Hello, have two different variables (both use Likert scales) in two groups. One variable has a four points scale and the other has a five point scale.
I want to compare ranks sums between them and I have used a Mann–Whitney–Wilcoxon-test (Wilcoxon rank-sum test) in R. The sample size for each group is rather small (n group 1 = 12, group 2 = 13).
Can I use this test here? How are the different scales to be treated in this example?
Any help would be appreciated.
Hello Martin,
Do you wish to compare:
1. DV#1 (4-pt response scale) between Grp1 and Grp 2, followed by DV#2 (5-pt response scale) between Grp 1 and Grp 2;
2. DV#1 vs DV#2 for Grp1, followed by a similar comparison for Grp 2;
3. Both; or
4. Something else?
If 1, then Mann-Whitney will work just fine. (Two such tests in all).
If 2, then Wilcoxon signed-ranks sum test may be used, after conversion of 4-pt values to 5-pt. values. (Again, two such tests).
If 3, then possibly the aligned ranks method.
If 4, then you'll need to explain further.
Conversion of 4-pt to 5-point, if needed (for #2):
1 --> 1
2 --> 2.333
3 --> 3.667
4 --> 5
Good luck with your work.
Hello David,
thank you for your answer. It is none of the four possibilities you listed. This is what I want to do:
I want to compare the DV#1 (4-pt response scale) only found in Grp1 with the DV#2 (5-pt response scale) only found in Grp 2. So it is a comparison of two different variables between two independent groups.
Is that feasible and if so, how could I do that?
Thanks,
Martin
Hello again Martin,
As long as you're convinced (and are willing to defend the claim) that DV#1 and DV#2 are interchangeable as measures of the same characteristic, save for the response scale, then Mann-Whitney would suffice.
If not interchangeable, then your comparison won't be "apples to apples," and won't make much sense, regardless.
Be sure to apply the scale conversion first.
Good luck with your work.
Thank you so much for your answer David, I will try that out.
Regards
Martin
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis