SPSS test for analysing likert scale data and categorical data

#1
Hello :)

I have recently begun analysing some data that had to be collected in a certain way, where participants had to give ratings for a set of images and then had to categorise the image.

I have collected and input data from a survey of two parts 1) a response scale (1-7) like a likert scale recording rating responses to an image, and 2) categorical data regarding the same images (whether the participants consider stimuli models in-group or out-group).

So far in SPSS i have the rating responses and each of the categorical questions that correspond to them, so for each image I have a variable showing the participant's ratings (1-7 ordinal scale data) and a variable for each image for participants answers to categorisation question (two categories, in or out-group, 1 or 2).

For the anaylsis I am trying to discern whether there is a significant difference between the way participants rated images of models that they considered in or out-group.

However, I am struggling to find an appropriate test for this style of data. I have looked into doing a Kruskal-Wallis test on my data, but SPSS will only allow one grouping variable, but the grouping of my participants is defined by their response to the categorical question for each stimulus, so I have a grouping variable for each image.

Does anyone know if any of the tests on SPSS are appropriate for this kind of data?

Many thanks!
 

CB

Super Moderator
#2
Are you trying to run an individual analysis for each individual image, or are you wanting to include all the images in one overarching analysis?

(I suspect it's the latter, just wanting to clarify)
 
#3
Well since the images are 50% happy images and 50% sad images, I wanted to see if the categorical variables (in/out-group answers) imapact on the answers given for the rating variables, so I was looking to do an over-arching analysis on all of the data.

I know that I can reformat my data into these different conditions (happy in-group responses, sad in-group responses, happy out-group responses and sad out-group responses) to compare the mean scores of each, but then I would lose the individual identity of each participant and the ability to analyse for other effects, like gender or age.

So I suppose the problem is getting the categorical data to correspond to the rating data in the program, so SPSS knows whether each rating given was considered to be an in/out-group model in the image... if that makes any sense :eek: