US tab

The Role of Parenting and Attachment in
Identity Style Development

By: Kaylin Ratner | Mentor: Dr. Steven Berman

Results

Hypothesis 1

To determine if parenting style was related to identity style, a Multivariate Analysis of Variance (MANOVA) was run with participant gender and parenting style as the independent variables and the identity styles as the dependent variables. Results showed that there was a significant main effect for maternal parenting style (F[9, 696] = 3.60, p < .001). The informational identity style, F(3, 232) = 4.62, p = .004, and the normative identity style, F(3, 232) = 6.12, p = .001, were found to be significantly dependent upon maternal parenting style. A Fisher's LSD post hoc analysis revealed that those with an authoritative or an indulgent mother had significantly higher informational and normative identity style scores than those with a negligent or an authoritarian mother.

Hypothesis 2

To determine if parental responsiveness and demandingness would predict their respective attachment scores, a multiple regression analysis was run with age and sex entered on Step 1, maternal responsiveness and demandingness entered on Step 2, with attachment as the dependent variable. This process was repeated for paternal prediction value. In terms of the maternal regression, results indicated that the overall model was significant, R2 = .75, Adjusted R2 = .75, F[4, 234] = 175.28, p < .001. When measuring for paternal effects, results showed that again the overall model was significant, R2 = .70, Adjusted R2 = .70, F[4, 187] = 110.51, p < .001. Results of these regressions, including changes in R2 at Step 2 and beta-weights, can be found in Table 2 (See Appendix). A one-way Analysis of Variance (ANOVA) showed a significant main effect for maternal parenting style on adolescent attachment score, F(3,238) = 79.94, p < .001. Adolescents who identified their mother as either negligent or authoritarian felt significantly less attached than adolescents who identified their mother as indulgent or authoritative (p < .001). Adolescents who reported their mothers as authoritative reported significantly higher attachment scores (p < .005) than adolescents who indicated an indulgent mother.

An ANOVA also showed a significant main effect for paternal influences on paternal attachment, F(3, 190) = 55.61, p < .001. Adolescents who reported their father as authoritative showed significantly higher attachment scores than those who claimed that their father was negligent (p = .001) or authoritarian (p = .002).

Hypothesis 3

No significant results were seen when the informational identity style and the diffuse-avoidant identity style were entered as the dependent variable; however, when the normative identity style was the dependent variable, results indicated that the overall model was significant (R2 = .32, Adjusted R2 = .10, F[4, 186] = 5.36, p < .001). At Step 2, the change in R-square was also significant (ΔF[2, 186] = 9.71, p < .001; ΔR2 = .09) with standardized beta coefficients reaching significance for maternal attachment (β = .31, t = 3.79, p < .001). No significance was found for paternal attachment and any of the identity styles.

Hypothesis 4

To test the final hypothesis, that parenting style predicts identity style development but this relationship is mediated by parental attachment, a series of multiple regression analyses were conducted. As proposed by Holmbeck (1997), three significant relationships must first be established prior to testing for mediation. Variable A (parenting style) must predict Variable B (attachment), Variable B must predict Variable C (Identity Style), Variable A must predict Variable C, and finally, Variable A and B together must be entered together to predict Variable C, but Variable A's relationship with Variable C should no longer reach significance once controlling for Variable B. That is, parental attachment should have a direct effect on identity style whereas parenting style has an indirect effect on identity style only through parental attachment. For each regression analysis, sex and age were entered on Step 1 with the appropriate predictor variable(s) entered on Step 2.

As tested in Hypothesis 2, Variable A (maternal responsiveness and demandingness) was shown to predict Variable B (maternal attachment). As tested in Hypothesis 3, Variable B (maternal attachment) was shown to predict Variable C (only the normative identity style).

For the next regression, only maternal parenting style was considered in regard to identity style due to Hypothesis 1's findings. When normative identity style was entered as the dependent variable, results indicated that the overall model was significant, R2 = .10, Adjusted R2 = .09, F[4, 236] = 6.83, p < .001. At Step 2, the change in R-square was also significant (ΔF[2, 236] = 12.90, p < .001; ΔR2 = .10) with standardized beta coefficients reaching significance for maternal responsiveness (β = .27, t = 4.27, p < .001).

In the final regression to test for mediation, Variables A and B (maternal responsiveness and maternal attachment) were entered on Step 2 to predict Variable C (the normative identity style). An overall significant model was shown, R2 = .10, Adjusted R2 = .08, F(4, 235) = 6.37, p < .001, as well as a significant change model, ΔF(2, 235) = 11.96, p < .001; ΔR2 = .09, but maternal responsiveness (β = .21, t = 1.71, p = .089) and attachment (β = .11, t = .86, p = .39) were no longer significant when entered together on this last step.

Discussion >>