Hello Everyone, I have got data that is not normally distributed and I need to run the Mann-Whitney U test instead of the Independent Samples T-Test, as well as the Kruskal-Wallis Test instead of * ANOVA. The problem is my data consist of five-items Likert scales ( I have several items that test a particular aspect of the study, they are organized in terms of scales and every scale consists of a number of items which are all Five Point Liker-scales, and the Cronbach Alpha is fine). My questions is, do I compute these items based on the mean to create one variable? Or do I need to compute them based on something else (Sum, median...) because the non-parametric tests use rankings? I do hope you would be so kind as to help me. Thank you.
The usual approach to creating scales from Likert-scored items is simply to add them together (or you can take an average, which is essentially the same thing because you are dividing by a constant). After that I would look at the distribution of your scales.
When using non-parametric tests like the Mann-Whitney U test or the Kruskal-Wallis test with Likert scale data, you have a few options for aggregating the individual item responses into a single variable. Here are two common approaches:
Both approaches have their advantages and considerations. Using the sum or total score provides a continuous variable that can be used with non-parametric tests, but it assumes equal intervals between the Likert scale points. Using the median or mode preserves the ordinal nature of the data but results in a discrete variable.
When using non-parametric tests, it's important to note that these tests work based on the ranks or relative positions of the observations rather than their specific values. Therefore, the choice of how to aggregate the Likert scale items should be based on theoretical and practical considerations, as well as the assumptions and requirements of your analysis.
Additionally, it's good practice to report the individual item responses alongside the aggregated scores to provide a comprehensive understanding of the data and allow for further analysis or interpretation if needed.
Remember to interpret the results of non-parametric tests cautiously and consider the limitations and assumptions of these tests, as well as the specific context of your study.
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Extraction