1. Nancy Eisenberg and colleagues (2001) conducted research to determine the effects of mothers’ emotions on their young children’s behavior problems and social competence. One interest in the research was whether these effects are mediated by children’s own emotional regulation. Figure 18.23 shows a model patterned after those in the article. It includes fewer variables but still includes many interesting aspects of the original research. Mother’s Positive Expressivity represents mothers’ expression of positive emotions with their children, both as rated by the mothers and as observed
during their work on a task. Child’s Regulation, Externalizing, and Social Competence are latent variables representing these child characteristics, and each was rated both by the mothers and by teachers. The model includes correlated errors between the child ratings by the mother and the child ratings by the teacher. The model is contained in the Amos file “Eisenberg et al 1.amw.” Simulated data designed to mimic the relevant portions of the correlation matrix presented in the article are contained within the
Excel file “Eisenberg et al 2001.xls” and the SPSS file “Eisenberg et al 2001.sav” on the accompanying Web site. The variable names in the data and the corresponding labels from the model are shown in Table 18.4.
a. Estimate the model as shown. Focus on the fit indexes and, if you judge them to be adequate, interpret the model.
b. Estimate a model without the correlated errors. What happens to the fit of the model? Were these parameters justified?
c. Compare the fit of this model with one in which Mother’s Expressivity is also allowed to have direct effects on the two child outcomes. Based on the change in fit, would you say that children’s Regulation completely or partially mediates the effect of Mother’s Expressivity on Behavior Problems and Social Competence?
d. Calculate and interpret direct, indirect, and total effects for your accepted model.
e. Test any additional alternative models that are of interest.
