0
$\begingroup$

I need to run a stats analysis on data at the bottom of this post. Essentially, I want to investigate whether Diet Group (categorical independent var) or # of Stitches (categorical independent var) affects Score (ordinal dependent var).

From what I understand, I would need to use a non-parametric test as my data doesn't seem to be normal / there isn't enough to convince me that I have a normal dataset. This would mean that I can't use a two-way ANOVA (which I was originally recommended) to investigate all three variables together.

From my discussions, I believe I should be conducting two Mann-Whitney U tests (one for Diet Group vs. Score and one for # of Stitches vs. Score. However, as my dependent var (Score) essentially falls mostly within two groups (either 1 or 2) even though the scoring scale itself is 0 - 4, I was informed that these ranked tests wouldn't be ideal as they don't do well when there are lots of ties within the ranking.

Are there any other statistical tests that I should be using with this non-ideal data set or is Mann-Whitney U the best I'd be able to do?



Raw Data:

animal diet group # stitches score
A ED 5 2
B ED 5 0
C ED 5 2
D ED 5 1
E ED 5 1
F ED 5 1
G ED 5 1
H ED 5 2
I ED 5 2
J WD 5 2
K WD 5 2
L WD 5 2
M WD 5 2
N WD 5 1
O ED 7 2
P WD 7 1
Q WD 7 2
R WD 7 2
S WD 7 2
T WD 7 2
$\endgroup$

1 Answer 1

0
$\begingroup$

You can correct for ties in the Mann-Whitney test, have you considered using this version? Some software, e.g., Statistica calculate this automatically.

A tie correction method can be found on Wikipedia: https://en.wikipedia.org/wiki/Mann–Whitney_U_test

Note, that it is still not ideal since your outcome/dependent variable only has three categories.

An alternative could be multinomial logistic regression but your sample size may be too small for this, especially if you originally planned for a t-test in your sample size calculation.

$\endgroup$
1
  • $\begingroup$ Your answer could be improved with additional supporting information. Please edit to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Oct 10, 2021 at 6:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.