Testing if Rank Data varies significantly from 0

[milnehouse]

Hi, I've done a survey with an 11 point Likert scale (-5 to +5, with 0).There are 30 questions total in the survey, with ~800 responses.

I have graphed the median response for each survey question, but I am looking for a way to test which questions are 'significantly' different from 0 (a neutral response).

I was looking at Mann-Whitney U test, but that doesn't seem to make sense if I don't actually have 2 distributions of data (one is just 0). Is there another test that I could use?

Thanks,
Andrew

 

[milnehouse]

Reading up more on the Mann-Whitney, I see that you can do it with one sample, but there seems to be an assumption that your data is symmetrically distributed around the mean.

My data is highly skewed, does anyone know of another test that I could use?