(1) each of their parents having sad behavior for 2 or more weeks and (2) their parents’ drug and alcohol problems during the respondents’ childhood (defined as up to the age of 18). These analyses were conducted incorporating the weighting and design effects from the NCS-R. Control variables in this framework included their parents’ immigrant status and ethnicity and the respondent’s own age at interview and relationship status. Potential mediating variables in this framework included the respondents’ recall of the level of closeness with each of their parents, their recall of each of their parents’ social problems, their report of parental neglect during their childhoods, and their report of the experience of trauma during childhood. We statistically tested the mediational effects of these factors and found that an alternate model in which these potential mediating variables were treated as covariates provided a better fit. Hence, we evaluated the independent and potential moderating effects of these variables in multiple logistic regression analyses. In these logistic regression models, we retained independent variables that had p-values less than 0.10 and examined interactions (effect modification) among the variables that met this criterion. Interaction terms with p-values less than 0.05 were retained.

In the models for both sons and daughters, the variables that consisstently predicted major depression within the past 12 months were recalling having a father who had experienced sadness for 2 or more weeks during the respondent’s childhood and having had an experience of trauma before age 18. As shown in Table 4-1, in the multiple logistic regression model, we found that males who recalled their fathers having experienced sadness for 2 or more weeks during their childhoods had rates of major depression in the past 12 months that were modified by their fathers’ drug or alcohol problems during their childhoods (effect modification statistically significant, p = 0.01). Paradoxically, those whose fathers had experienced sadness and also had drug or alcohol problems during the respondents’ childhood had lower rates of major depression in the past 12 months compared to those who fathers had neither risk factor, though this difference was not statistically significant. In contrast, those whose fathers experienced 2 or more weeks of sadness alone and those whose fathers had drug or alcohol problems alone had elevated rates of major depression compared to those with neither risk factor, with odds ratios of 3.65 and 1.73, respectively. In addition, the summary score of paternal closeness during childhood was highly significantly associated with major depression in the past 12 months (p = 0.01), with those who perceived less closeness to their fathers more likely to have major depression. Also, males who reported having experienced trauma before the age of 18 were approximately 3.8 times more likely to have had major depression in the past year (p = 0.02).

Among females, we again found that having a father who had sadness for 2 or more weeks during the respondent’s childhood was a strong (odds



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement