use of various verbal and physical discipline techniques with their child. Each item was assessed on a 4-point scale to assess frequency of use ranging from strongly disagree (1) to strongly agree (4). Higher scores indicate higher use of maltreatment behaviors. For this sample, an analysis of internal consistency yielded an alpha coefficient of .87.
Depression. The Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977) was administered to the participants in the second wave of the study. This 20-item scale measures depressive symptomology in the general population. The respondents report on the frequency of various feelings and behaviors over the past week. Responses range from rarely or none of the time (less than 1 day a week, value = 1) to most or all of the time (5–7 days a week, value = 4). Higher scores indicate more depressive symptoms. This scale has been shown to be reliable and valid over time and has been used with a number of populations (Ensel, 1982). The scale has average reliability coefficients at .85 in non-clinical samples, and .90 in psychiatric populations (Simmons, Huddleston-Casas, & Berrv, 2007). For this sample, an analysis of internal consistency yielded an alpha coefficient of .87. In the post hoc analysis, criteria previously developed in similar populations (Cohen, Goh, & Gustave, 2009) were used to divide depression scores into clinical depression (CES-D ≥ 36) and moderate depression (CES-D score of ≥ 28) and subclinical depression (CES-D score < 28). Dummy variables were coded for each level of depression, with (0) representing the control group and (1) representing the index group.
Parenting Sense of Competence. The Parenting Sense of Competence Scale (PSOC) was administered in the second wave of the study (Gibaud-Wallsoton & Wandersman, 1978) was administered in the second wave of the study. The PSOC has 17 items measuring two dimensions of competence: parenting satisfaction and parenting efficacy. Each item is answered on a 6-point scale, ranging from strongly agree (1) to strongly disagree (6). Higher scores indicate greater perceived competence. The scale has reliability coefficients ranging from .75 to .82 for the satisfaction subscale .70 to .76 for the efficacy subscale, and between .75 and .88 for the total scale (Johnston & Mash, 1989). For this sample, the alpha coefficient was .81.
Control Variables. The age and sex of the mother’s first child at time of interview were used as control variables in this study. Age was coded as the last full year of the child’s life. Thus, infants who were less than 1 year are coded as 0. The sex of the child was coded as a dichotomous variable with (0) representing males (1) representing females.
Path analysis using AMOS (Arbukle, 1997) was used to test the research questions. A critical element of meaningful results is adequate statistical power (Cohen, 1992). Thus, a power analysis was conducted to ascertain if sufficient statistical power = .80 would be achieved (Cohen). The full model had 24 parameters to be estimated. Kline (2005) suggests that for statistical power = .80 in path analysis, the ratio of sample size to parameters must at minimum be 10:1 cases for each parameter estimated. The full model for this study included 24 parameters and 265 participants and the ratio of sample size to parameters was 11:1. Path analysis was used to examine the major research questions, and missing data were addressed using Full Information Maximum Likelihood (FIML) estimation in AMOS (Arbukle, 1997).
The first step in path analyses is to specify the model (Kline, 2005). History of CSA was dummy coded (0 = no sexual abuse, 1 = childhood history of sexual abuse) and child sex was dummy coded (0 = male sex, 1 = female sex). Child age, maternal depression, parenting sense of competence, and parenting stress were composite variables in this study. Then the structural model was tested in a series of four steps testing mediation (Baron & Kenny, 1986). Maximum likelihood estimation procedures were used, and standardized parameter estimates are presented in the results section. The goodness-of-fit of the models was examined by using the comparative fit index (CFI), the normed fit index (NFI), and the root mean square error of approximation (RMSEA). The chi-square difference compares the fit of successive nested models.
Separate structural models were evaluated in the four steps for establishing mediation discussed by Baron and Kenny (1986). In the first model, the association between CSA and parenting outcomes was evaluated; in the second model, the association between CSA and the mediators (maternal depression and parenting sense of competence) was evaluated; in the third model, the association between the mediators and parenting outcomes was evaluated, controlling for CSA, child age, and child sex. The parameter estimates were evaluated to examine the research question: would the association between CSA and parenting outcomes be reduced after controlling for the effect of the mediators on the outcome variables. The fourth model was modified to fit the question: would the association between CSA and parenting be mediated by maternal depression and parenting sense of competence (the direct paths from CSA to the two parenting outcomes were set to zero) and the fit indices for this final hypothesized model (model 4) were examined. A fifth model was conducted to adjust for parsimony. The chi-square difference test was used to compare the fit of successive nested models (model 3, model 4, and model 5).
Separate path models were evaluated in the four steps for establishing mediation (Baron & Kenny, 1986). Chi-square is reported in the results below because of the influence of chi-square on most model fit indices (Kline, 2005). For models with between 75 and 200 participants, chi-square is a reasonable measure of model fit (Kenny, 2010). However, for models with more than 200 participants, the chi-square is regularly statistically significant, indicating that the null hypothesis cannot be rejected. Chi-square is also affected by the size of the correlations in the model: the larger the correlations, the poorer the fit (Kenny). For these reasons, alternative measures of fit have been developed and are reported below. The Comparative Fit Index (CFI) was used to assess the improvement in fit of successive nested models in comparison with the baseline model. Values of CFI greater than .90 indicate reasonably good fit of a model (Kline, 2005). The Normed Fit Index (NFI) defines the null model as a model in which all of the correlations or covariances are zero (Kenny). NFI values below .90 indicate a need to respecify the model, values between .90 and .95 are considered acceptable, and values above .95 are considered good (Kenny). The Root Mean Square Error of Approximation (RMSEA) is a “badness of fit index,” values ≤ .05 indicate close approximate fit, values between .05 and .08 indicate a reasonable error of approximation, and values above .08 indicate poor model fit (Kline, 2005, p. 138).
Results from the first model indicated that CSA was not significantly associated with either parenting outcomes: parenting stress β = −.07, p = ns and maltreatment behavior β = .06, p = ns. Mothers with a history of sexual abuse were no more likely to have parenting stress than mothers not reporting such histories and were also no more likely to report maltreatment behaviors. Similarly, neither child age nor sex was significantly associated with parenting outcomes. Standardized estimates for child age were β = −.04, p = ns for parenting stress and β = .08, p = ns for maltreatment behavior. Standardized estimates for child sex were (β = .01, p = ns) for parenting stress and β = .10, p = ns for maltreatment β = .52, p < .001. This model did not fit the data well, χ2(14) = 189.37, p < .001, CFI = .084, NFI = .12, RMSEA = .218. The first research question: do mothers with a history of CSA have significantly higher levels of parenting stress and maltreatment behaviors than mothers without CSA histories, controlling for child age and sex was not supported, as the null hypothesis was not rejected. A direct path from the independent variable to the outcome variables (step 1 of mediation) is not required for mediation to occur (MacKinnon, Fairchild, & Fritz, 2007) because the path from the independent variable to the outcome variables is implied when steps 2 and 3 of mediation are met (Kenny, Kashy, & Bolger, 1998).
The second model tested research questions two and three, do mothers with a history of CSA have higher levels of depression and lower levels of self-perceived parenting competence than mothers without CSA histories. Results from the second model indicated that CSA was associated with higher maternal depression β = .12, p = .05 and lower parenting sense of competence at β = .11, p = .05. Mothers with a childhood history of sexual abuse were more likely to be depressed and to perceive their parenting abilities as less competent. Standardized estimates for child age were β = −.01, p = ns for depression and β = −.02, p = ns for parenting sense of competence. Standardized estimates for child sex were β = .05, p = ns for depression and β = .02, p = ns for parenting sense of competence. The second and third research questions, do mothers with a history of CSA would have higher levels of depression and lower levels of parenting sense of competence than mothers without CSA histories, controlling for child age and sex were not fully confirmed. Although significant relationships between CSA and the mediating variables were confirmed, this model did not fit the data well, χ2(15) = 209.17, p < .001, CFI = .00, NFI = .05, RMSEA = .22.
The third model included paths from CSA to the mediators, paths from CSA to parenting outcomes, and paths from the mediators to parenting outcomes. Parameter estimates from the mediators to parenting outcomes were examined to see whether there was a significant association between the hypothesized mediators and parenting outcomes. Once again, paths from CSA to the parenting outcomes were not statistically significant: parenting stress β = .01, p = ns and maltreatment behavior β = .02, p = ns. Likewise, standardized estimates for child sex were not significantly associated with either parenting outcomes, β = .01, p = ns, for parenting stress and, β = .07, p = ns, for maltreatment behavior. Standardized estimates for child age were, β = −.04, p = ns, for parenting stress and, β = .08, p = .05, for maltreatment behavior. Both mediator variables were significantly associated with the parenting stress and maltreatment behavior. Maternal depression was significantly associated with both parenting stress (β = .38, p < .001) and maltreatment behavior (β = .34, p < .001). Depressed mothers had more parenting stress and were more likely to report maltreatment behavior with their female children. Higher parenting sense of competence was also significantly associated with lower reported parenting stress (β = −.55, p < .001) and was also associated with lower reported maltreatment behaviors (β = −.22, p = .01). The fourth research question, what is the extent to which maternal depression influences the relationship of CSA on parenting stress and maltreatment behaviors, controlling for child age and sex? was confirmed as significant. Although there were statistically significant relationships between variables, indicators of overall model fit were poor, χ2(10) = 46.52, p < .001, CFI = .81, NFI = .79, RMSEA = .12.
The fourth model included paths from CSA to the mediators and paths from the mediators to parenting outcomes. The direct paths from CSA to the outcome variables were set to zero (Baron & Kenny, 1986). This model indicated an improved, yet insufficient, model fit, χ2(6) = 36.02, p < .001, CFI = .84, NFI = .84, RMSEA = .14. Research questions five and six, to what extent do depression and parenting competence would mediate the relationship between CSA on parenting stress and maltreatment behaviors, controlling for child age and sex were not fully confirmed. Although there were statistically significant relationships between variables, indicators of overall model fit were poor, χ2(6) = 36.02, p < .001, CFI = .84, NFI = .84, RMSEA = .14.
Finally, in model 5, the paths from CSA to parenting outcomes were set to zero, additionally, all other insignificant direct paths were set to zero. Direct paths from CSA to depression and parenting sense of competence, direct paths from depression and parenting sense of competence to parenting stress and maltreatment behavior, a direct path from parenting stress to maltreatment behavior and a direct path from child age to maltreatment behavior were included in model 5. This model resulted in an insufficient improvement in fit, χ2(7) = 36.18, p < .001, CFI = .85, NFI = .83, RMSEA = .13, however, it was a more parsimonious model. Standardized estimates are provided in Table 1 for models 4 and 5, and the final model is depicted in Figure 1.
|Depression||PSOC||Parenting Stress||Maltreatment Behavior|
|Child Sexual Abuse||.12*||.11*||.07||.06|
Post Hoc Analyses
In the path analyses conducted, depression was shown to be significantly associated CSA and with both parenting stress and maltreatment behavior. Because of poor model fit, mediation could not be established. However, estimates indicated that a maternal history of child sexual abuse predicted increased depressive symptoms and that depressive symptoms predicted increased parenting stress and increased maltreatment behavior. Although the history of sexual abuse interacted with depression significantly in predicting parenting stress and maltreatment behavior, the mediation model had poor model fit and did not account for whether child sexual abuse predicts negative parenting outcomes beyond that which could be explained by the increased likelihood of depressive symptoms in child sexual abuse survivors. Thus, the relationship between child sexual abuse and depression was further explored by examining the relationship between child sexual abuse and the parenting outcomes for women who reported levels of clinical depression, moderate depression, and sub-clinical depression.
ANOVA analyses were conducted to examine differences in the relationship between clinically depressed mothers in the index group and clinically depressed mothers in the control group and the outcomes variables: parenting stress and maltreatment behavior. Additional analyses were conducted to examine differences between moderately depressed mothers in the index and control groups and the outcome variables (See Table 2).
|Depression Level||Index Group||Control Group|
|N||M (SD)||Percent||N||M (SD)||Percent|
|Clinical||40||46.23 (7.95)||15.2||43||44.47 (8.17)||16.3|
|Moderate||44||31.11 (2.65)||16.7||66||30.6 (0.49)||25.1|
|No Depression||19||25.30 (1.49)||7.2||47||25.24 (1.49)||17.9|
For the clinically depressed portion of the sample, CSA was not significantly correlated with parenting stress, F(1, 20) = 0.05, p = ns. Nor was CSA significantly correlated with maltreatment behavior, F(1, 81) = 0.04, p = ns. However, CSA was significantly correlated with parenting sense of competence, F(1, 81) = 3.90, p < .05. Thus, clinically depressed women with histories of CSA are no more likely to experience parenting stress or report maltreatment behavior than clinically depressed women without reported histories of CSA, but are more likely to report a decreased sense of parenting competence.
For the moderately depressed portion of the sample, CSA was not significantly correlated with parenting stress, F(1, 32) = 0.11, p = ns, maltreatment behavior, F(1, 109) = 1.52, p = ns, or parenting sense of competence F(1, 105) = 0.17, p = ns. Results indicate that moderately depressed women with histories of CSA are no more likely to experience parenting stress or report maltreatment behavior than moderately depressed women without histories of CSA.
For the non-clinically depressed portion of the sample, CSA was not significantly correlated with parenting stress, F(1, 19) = 1.45, p = ns, maltreatment behavior, F(1, 68) = 1.51, p = ns, or parenting sense of competence F(1, 67) = 1.24, p = ns. Results indicate that women with histories of CSA and minimal depressive symptoms are no more likely to experience parenting stress or report maltreatment behavior than women without reported histories of CSA and minimal depressive symptoms. The only significant finding in the post hoc analysis was the relationship between child sexual abuse and parenting sense of competence in clinically depressed participants. This finding suggests that CSA influences parenting sense of competence above and beyond the influence of depression on mothers’ sense of competence.
This study was conducted to identify the processes through which an experience of child sexual abuse might later lead to parenting stress and maltreatment behavior. Recent literature suggests that maternal history of sexual abuse is associated with parenting outcomes such as increased use of physical discipline and child abuse potential (Banyard, 1997; Craig & Sprang, 2007). However, the predisposition of research questions to assume a deficiency in mothers with histories of sexual abuse ignores the potential ways in which many mothers with histories of childhood sexual abuse successfully parent their children, and often fails to consider potential indirect processes through which negative outcomes are determined (Breckenridge, 2006). Symbolic interactionism was used to examine meanings that may mediate the association between mothers’ histories of child sexual abuse and later parenting outcomes, specifically mother’s beliefs in their own parenting competence and their experiences of depression.
In this study, mothers with a history of CSA did not have significantly higher levels of parenting stress or maltreatment behaviors compared to mothers without CSA histories. However, mothers with a history of CSA had higher levels of depression than mothers without CSA histories, controlling for child age and sex. These findings can make important contributions informing clinical practice. Results indicate that the relationship between CSA and parenting outcomes is an indirect function of internal belief systems (i.e., depression and parenting sense of competence) rather than a direct relationship as frequently cited in past research (e.g., Banyard, 1997; Craig & Sprang, 2007).
The elucidation of indirect processes through which negative parenting outcomes of survivors of child sexual abuse may be determined is in congruence with the theoretical framework of symbolic interactionism. Blumer (1969) theorized that the meanings that people have for things are central to the choices that people make and to how they behave in any given situation. Symbolic interactionism explains how beliefs associated with depression and parenting sense of competence are central to the relationship between child sexual abuse and parenting outcomes.
Prior studies have recruited participants from clinical settings, which may produce bias toward participants with heightened mental health concerns. It may be that the use of a community sample allowed for a wider range of symptoms and experiences within the sample. Maternal CSA survivors in this sample may have already sought mental health support prior to the study and/or experienced lower levels of depressive symptoms and negative internal beliefs than maternal CSA survivors recruited from mental health facilities and parenting programs (Craig & Sprang, 2007; DiLillo et al., 2000; Douglas, 2000; Fitzgerald, Shipman, Jackson, McMahon, & Hanley, 2005; Ruscio, 2001). Additionally, past research has been plagued by low statistical power. It may be that past research was influenced by type one error and that the null hypothesis was rejected when it was, in fact, true (DiLillo & Damashek, 2003). This study had statistical power over .80, making type one error less likely.
In this sample, the means of depressive symptoms in maternal CSA survivors were consistently higher across levels of depression: clinical depression, moderate depression, and no clinical depression. Prior research has consistently indicated that child sexual abuse is significantly associated with increased depressive symptoms (Neumann, Houskamp, Pollock, & Briere, 1996). People interpret social behavior and attach meanings to their experiences with others by how they interpret the various social behavior they encounter (Blumer, 1969). Symbolic interactionism may account for this difference in depression levels in that mothers who have experienced child sexual abuse may have made meanings about themselves and their experiences that are more hopeless than mothers without histories of CSA.
Mothers with a history of CSA had lower levels of self-perceived parenting competence than mothers without CSA histories. Prior studies have indicated a significant association between CSA and parenting sense of competence (Fitzgerald et al., 2005; Zuravin & Fontanella, 1999). For example, Fitzgerald et al. (2005) found a significant relationship between a history of incest and parenting sense of competence. However, behavioral observations indicated that survivors’ interactions with their children were positive overall and comparable to those of mothers not reporting incest histories. Symbolic interactionism, and specifically the concept of role taking, can help explain these findings. Symbolic interactionism asserts that individuals are actors in their world by trying to behave in the ways that others see themselves (White & Klein, 2008). Additionally, the theory posits that people recognize themselves as both the subject and object of their behavior by looking at their behavior as an object and take on the role of the other as a subject (Blumer, 1969). How we acquire the symbols, attitudes, and beliefs of our culture is attributed to a process of socialization (Blumer). For example, a CSA survivor’s ability to be aware of how her actions impact herself (role take) may have been impaired by virtue of her perhaps frequent losses of autonomy during experiences of abuse, and thus as a parent she fails to connect to how her actions impact herself and her children.
Increased levels of depression and lower levels of self-perceived parenting competence were associated with increased levels of parenting stress and maltreatment behavior. Past research indicates that depression is associated with parenting outcomes (DiLillo & Damashek, 2003; Noll, 2008). However, this study is the first study to use parenting sense of competence (a belief) as a predictor of parenting stress and child maltreatment (indicators of behavior). Symbolic interaction provides a framework through which behavior can be explained by the generation of meaning (White & Klein, 2008). This research question was theoretically based on the influence of survivors’ meaning making (depression and parenting sense of competence) on their self-reported behavior (parenting stress and maltreatment behavior). In this sample, negative parenting outcomes were associated with specific negative beliefs.
Results indicated no direct relationship between CSA and the parenting outcomes, parenting stress or maltreatment behavior. Yet, the mediation variables, maternal depression and parenting sense of competence, were significantly associated with both CSA and the outcome variables. These results underscore that the pathways for increased risk in parenting outcomes for CSA survivors may be indirect and associated with beliefs of their own sense of competence and depression, such as hopelessness, opposed to any direct association with the sexual abuse itself.
Although mediation of depression and parenting sense of competence was not supported, results indicate that CSA was associated with higher depression and lower parenting sense of competence and that high depression and low parenting sense of competence were both associated with higher parenting stress and reported maltreatment behavior. These associations indicate that clinical assessment and intervention of adult CSA survivors must extend to the beliefs associated with survivors’ interpersonal relationships with their children.
Although this study was designed to address several of the limitations of previous research such as insufficient statistical power, lack of meditational models to test indirect effects, and minimal theoretical basis for research hypotheses, there remain limitations inherent to this study. First, the use of a convenience sample was a limitation. One of the primary weaknesses associated with convenience sampling is the potential for a non-generalizable sample. Census data were reviewed to confirm the generalizability of this sample in relation to the location of data collection. Second, sexual abuse was a dichotomous variable in the PAWSAIC dataset and therefore is a limitation of this study. To accurately and more specifically determine associations between CSA and parenting outcomes, it is important to use measures of CSA that yield continuous scores for the different commonly accepted subtypes of CSA such as severity and duration of abuse and relationship to perpetrator (Newcomb & Locke, 2001). However, the PAWSAIC dataset provided minimal information regarding survivors’ relationships to the perpetrator, severity of abuse, or duration of abuse. Prior access to mental health intervention was not assessed in original data collection, and thus was not controlled for in statistical analysis. Participant use of mental health interventions and resources could moderate the influence of CSA on parenting outcomes. Finally, this study did not include paternal survivors of child sexual abuse because of the limitations associated with the use of secondary data. A considerable void in the literature is the absence of sexually abused fathers, given the integral role that many fathers play in parenting (DiLillo & Damashek, 2003) and the high rate at which boys are sexually abused (1 in 6; Centers for Disease Control and Prevention, 2010).
The results of this study are meaningful for clinical intervention and practice. The lack of support for the hypothesis that mothers with a history of CSA would have significantly higher levels of parenting stress and maltreatment behaviors than mothers without CSA histories is a source of hope for both maternal CSA survivors and those who treat survivors of CSA. This indicates that the relationship between CSA and parenting outcomes is an indirect function of internal belief systems. Given that belief systems are associated with negative parenting behaviors, these beliefs could place survivors’ children at an increased risk of victimization (DiLillo & Damashek, 2003). Thus, intervention for maternal survivors of CSA must be focused on the beliefs that survivors have about themselves, their children, and the world in which they live, in addition to addressing presenting symptoms associated with CSA. Intervention centered on changing belief systems may be the catalyst for preventing the transmission of abuse from one generation to the next.
Because the results of this study do indicate strong relationships between beliefs associated with depression and parenting sense of competence to parenting stress and maltreatment behavior, it is important to have interventions that specifically assess and address the beliefs survivors have about themselves and their interpersonal relationships, including their beliefs about parenting. One clinical intervention designed for adult survivors with an interpersonal focus is emotion focused couple therapy for trauma survivors. Emotionally focused couple therapy (EFT) theory asserts that couple therapy plays a critical role in addressing the interpersonal effects of trauma, such as child sexual abuse (Johnson, 2002).
In a study exploring the use of EFT with couples, in which a minimum of one partner was a survivor of CSA, 50% of the CSA survivors in the study reported clinically significant decreases in trauma symptoms (MacIntosh & Johnson, 2008). Thematic analysis in this study elucidated areas where trauma survivors struggled to engage in the therapy process, such as beliefs related to feelings of shame and hyper vigilance (MacIntosh & Johnson). Although this intervention warrants further study for its use with CSA, emotion focused couple therapy shows promise in both couple relationship satisfaction and decreases in trauma symptoms for CSA survivors and their partners (MacIntosh & Johnson, 2008). Nevertheless, couple therapy may not fully address some of the parenting beliefs that arise for CSA survivors, particularly survivors of parental incest. Thus, interventions focused beyond the individual survivor and her partner to her relationship with future and current children are needed to promote resilience in survivors and prevent potentially harmful symptoms of CSA from negatively impacting survivors, partners, and their children.
Results of this study may suggest that as therapists begin treatment with maternal survivors of sexual abuse, they must not only validate feelings and address presenting symptoms, but also assess survivors’ beliefs about themselves, their beliefs and expectations of childhood behavior, and their feelings about their children. A strengths based clinical intervention is a critical component of resilience and empowerment (Cowger, 1997). However, results indicate that intervention must extend beyond empowerment, to prevention via assessment of deleterious beliefs, and ultimately intervention for harmful behaviors.
Group therapy is another promising intervention for survivors of child sexual abuse in relation to parenting outcomes. Hiebert-Murphy and Richert (2000) found that maternal survivors of CSA reported increased parenting self-esteem and improvement in attitudes toward their children following a solution-focused 12-week group intervention. CSA survivor groups could be developed to address a number of long-term symptoms associated with CSA, including negative parenting outcomes such as decreased sense of competence and increased maltreatment behavior (DiLillo & Damashek, 2003). The use of treatment protocols that extend beyond basic parenting training to encourage insight about the ways in which survivors histories of CSA and other related family-of-origin issues may be related to current parenting issues, and challenge beliefs associated with experiences of CSA could provide both attitudinal and behavior change.
Future research could address some of the aforementioned limitations of this and past studies. Measuring type, severity, duration, relationship to perpetrator, age and gender of the perpetrator, and age of onset, for example, could further reveal abuse related factors associated with an increased risk of negative parenting beliefs and behaviors in maternal survivors of CSA. Although most agree that certain actions constitute child sexual abuse, such as a father-child rape, there is little consensus regarding the validity of labeling other sexual activities as child sexual abuse, such as sexual intercourse between a 15-year-old girl and her 21-year-old boyfriend (DiLillo & Damashek, 2003). Definitions of child sexual abuse could be further clarified by assigning all women abused after the age of 18 to an “adult sexual abuse” group, assigning all women who experience date rape or statutory rape in adolescence in an “adolescent sexual abuse” group, and all women who experienced at least one contact episode of sexual abuse before the age of 18, perpetrated by someone in a position of power as indicated by age, position, type of relationship, or use of force during the sexual abuse in a “child sexual abuse” group. Clear definitions of sexual abuse may help reveal symptoms associated with sexual abuse at qualitatively different developmental ages and improve assessment and intervention. Research could be also improved by increasing sample size to accommodate the inclusion of moderating and/or mediating variables. In addition to the internal beliefs that mediate the relationship between CSA and parenting outcomes, there may be external processes such as, family support, job stability, and marital quality that moderate the relationship of CSA on parenting outcomes. The use of multiple respondents to better assess behavioral processes would also strengthen existing research. For example, the use of co-parent observations could improve the discriminant validity of parenting outcomes (Jacobson & Moore, 1981). Finally, research could use observational data of mothers’ behavior with their children. Self-report measures do not adequately assess family processes such as family power or parent–child emotional support; observational research allows for the collection of reliable, valid data unavailable through traditional assessments (Markman & Notarius, 1987).
Depression and parenting sense of competence influence the relationship between history of child sexual abuse and parenting stress and maltreatment behavior, controlling for child age. After controlling for the levels of depression, the relationship between CSA and parenting stress and maltreatment behavior is not significant. The relationship between CSA and parenting sense of competence is significant for clinically depressed participants, but not significant for moderately depressed or sub-clinical participants. These results highlight that the pathways for increased risk in parenting outcomes for child sexual abuse survivors may be indirect and associated with beliefs of their own sense of competence and depression as opposed to a direct association with sexual abuse itself.
- 1995). Parenting Stress Index: Professional manual. Odessa, FL: Psychological Assessment Resources. (
- 1997). AMOS users’ guide (version 4.0). Chicago: SmallWaters. (
- 2009). Multiple child maltreatment recurrence relative to single recurrence and no recurrence. Children and Youth Services Review, 31, 617–624. , , & (
- 1997). The impact of childhood sexual abuse and familyfunctioning on four dimensions of women’s later parenting. Child Abuse 21, 1095–1107. (
- 2003). The impact of complex trauma and depression on parenting: An exploration of risk and protective factors. Child Maltreatment, 8, 334–349. , , & (
- 1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. , & (
- 1998). Parenting among women sexually abused in childhood. Final report submitted to the Department of Health and Human Services. Grant #90 CA-1544. (
- 1969). Symbolic Interactionism: Perspective and method. Engelwood Cliffs, NJ: Prentice-Hall. (
- 2006). “Speaking of mothers…” how does the literature portray mothers who have a history of child sexual abuse. Journal of Child Sexual Abuse, 15(2), 57–74. (
- 1992). Methodological issues in the study of sexual abuse effects. Journal of Consulting and Clinical Psychology, 60, 196–203. (
- 1992). Childhood history of abuse and child abuse screening. Child Abuse and Neglect, 16, 647–659. , & (
- Centers for Disease Control and Prevention. (2010). Adverse Childhood Experiences. Retrieved January 10, 2010, from http://www.cdc.gov/nccdphp/ace/prevalence.htm.
- 1992). A power primer. Psychological Bulletin, 112, 155–159. (
- 2009). A prospective study of outcome and predictors of subclinical and clinical depression in an older biracial sample of psychiatric outpatients. Journal of Affective Disorders, 6, 460–468. , , & (
- 1989). Incest survivors: The relation of their perceptions of their parents and their own parenting attitudes. Child Abuse and Neglect, 13, 409–416. , & (
- 1992). Parenting difficulties among adult survivors of father-daughter incest. Child Abuse and Neglect, 16, 239–249. , , , & (
- 1997). Assessing client’s strengths: Assessment for client empowerment. In D. Saleebey (Ed.), The strengths perspective in social work practice (2nd ed., pp. 59–73). White Plains, NY: Longman. (
- 2007). Trauma exposure and child abuse potential: Investigating the cycle of violence. American Journal of Orthopsychiatry, 77, 296–305. , & (
- 2004). Psychosocial factors associate with types of maltreatment. Child Welfare, 83, 69–99. (
- 2003). Parenting characteristics of women reporting a history of childhood sexual abuse. Child Maltreatment, 8, 319–333. , & (
- 2000). Linking childhood sexual abuse and abusive parenting: The mediating role of maternal anger. Child Abuse and Neglect, 24, 767–779. , , & (
- 2000). Reported anxieties concerning intimate parenting in women sexually abused as children. Child Abuse and Neglect, 24, 425–434. (
- 1982). The role of age in the relationship of gender and marital status to depression. Journal of Nervous and Mental Disease, 170, 536–543. (
- 2005). Perceptions of parenting versus parent-child interactions among incest survivors. Child Abuse and Neglect, 29, 661–681. , , , , & (
- 1999). Recurrence of maltreatment: An application of the national child abuse and neglect data system (NCANDS). Child Abuse and Neglect, 23, 633–650. , , & (
- 1997). Intimate violence in families (3rd ed.). Thousand Oaks, CA: Sage. (
- 1978). Development and Utility of the Parenting Sense of Competence Scale. Paper presented at the meeting of the American Psychological Association. Toronto. , & (
- 2007). A meta analytic investigation of therapy modality outcomes for sexually abused children and adolescents: An exploratory study. Child Abuse and Neglect, 31, 125–141. , , & (
- 2000). A parenting group for women dealing with child sexual abuse and substance abuse. International Journal of Group Psychotherapy, 50, 397–405. , & (
- 2002). Evaluation of a clinic-based parent education program to reduce the risk of infant and toddler maltreatment. Public Health Nursing, 19, 377–389. (
- 1999). Predicting abuse prone parental attitudes and discipline practices in a nationally representative sample. Child Abuse and Neglect, 23, 15–29. , , , , , , et al. (
- 1981). Spouses as observers of the events in their relationship. Journal of Consulting and Clinical Psychology, 49, 269–277. , & (
- 2002). Emotionally focused couple therapy with trauma survivors: Strengthening attachment bonds. New York: The Guilford Press. (
- 1989). A measure of parenting satisfaction and efficacy. Journal of Clinical Child Psychology, 18, 167–175. , & (
- 2010). Measuring Model Fit. Retrieved January 22, 2010, from http://davidakenny.net/cm/fit.htm. (
- 1998). Data analysis in social psychology. In D. Gilbert, S. Fiske & G. Lindzey (Eds.), The handbook of social psychology (Vol. 1, 4th ed., pp. 233–265). Boston, MA: McGraw Hill. , , & (
- 2005). Principles and practice of structural equation modeling (2nd ed.). New York: The Guilford Press. (
- 2008). Childhood abuse and later parenting outcomes in two American Indian tribes. Child Abuse and Neglect, 32, 195–211. , , , , & (
- 2008). Emotionally focused therapy for couples and child sexual abuse survivors. Journal of Marital and Family Therapy, 34, 298–315. , & (
- 2007). Mediation analysis. Annual Review of Psychology, 58, 593–614. , , & (
- 2006). The effects of sexual abuse as a child on the risk of mothers physically abusing their children: A path analysis using systems theory. Child Abuse and Neglect, 30, 1293–1310. (
- 1987). Coding marital and family interaction: Current status. In T. Jacob (Ed.), Family interaction and psychopathology: Theories, methods, and findings (pp. 329–338). New York: Plenum Press. , & (
- 2002). Association between self-reported childhood sexual abuse and adverse psychosocial outcomes: Results from a twin study. Archives of General Psychiatry, 59, 139–145. , , , , , , et al. (
- 1996). The long term sequelae of childhood sexual abuse in women: A meta-analytic review. Child Maltreatment, 1, 6–16. , , , & (
- 2001). Intergenerational cycle of maltreatment: A popular concept obscured by methodological limitations. Child Abuse & Neglect, 25, 1219–1240. , & (
- 2008). Sexual abuse of children – unique in its effects on development? Child Abuse and Neglect, 32, 603–605. (
- 1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. (
- 2001). Predicting the child-rearing practices of mothers sexually abused in childhood. Child Abuse and Neglect, 25, 369–587. (
- 1983). The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female children. Child Abuse and Neglect, 7, 133–146. (
- 2005). The relationship between sexual abuse during childhood and parenting outcomes: Modeling direct and indirect pathways. Child Abuse and Neglect, 29, 645–659. , & (
- 2007). Low-income rural women and depression: Factors associated with self-reporting. American Journal of Health Behavior, 31, 657–666. , , & (
- 1990). Test scores and sociodemographic information as predictors of child abuse potential scores in young female adults. Journal of Social Behavior and Personality, 5, 199–208. , , , & (
- 2008). Family Theories (3rd ed.). Los Angeles: Sage Publications. , & (
- 1987). Child abuse: Implications for child development and psychopathology. Newbury Park, CA: Sage. (
- World Health Organization. (2008). Child abuse and neglect 2006. Retrieved February 7, 2008, from http://injury_prevention/violence/neglect/en.
- 1999). Parenting behaviors and perceived parenting competence of child sexual abuse survivors. Child Abuse and Neglect, 23, 623–632. , & (