Sent my advisor a revised Chapter 4 on Sunday. He finished editing it and I need to go pick it up today. It was 41 pages! I'm trying to get Chapter 5 worked out before the week is up: It's the chapter on the conditioning effects of depression and self-efficacy on strain, and how different levels of those variables affect the different kinds of outcomes I'm studying. The findings are interesting -- depression amplifies the effect of strain on outcomes like running away and suicidal behavior, but buffers the effect of strain on more externalized outcomes like aggressive delinquent acts and shoplifting. Those are the findings for girls. There are a couple of similarities in the boys' findings, but generally they don't hold for them. Self-efficacy seems to act as a buffer for both girls and boys, but more so for girls. Fascinating.
I had to re-run the interaction models because the first time I ran them (months ago) I didn't "mean center" the variables involved in the interaction. Apparently this is what you're supposed to do. I think it's kind of optional, but customary. It doesn't affect the results, just the interpretation of the results, and it also reduces multicollinearity between the interaction term and the variables that are multiplied together to create the interaction term. My adviser didn't think this was necessary, but I'm anticipating the comments from the rest of my committee. Anyway, so I created the new mean-centered variables for the girls and re-ran the analysis, updated the tables. Tonight I will do the same for the boys' sample, and then I can start augmenting Chapter 5 with some exposition about the findings. I need to get this DONE this week so I can focus on the conclusion -- what are the big findings, why, what do they mean, so what, etc. I will soon be going from "serious" mode into "Red Alert" mode (stole that from the blog
Reassigned Time). Which means no more staying up until 1am watching non-stop episodes of "24."