# Overall significance test for the effect of an independent continuous variable on a categorical dependent variable

If I have a dependent variable having more than two categories (the categories can be ordered) and a few independent variables which are all continuous, then how can I see whether the independent variable affects the dependent variable significantly or not?

I can see that if I run an ordinal logistic regression, then compared to the reference category, a beta coefficient and its significance level is given for each of the other categories of the dependent variable. But my purpose is to comment in an overall manner whether the independent variable affects the dependent variable. For example, whether income(continuous) will affect social status(categorical) significantly. I don't want to find out in comparison to any reference category, I just want to make an overall comment.

If this is possible, then how can I find that overall significance?

-
What software are you using? Such overall tests are part of the output in SAS PROC LOGISTIC. I am sure you can also get them from R (although it's been a while since I did an ordinal logistic regression in R). –  Peter Flom Aug 5 '12 at 12:22
If you really have a single categorical outcome variable and all the other variables are continuous, then the situation you describe as undesirable does not fit. Ordinal logistic regression would place the categorical variable as the dependent, and there are no predictors that can have reference categories. So ordinal logistic regression should give you exactly what you need: a single coefficient for each predictor. –  rolando2 Aug 5 '12 at 23:05
add comment

## 1 Answer

It's very important to first define the nature of your dependent variable. If qualitative ordinal, then an ordinal probit(or logit) model is the right choice. With this model you will have a unique slope parameter per explanatory variable whatever the category as only the constant changes with categories. If your dependent variable is social status then it can be easily considered as ordinal. Thus, inference on an independent variable effect becomes straighforward.

-
add comment