Misc. Stats Topics Discussion > Significance in continuous and unordered categorical data
Don't worry folks, it's sorted. Thanks!
October 31, 2012 |
Alice
Alice
Sorry for the delay in response, Alice, but I'm glad you got it worked out. Which analysis you use depends on how you conceptualize your question. Do you want to know the impact of the IVs on the two DVs together (overall), or their impact on each DV respectively?
If the former is true, you should be able to use a multivariate general linear model (or MANCOVA). If the latter is true, you should be able to use separate univariate general linear models ( aka regressions; aka ANCOVA).
November 3, 2012 |
Jeremy Taylor
Jeremy Taylor
I've got a very complicated dataset, and I've run a few different stats tests on it but I'm not sure if they are the correct ones to use. I have 2DVs which are continuous (they are measures of carbon and nitrogen isotopes), and 5IVs which are categorical (but unordered, things like soil type). I want to know which of the IVs has the most significant impact on the DVs, taking into account the DVs interaction effects. I've tried principal components analysis, but also multinomial generalized logistic regression (creating a dummy variable?), and standard glms and gams. I'm sure half of these aren't right, and I'm completely bewildered, please help!
Thanks,
Alice
PS: I use R