0
$\begingroup$

I am comparing measurements of 12 different groups to one another. My distributions are not normal and my Levene's test is significant - even after data transformation. I am therefore doing a Kruskal-Wallis instead of a one way ANOVA.

I have done the Kruskal-Wallis and the pair-wise comparisons available in the latest version of SPSS. They all look fine and seem to match the plots I have made. My concern is the output - the Chi squared value is huge.

As an example - I am comparing all of the lengths to each other and get the following:

Chi-Squared: 624.453 DF: 11 Asymp sig: .000

The widths gives a similar Chi Squared value of just over 400. These values seem very large - are they wrong? There is quite a lot of variation in my groups, but not an absurd amount. Some range from 1-1000 nanometers, other groups hit the 3-4000 nanometers. Is this why my Chi-squared value is so high? I am not even sure if it should be a concern to me?

This is for my final year project and SPSS is all I can use.

$\endgroup$
  • $\begingroup$ We can't comment precisely on what might be right or wrong without seeing your data (and possibly also seeing the exact syntax you used). But there is nothing inherently wrong in such chi-square values. Indeed, high-looking chi-square and very low values for significance are also compatible with a high degree of overlap between groups. (No need or gain in your telling us you don't like statistics: that is more likely to repel answers than to attract them. I edited extra wording out of the question.) $\endgroup$ – Nick Cox Jan 6 '16 at 15:15
  • $\begingroup$ Thanks, Nick. You're right - I think 10 hours of confusing statistics provoked my hasty and harsh statement. Thanks very much for taking the time to answer, and for simplifying my post. Your point about overlapping between groups is a likely answer, from looking at my data. Thanks very much again. Carley $\endgroup$ – carley Jan 6 '16 at 15:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.