• Keine Ergebnisse gefunden

Effect of Culture-Level Clusters on the Intention to Have Children

3. Method

4.3 Relating Family Model Value Profiles to Behavioral Intentions

4.3.3 Family Model Value Profiles and the Intention to Have Children

4.3.3.1 Effect of Culture-Level Clusters on the Intention to Have Children

In the full-factorial model the interaction of Culture-Level Cluster x Gender was significant while the interaction of Culture x Gender was non-significant. Thus, the interaction between Culture and Gender was fully explained by the interaction between Culture-Level Cluster and Gender (since Culture is nested in Culture-Level Cluster). In the next model without the previously non-significant interaction effect the interaction effect of Culture-cluster x Gender was again significant, but a slightly higher AIC resulted and the model was also non-fitting. Therefore, the full-factorial model was selected as the final model. As expected, the additional main effects model was non-fitting and showed a higher

AIC than the full factorial model (see Table 35). A further additional model controlling for adolescents’ age yielded a non-significant effect of age with Wald χ2 = 0.86 (df = 2, p = .65) and no differences resulted with regard to the other effects in the final model.

The single comparison contrasts of the main effect of Culture-Level Cluster (in the main effects model) were only partly significant (see Table 36). The Culture-Level Cluster-contrast of (total) interdependence vs. emotional interdependence had significant effects both on the proportion of adolescents who planned to have children as compared to those who did not know yet (this effect was also significant for the contrast of (total) interdependence vs.

independence); as well as on the proportion of adolescents who did not plan to have children as compared to those who did not know. The contrast of emotional interdependence vs.

independence did not have any significant effects on any contrasts of the dependent variable (see Table 36).

Table 35

Multinomial Logistic Regression for the Intention to Have Children

Model Full Factorial Reduced Main Effects

Effects Wald df P Wald Df P Wald df p

Culture-Level Cluster 35.46 4 <.001 47.77 4 <.001 44.00 4 <.001 Culture [C-L Cl.] 175.03 14 <.001 216.10 14 <.001 212.07 14 <.001

Gender 1.16 2 .56 4.98 2 <.10 4.37 2 .11

C-L Cl. x Gender 15.60 4 <.01 30.95 4 <.001 Culture x Gender [C-L Cl.] 17.34 14 .24

Criteria Model Test (LR) Model Fit (LR) Model Comparison AIC R2 (U) Chi2 Df p Chi2 df p Intercept only 551.67

Full Factorial 235.67 .192 392.01 38 <.001 saturated model

Reduced 235.81 .179 363.87 24 <.001 28.14 14 <.05 Main Effects 259.74 .164 331.93 20 <.001 60.07 18 <.001

Note. C-L Cl. = Culture-Level Cluster. The acronym in squared brackets signifies that the preceding effect is nested in the factor Culture-Level Cluster. AIC = Akaikes Information Criterion. R2 (U) = Nagelkerke’s Pseudo R2 (Reduction of Uncertainty). The bold printed model is selected as the best (and final) model according to the AIC. The factors Culture-Level Cluster, Culture, and Gender were used for the definition of subpopulations.

Table 37 shows the distribution of adolescents’ answers with regard to their Intention to Have Children by Culture, Culture-Level Cluster, and Gender. From Figure 19 it becomes clear that the differences between culture-level clusters did not explain much of the cross-cultural differences. Cross-cross-cultural differences within culture-level clusters were especially

pronounced in the cluster representing the family model of (total) interdependence (India as exception) and in the cluster representing the family model of emotional interdependence (China as exception), while cultures in the cluster representing the family model of independence were rather homogenous. The highest approval rates and the lowest disapproval rates for having children were found in the cluster representing the family model of (total) interdependence. Within this cluster, Indian adolescents were less sure about having children in the future than Indonesian and South African children (see Figure 19). Adolescents from cultures belonging to the cluster representing the family model of emotional interdependence were somewhat less sure about having children in the future (less approval and more “don’t know”) than those from the cluster representing the family model (total) interdependence, but the disapproval rate was also very low. Here, China differed greatly from the other cultures in the cluster: a high proportion of Chinese adolescents were undecided with respect to having children, and the proportion of those who did not plan to have children was twice as high as for Turkish adolescents and even four times as high as for Israeli adolescents. Though relatively few adolescents from cultures belonging to the cluster representing the family model of independence did not want to have children, this proportion was somewhat higher than in the cultures belonging to the other two culture-level clusters. Within this independence-cluster, French adolescents stand out because they reported an especially high approval rate and an especially low disapproval rate regarding having children as compared to the other cultures in the cluster.

Table 36

Intention to Have Children: Contrasts for Culture-Level Cluster

Yes vs. Don’t Know No vs. Don’t Know Yes vs. No Culture-Level Cluster B SE Wald B SE Wald B SE Wald Interdependence vs.

Emot. Interdependence 1.14 .30 14.80*** 1.26 .44 8.31** -0.12 .34 0.12 Interdependence vs.

Independence 1.22 .37 11.14*** 0.63 .51 1.55 0.59 .39 2.25 Emot. Interdependence

vs. Independence 0.08 .31 0.07 -0.63 .48 1.69 0.71 .41 2.94+

Note. Single contrast effects for the main effect of Culture-Level Cluster on the Intention to Have Children from the main effects model.

+ p < .10. * p < .05. ** p < .01. *** p < .001.

Since only the interaction between Culture-Level Cluster and Gender was significant (but not that of Culture x Gender), only this interaction will be illustrated here (see Figure 20).

In the cluster representing the family model of (total) interdependence female adolescents were somewhat more skeptical about having children than male adolescents. While gender differences in the cluster representing the family model emotional interdependence were small, there were considerable gender differences in the cluster representing the family model of independence. Here, male adolescents were more skeptical or insecure about having children than female adolescents as reflected in the lower approval and higher “don’t know”-rates for male adolescents (see Figure 20).

Table 37

Intention to Have Children: Culture, Culture-Level Cluster, and Gender Differences

Males Females All

With regard to Research Question 2.1.2 the results were partly in line with the expectations based on the theory of family change. Adolescents from cultures represented by the family model of (total) interdependence were most likely to report that they wanted to have children in the future while adolescents from the other two culture-level clusters were substantially lower in this respect. No significant differences were found for the proportion of adolescents not wanting to have children. Furthermore, no significant differences were found

between the culture-level clusters of emotional interdependence and of independence (see Table 36). The effect of Culture-Level Clusters was overridden by culture-specific effects.

Figure 19. Intention to Have Children across Cultures and Culture-Level Clusters.

Figure 20. Intention to Have Children across Culture-Level Clusters and Gender.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

No

Don’t know Yes

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Male Female Male Female Male Female

Interdependence Emotional Interdep. Independence

No

Don't know Yes

Multilevel Analysis to Estimate the Between-Culture Variance Explained by Culture-Level Cluster Membership. The results of the null model showed significant between-variance-components of the intercept of Category 1 (Yes vs. Don’t Know) with u0j (1) = 0.665, (χ2 = 261.58, df = 9, p < .001) and of the intercept of Category 2 (No vs. Don’t Know) with u0j (2)

= 0.371 (χ2 = 47.41, df = 9, p < .001), indicating significant cross-cultural differences with respect to the outcome variable. In the model including the culture-level effects of the dummy variables representing culture-level cluster membership, none of the four culture-level effects was significant (see Table C-1 in Appendix C). Nevertheless, the proportion of variance accounted for by the culture-level predictors will be reported here: the between-variance-components were reduced to u0j (1) = 0.512 (χ2 = 230.58, df = 7, p < .001) and u0j (2) = 0.306 (χ2 = 37.47, df = 7, p < .001), respectively. Thus, culture-level cluster membership accounted for 23.0 % of the between-culture variance for Category 1 [(0.665-0.512) / 0.665 = .230] and for 17.5 % of the between-culture variance for Category 2 [(0.371-0.306) / 0.371 = .175].

Thus, overall about 20 % of the between-culture variance with respect to Intention to Have Children was explained by culture-level cluster membership.