How do you analyse Likert data for four paired groups?

[freckle21]

I asked a group of 90 people the same four questions about four different artworks. The group viewed the four artworks in-situ in a gallery setting experiencing the artworks one after the other in the same order. Each of the four questions asked was a Likert item, designed to form a Likert scale. Therefore, I have 90 responses to each of the four Likert scales. It is my understanding that the data is paired or dependent as all responses were generated by the same group of people. And, that this therefore means that I am unable to use the Kruskall-Wallis non-parametric data test which would normally be used to assess ordinal data for more than one group. My hypothesis is that one of the four artworks is more impactful than the others and I therefore want to compare the data from the four scales. Can anyone offer any advice?

 

[katxt]

There is Friedman's test which is nonparametric repeated measures. I don't know whether this is what you want because I'm not entirely clear as to how you have arranged things.

 

[freckle21]

I have had a quick look and this appears to fit what I need. Thanks so much. Who knew it would be this difficult!

I have a great many ties which is proving rather boring in terms of ranking. I have done these by hand. I am wondering if it would be easier to sort in SPSS?

 

[Karabiner]

What do you mean by "ranking by hand" and "sort in SPSS"? What does this refer to, and what do you want to achieve?

With kind regards

Karabiner

 

[katxt]

You have a lot of data so probably a standard repeated measures anova will work OK, even if the purists disagree.