I wanted to quickly reference with someone who may know more about logistic type analyses.

I am doing an analysis of a very large data set (n > 3000), and i have a categorical IV with 4 levels and a dichotomous DV (0 v 1). What i want to do is see whether there is a main effect of the IV on the DV, and whether there are any differences between levels of the IV in predicting the outcome of the DV.

I came across an old publication which states that a one way anova would be appropriate to use if the sample size is large enough and I meet certain proportional criteria (which are met in the data).

D'agostino (1971) A second look at analysis of variance on dichotomous data.
Journal of Educational Measurement, 8 (4), 327-333.

This article suggests that it may be appropriate for me to use the one way anova (vs. a logistic regression) to analyze the main effect and do possible multiple comparisons.

Does anyone have experience with these types of analyses and whether the anova would be appropriate?

If it is inappropriate to do the anova, i have a query on the interpretation of a dummy coded IV in the logit regression. As the variable has 4 categories, three dummy codes were made along with a null, which was the comparison category. In the SPSS output, is this comparison category the constant?

For example: the output may look like:


Thanks in advance!