Hi,
I am now conducting binary logistic models to predict what factors can affect whether people get depressed (the data is from BCS70). Get confused on the results.
I try to add explanatory variables gradually. First, the result illustrate that if respondents' mother had a degree when respondents were age 5 then they are less likely to get depressed at age 42. Then, when respondents' own educational factors were added, mother's education is not significant (I can understand this change, maybe because mother's education is indirect effect). However, after I add respondents' social class, neither mother's or respondents' own education factor is significant any more. But social class factor is also not significant. And I cannot understand the causal logic underlie this change.
If someone know the answer, give me some inspiration.
Many thanks~