CRAP I hate applying for funding. For the Ontario Graduate Scholarship, we basically have to write a proposal detailing our research focus for the next four years (PhD). The problem is, we’re not even supposed to pick an area of emphasis until this coming April. But it’s a good thing, I think. It’s forcing me to actually think about exactly what I want to do with my philosophy degree.
What I’m really interested in, thanks to my thesis work, are the philosophical ramifications of assessing model fit, particularly in structural equation modeling (‘cause it’s what I’m most familiar with now), but also in things like factor analysis and regression. What are the best methods to determine appropriate model fit? Should a fit index show better fit for a model with two factors when the factors are nearly orthogonal but the observable variables are all somewhat equally correlated, or should a fit index show better fit for a model with two factors that are more closely correlated but whose observable variables are more correlationally (is that a word?) separated between the two factors? What components of a model should weigh most heavily when determining model fit? Is there an “ideal” index in that sense?
It might sound weird or obscure or pedantic or whatever, but it’s interesting to me. And I think it’s very important that we start looking at the philosophical side of statistics now that we’ve got the software to run mega simulations and Monte Carlos and number orgies and sexy graphs and…
Sounds like a not often looked at area of study. Go for it, sounds fun.