Help analyzing nonnormal data

[psychmajor]

Hi everybody,

I'd like to perform an ANOVA or similar with data whose residuals are nonnormal. My dependent variable is 1-7, but about 30% of my respondents provided 1's and the plurality of the rest provided 2's. I've tried log and square root transformations with no luck. Any advice on how to proceed?

Basically, the distribution looks like this (with each 1 representing an interval of 10 cases):

1
1
1
1
1
1
11
11
11
11 1
11111
1111111
1111111

Gratefully,
Psychmajor

 

[hlsmith]

What is your sample size? Do you need to control for other variables? Tell us more about the context if possible.

 

[psychmajor]

Thanks for asking, hlsmith! Sample size is 415. I'm trying to conduct a 2x3 ANOVA. And to clarify my earlier approach at a graphic, that represents the raw scores, not the residuals. Please let me know what other information would be helpful, and thank you for asking me to clarify.

 

[hlsmith]

Can you provide a qqplot and scatterplot of the residuals? Could the outcome be better defined as a negative binomial regression or other count model.

the reason I asked about controlling for other variables was that the kruskal wallis test is another option.

 

[psychmajor]

I hope these are informative! I'm also going to look into a negative binomial regression--thank you for the advice!.

 

[psychmajor]

Also, I did a kruskal wallis, too, and it didn't look good. thanks!

 

[hlsmith]

How can a Kruskal Wallis test - "didn't look good"? Based on what criteria?

 

[psychmajor]

Sorry, just that the Kruskal Wallis indicated nonnormality.

 

[hlsmith]

How can K-W indicate non-normality?

Thanks!