Is Pearson's chi-sqaure test appropriate to use for Likert items? I want to find out whether there is any difference in the responses from 3 different groups?
If it is Kruskal-Wallis Test, how would I analyse the scales (means of several items combined)?
I am using SPSS.
Many thanks :)
Hi,
Question: Appropriate statistical test for Likert items and scales to find out if there is any significant difference in the responses of 3 different groups?
For your data analysis you can use Mann Whitney test or Kruskal Wallis test.
Kindly refer the following:
https://www.st-ndrews.ac.uk/media/capod/students/mathssupport/OrdinalexampleR.pdf
Hello Shane,
I would like to agree with the previous response, but in order to run Mann Whitney test or Kruskal Wallis test - make sure to test for normality. If you do not know how, there are many videos on you tube that tells you how to do using spss, for example.
If your groups are normally distributed your could run an ANOVA because you have 3 independent groups. Yet, your never indicated whether you are doing it as a longitudinal design, it means you would use repeated measures ANOVA.I am assuming this is a cross-sectional design.
I am sure this website attached will help you to find your questions to run the correct analysis.
Important: a Likert scale: can have different structures - please find a file attached.
You must transform each possible answer in a corresponding number to that answer:
Strongly agree (5), agree (4), neutral (3), disagree (2) and strongly disagree (1) - the scores will depend on your item statements and what kind of questions you have in mind. Make sure the number will not be on your response, this is just an example.
Good luck and I hope it was helpful.
Kamary
http://www.how2stats.net/p/home.html
Hello Shane. I agree with the previous answers. However, as you do not give essential details so that I can give you a certain answer about what you should do, I can roughly give you the following guidelines: You can use ANOVA to compare between means for more than two independent variables. The equivalent non-parametric criterion is Kruskall-Wallis which is mainly used when small samples are examined (N
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS