What does a chi-square test mean when no p-value is returned?

Question

I did multinomial logistic regression using SPSS chi-square is .000 , df is 0 and significance =.

So what does it mean significance = .?

What is the exact message in the output? If it has a number between 0 and 1 after the equal sign then it is probably giveing a p-value. If it says yes or no it could be comparing the p-value to a given significance level like 0.05. But if it literally says "chi-square =.000, df is 0 and significance =." then something must be wrong. The chi-square statistic would always be >0 the df would be greater than 0 and there would be a p-value betwen 0 and 1. One thing that could go wrong would be a perfect fit giving no error variance.
That could give you 0 for the chi square statistic, and no room for degrees of freedom for error. These things happen when you have too many parameters. For multiple linear regression this happens when the number of data points is exactly equal to the number of parameters being estimated. So for a simple straight line if you have two data points you can choose the slope and intercept so that they go through the two points leaving no error.

Henry · Answer 1 · 2012-09-14 20:47:37Z

up vote 7 down vote

It means that SPSS was unable to do the calculation, roughly equivalent to trying to calculate $\frac{0}{0}$.

"df is $0$" is an indication that your regression model may have too many explanatory variables.

answered Sep 14 '12 at 20:47

Henry
8,2551720

+1 exactly what I thought – Michael Chernick Sep 15 '12 at 1:27

1 Answer