I am looking for an appropriate method to analyse correlation between results of a patient survey and patient characteristics e.g. age, sex, education level. Would the Mann-Whitney-U test be appropriate for such an analysis?
Do a regression analysis. mann-Whitney is NOT appropriate.
If you analyze using SPSS (I assume SPSS), you put any number, it will display the result. But you must always go back what statistic says. Between characteristics (I assume categorial data) and Likert, normally, we normally look for differences. Check if your data parametric or non-parametric only you can decide on your choice of analysis. For example if your data are parametric then you should look for how many group of characterics. If gender, male & female (2 groups), you should use t-test. If more, you can use ANOVA. Meaning to say, whatever analysis you decide, please check which statistic test is appropriate to answer your research objective. Hope my answer will help.
In regression analysis you do not have to check for normality.
Check the regression assumption. Actually every statistical analysis has their own assumptions.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey