Child maltreatment in Kenya, Zambia, and the Netherlands : a cross-cultural comparison of prevalence,
psychopathological sequelae, and mediation by PTSS
Mbagaya, C.V.
Citation
Mbagaya, C. V. (2010, December 1). Child maltreatment in Kenya, Zambia, and the Netherlands : a cross-cultural comparison of prevalence,
psychopathological sequelae, and mediation by PTSS. Retrieved from https://hdl.handle.net/1887/16193
Version: Not Applicable (or Unknown)
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/16193
Note: To cite this publication please use the final published version (if applicable).
CHAPTER 8
Similarity in Diversity or Divergence in Diversity:
Comparing results of the Kenyan, Zambian
and Dutch Studies
Similarity in Diversity or Divergence in Diversity
117
8.0. Introduction
In this chapter we present comparative results of our study on child maltreatment in Kenya, Zambia and The Netherlands. We selected female respondents in Kenya and Zambia in order to effectively compare them to the all-female Dutch sample.
We compared mean scores on age, social desirability, child maltreatment variables and PTSS. In addition, we present results of the multinomial logistic regressions comparing Kenya, Zambia and The Netherlands on child maltreatment variables.
We also compare the prediction of PTSS by the child maltreatment variables in the three samples.
8.1. Descriptives
Table 8.1 presents the means and standard deviations of age, social desirability, childhood maltreatment variables and posttraumatic stress symptoms (PTSS) of the three countries and the total group. The students from the three countries did not differ in age. Dutch students had higher scores on social desirability than Kenyan and Zambian students, F (2, 572) = 14.85, p <.01. The depressive symptoms scores were significantly lower in The Netherlands sample than in the Zambian and Kenyan samples, F (2, 572) = 11.83, p <.01. The child maltreatment variables were also significantly different among the three countries. Physical abuse, history of neglect and witnessing interparental violence scores were lower in The Netherlands sample than in both the Zambian and Kenyan samples, F (2, 572) = 130.61, p <.01; F (2, 572) = 15.89, p <.01; and F (2, 572) = 30.85, p <.01, respectively. Sexual abuse scores varied significantly among the three groups with The Netherlands having lower scores than Zambia and Kenya, and Zambia having higher scores than Kenya, F (2, 572) = 30.85, p <.01. Similarly, there were significant differences in the PTSS scores among the three groups. The students from The Netherlands reported lower PTSS scores than did those from Zambia and Kenya F (2, 572) = 184.91, p <.01.
8.2. Bivariate associations between child maltreatment variables, cova- riates and PTSS
Tables 8.2, 8.3, 8.4, 8.5 present the bivariate associations between age, social desirability, childhood maltreatment variables and PTSS within the total group, the Dutch sample, the Zambian sample and the Kenyan sample. Most of the correlations within the total group were significant except for the associations between age on the one hand and social desirability, depressive symptoms and sexual abuse on the other hand (Table 8.2). The association between age and the child maltreatment variables was stronger in The Netherlands sample (Table 8.3), where age showed significant correlations of r = .13 with physical abuse and r = .11 for sexual abuse. Similarly, the correlations between age and a history of neglect (r = .23) and between age and witnessing violence, (r = .16) were significant. These
Chapter 8
118
associations indicate that in The Netherlands, older respondents were more likely to report the various forms of abuse than were younger respondents. In the Zambian sample, age was significantly associated with PTSS, r = .11 (Table 8.4) with older respondents reporting more PTSS. The correlations between the various types of child maltreatment within the total group were mostly significant (Table 8.2). This means that those who reported one type of abuse were likely to also report another type of abuse. There were variations however within the different groups. For instance, among the Dutch students, childhood sexual abuse did not have significant associations with a history of neglect; childhood physical abuse and with witnessing interparental violence. The Zambian sample did not show significant associations between a history of neglect and sexual abuse and a history of neglect and witnessing interparental violence. Among Kenyan students (see Table 8.5) this association was significant for all except for the association between physical abuse and a history of neglect. Within the total group, social desirability was negatively significantly related to all the variables in the total group except physical abuse during childhood. Thus, the higher the respondents scored on social desirability, the less they scored on depressive symptoms, PTSS, childhood sexual abuse, childhood neglect and witnessing interparental violence. Among the Zambian students however, the association between social desirability and the various variables was only significant with neglect (r = .31). This means that the higher the Zambian students scored on social desirability the lower they scored on a history of neglect. Virtually all childhood maltreatment variables had significant associations with PTSS, both in the total group and within the three samples. There was one exception, for the Zambian sample, in which there was no association between PTSS and neglect. With this exception, the higher the respondents scored on the child maltreatment variables, the higher they scored on PTSS.
Table 8.1
Differences in Child Maltreatment Variables and Covariates between Dutch, Zambian and Kenyan Students
Total (N=573)
Netherlands (NL) (n=305)
Zambia(Z) (n=86)
Kenya(K)
(n= 182) F Country differences
M SD M SD M SD
Physical abuse 1.50 .71 1.12 0.42 1.83 0.75 2.0 .74 130.61** NL<Z/K Sexual abuse 1.31 0.52 1.21 0.39 1.55 0.68 1.38 .57 17.25** NL<K<Z Neglect 1.47 0.40 1.39 0.37 1.62 0.39 1.54 .42 15.89** NL<Z/K Witnessing violence 1.33 0.77 1.12 0.53 1.73 1.00 1.50 .87 30.85** NL<Z/K PTSS 2.09 .64 1.71 .49 2.55 .50 2.51 .53 184.91** NL<Z/K Depression 1.80 0.48 1.71 0.46 2.00 0.46 1.86 .48 11.83** NL<Z/K Social desirability 2.57 0.44 2.67 0.38 2.52 0.36 2.45 .52 14.85** NL>Z/K
Age 4.10 1.69 3.98 1.92 4.34 1.68 4.19 1.19 1.88
**p < .01
Similarity in Diversity or Divergence in Diversity
119 Table 8.2
Bivariate Correlations between Child Maltreatment Variables, Covariates and PTSS for the Total Group
Age Social desirability
Depressive symptoms
Physical abuse
Sexual
abuse Neglect Witnessing violence Social desirability .02
Depressive symptoms .02 -.36**
Physical abuse .09* -.06 .28**
Sexual abuse .06 -.17** .28** .25**
Neglect .17** -.10* .38** .28** .21**
Witnessing violence .09* -.16** .19** .47** .16** .20**
PTSS .05** -.37** .47** .53** .39** .31** .32**
*p < .05 **p < .01 Table 8.3
Bivariate Correlations between Child Maltreatment Variables, Covariates and PTSS for The Netherlands
Age Social desirability
Depressive symptoms
Physical abuse
Sexual
abuse Neglect Witnessing violence Social desirability .03
Depressive symptoms .07 -.41**
Physical abuse .13* -.06 .19**
Sexual abuse .11* -.02 .12* -.02
Neglect .23** -.13* .36** .27** .07
Witnessing violence .16** -.08 .18** .62** -.01 .19**
PTSS .08 -.24** .47** .25** .30** .34** .20**
*p <.05 **p < .01 Table 8.4
Bivariate Correlations between Child Maltreatment Variables, Covariates and PTSS for Zambia
Age Social desirability
Depressive symptoms
Physical abuse
Sexual
abuse Neglect Witnessing violence Social desirability .01
Depressive symptoms -.15 -.31**
Physical abuse .11 .32 .34**
Sexual abuse .13 -.17 .37** .22*
Neglect .03 .01 .30** .31** .11
Witnessing violence -.03 -.15 .10 .43** .08 .03
PTSS .22* .09 .32** .34** .38** .00 .16
*p <.05 **p < .01
Chapter 8
120 Table 8.5
Bivariate Correlations between Child Maltreatment Variables, Covariates and PTSS for Kenya
Age Social
desirability Depressive
symptoms Physical abuse Sexual
abuse Neglect Witnessing violence Social desirability .07
Depressive symptoms .10 -.27**
Physical abuse -.05 -.25** .23**
Sexual abuse -.14 -.23** .32** .25**
Neglect .07 -.01 .36** .10 .29**
Witnessing violence .02 -.11 .12 .26** .18* .17*
PTSS -.06 -.45** .51** .29** .35** .16* .17*
*p <.05 **p < .01
8.3. Multinomial logistic regression predicting child maltreatment in Kenya and Zambia with The Netherlands as the reference category Table 8.6 shows the results of the multinomial logistic regression to assess the prediction of child maltreatment (physical abuse, sexual abuse, neglect, and witnessing interparental violence) in Zambia and Kenya with The Netherlands as the reference category. Age, social desirability and depressive symptoms were first entered in the regression as control background variables and also because of their likelihood to influence answering patterns. Similar analyses were also conducted with Kenya as the reference category in order to determine the prediction of child maltreatment variables in Zambia (see Table 8.6). In doing this, we were able to determine the country in which the reported level of the various types of maltreatment was most likely to be significantly different from the reference category (country). In all analyses, the control variables were nonsignificant. The results showed that the reports of physical abuse, sexual abuse and witnessing interparental violence were significantly higher in Zambia than in The Netherlands. This prediction was significantly better for physical abuse (7.61, 95% CI 4.30 – 13.34, Wald = 50.19) than for sexual abuse (2.11, 95%
CI 1.27 – 3.56, Wald = 8.15). The difference in level of witnessing interparental violence was smaller between Zambia and The Netherlands, with the odds at 1.56, 95% CI 1.06 – 2.28, Wald = 5.02.
Among the four childhood maltreatment variables, the likelihood of having experienced physical abuse was higher in Kenya than in The Netherlands, odds
= 12.89, 95% CI 7.69 – 21.61, Wald = 94.21. In comparing Kenya and Zambia, the chance of being physically abused was lower in Zambia than in Kenya, (0.62, 95% CI 0.39 – 0.88, Wald = 12.88). On the other hand, the risk of sexual abuse and witnessing interparental violence was higher in Zambia than in Kenya. With the odds of sexual abuse at 1.57, 95% CI 1.01 – 2.47, Wald = 8.28, this prediction was better than the prediction of witnessing violence (1.28, 95% CI 1.07 – 1.98, Wald
= 5.93).
Similarity in Diversity or Divergence in Diversity
121 Table 8.6.
Multinomial Logistic Regression with The Netherlands as Reference Category
B SE Wald Odds (95%CI) p
The Netherlands - Zambia
Intercept -6.1 1.3 22.03 <.01
Age .06 .08 .45 1.06 (.90 – 1.25) .50
Social desirability -.18 .34 .37 .08 (.43 – 1.64) .60
Depression .01 .33 .00 1.01 (.53 – 1.94) .97
Physical abuse 2.03 .29 50.19 7.61 (4.30 – 13.34) <.01
Sexual abuse .75 .26 8.15 2.11 (1.27 – 3.56) <.01
Neglect .47 .38 1.56 1.60 (.76 – 3.36) .21
Witnessing violence .44 .20 5.02 1.56 (1.06 – 2.28) .03
The Netherlands - Kenya
Intercept -2.5 1.09 5.29 .02
Age .02 .07 .06 1.02 (.89 – 1.65) .80
Social desirability -.72 .29 6.37 .49 (.28 – .85) .01
Depression -.42 .29 2.07 .66 (.38 – 1.16) .97
Physical abuse 2.55 .26 94.21 12.89 (7.69 – 21.61) <.01
Sexual abuse .29 .25 1.32 1.34 (.81 – 2.20) .25
Neglect .32 .34 .89 1.37 (.71 – 2.64) .35
Witnessing violence .07 .19 .12 1.07 (.73 – 1.56) .73
Kenya – Zambia
Intercept -3.15 .88 12.84 <.01
Age .10 .06 2.92 1.2 (.88 – 1.23) .09
Social desirability .42 .23 3.39 1.52 (.88 – 3.33) .07
Depression .48 .23 4.09 1.60 (.83 – 2.85) .04
Physical abuse -.47 .43 12.88 .62 (.39 – .88) <.01
Sexual abuse .45 .16 8.28 1.57 (1.01 – 2.47) <.01
Neglect -.09 .24 .14 .91 (.60 – 2.28) .71
Witnessing violence .25 .10 5.93 1.28 (1.07 – 1.98) .02
Chapter 8
122
8.4. Cross-validation
In order to examine differences between countries in predicting PTSS on the basis of the various types of abuse, we conducted multivariate hierarchical regressions for each country separately with PTSS as criterion, and age, social desirability and depression as control variables, and the four types of abuse as predictors.
The results of these regressions are presented in Table 8.7. We found that for all countries the second and most important step in the regression was significant, that is, the contribution of the four types of abuse to the prediction of PTSS was significant after the inclusion of the control variables. However, for each of the countries the second step contained different predictors that significantly contributed to the equation. As expected for all three regressions sexual abuse was a significant predictor of PTSS. Witnessing of interparental violence, however, did not contribute to this prediction in any of the three countries. The role of childhood physical abuse and a history of neglect in predicting PTSS varied across countries (see Table 8.7).
In order to test whether the three countries would show similar fit of the regression model, the regression equations were cross-validated. The Netherlands regression equation was taken as the reference to which the fit in the other countries was compared. This equation of the prediction of PTSS for the Dutch sample was the following:
Y PTSS Netherlands = 0.514 -.007*age - .084 * social desirability + .341 * depression + .151 * physical abuse + .318 * sexual abuse + .205 * neglect + .026 * witnessing violence.
Within The Netherlands sample, the correlation of the predicted PTSS scores with the observed PTSS scores was r = .58. Using The Netherlands equation to predict the Zambian PTSS scores, we found a correlation of r = .44 between the predicted scores and the observed scores. Using the same equation for the Kenyan sample, the correlation between the predicted PTSS scores and the real PTSS scores of the Kenyan students was r = .55. The correlations between the predicted scores and real scores based on The Netherlands equation did not differ much from the correlations found between the predicted and real PTSS scores within each of the other countries.
In the Zambian sample the regression equation for the prediction of PTSS scores was the following:
Y PTSS Zambia = 1.795 -.081*age + .163 * social desirability + .141 * depression + .234
* physical abuse + .238 * sexual abuse - .229 * neglect - .005 * witnessing violence.
Within the Zambian sample, the correlation between the predicted and real scores was r = .58 which is not substantially different compared to the r = .44 when we used The Netherlands equation.
Similarly, the Kenyan prediction of the PTSS scores was
Y PTSS Kenya = 2.117 + .011*age - .294 * social desirability + .421 * depression + .063 * physical abuse + .133 * sexual abuse - .046 * neglect + .031 * witnessing violence.
Similarity in Diversity or Divergence in Diversity
123 The correlation within the Kenya sample between the predicted and real PTSS scores amounted to r = .63, and again compared to the r = .55 based on the regression equation of the Dutch sample, this difference is not substantial.
As the differences between the cross-validation correlations and the multiple correlations within the two countries was below .15, the equations were largely the same and the shrinkage was probably due to overfitting of the regression equation to its own sample rather than an inappropriateness of the equation from The Netherlands sample.
Within the Zambian sample, we also computed the correlation between the predicted PTSS scores based on The Netherlands equation and the predicted PTSS scores based on the Zambian equation, and this correlation was high, r = .76. For the Kenyan sample, a similar procedure led to a correlation between both predicted PTSS scores of r = .87, again a very strong association, reflecting the similarity of the predictive patterns in the three countries.
Chapter 8
124 Table 8.7.
Summary of Hierarchical Regression Analysis Predicting PTSS
B SE Beta1 T p
Constant
The Netherlands .51 .25 2.10 <.03
Zambia 1.80 .50 3.57 <.01
Kenya 2.12 .28 7.60 <.01
Total 1.08 1.91 5.67 <.01
Step 1
AgeThe Netherlands -.01 .01 -.03 -.54 .59
Zambia -.08. .03 -.27 -2.82 <.01
Kenya .01 .03 .02 .41 .69
Total .00 .01 -.01 -.28 .78
Social desirability
The Netherlands -.08 .07 -.07 -.128 .20
Zambia .16 .14 .12 1.16 .25
Kenya -.29 .07 -.29 -4.56 <.01
Total -.20 .05 -.14 -4.04 <.01
Depressive symptoms
The Netherlands .34 .06 .33 5.89 <.01
Zambia .14 .12 .13 .17 .25
Kenya .42 .08 .38 5.62 <.01
Total .33 .05 .24 6.61 <.01
Step 2
Physical abuse
The Netherlands .15 .07 .13 2.11 .04
Zambia .23 .08 .35 3.00 <.01
Kenya .06 .05 .09 1.39 .17
Total .30 .05 .24 6.61 <.01
Sexual abuse
The Netherlands .32 .06 .26 5.37 <.01
Zambia .24 .07 .33 3.20 <.01
Kenya .13 .06 .14 2.18 .03
Total .24 .04 .19 5.85 <.01
Neglect
The Netherlands .21 .07 .16 2.94 <.01
Zambia -.23 .13 -.18 -1.79 .08
Kenya -.05 .08 -.04 -.56 .58
Total .10 .03 .05 1.35 .18
Witnessing violence
The Netherlands .03 .06 .03 .46 .64
Zambia -.01 .05 -.01 -.10 .92
Kenya .03 .04 -.04 -.56 .58
Total .04 .03 .05 1.35 .17
The Netherlands (R .58; R2 .37; R2Ch .11; Fchange ; 12.26; df 4, 297)**
Zambia (R .58; R2 .33; R2Ch .20; Fchange ; 5.92; df 4, 78)**
Kenya (R .63; R2 .40; R2Ch .03; Fchange ; 2.45; df 4, 174)*
Total Group (R .67; R2 .45; R2Ch .18; Fchange ; 45.96; df 4, 565)**