# ANOVA or Chi-squared Test, which test should I use?

I have an depentent variable which means agreement, it is a value between 1 and 6 (integer) 6 means more agreement than 1. I have two categorical variables:

• Level of experience. Value from 1 to 4. 1 means less experience than 4.
• Profession. 4 different values, let's say: A, B, C and D.

I want to see if there is a difference in agreement in one statement, given the level of experience. I also want to see if there is a difference in agreement given the profession. Of course seeing the interactions between the categorical variables would be great too.

Which test should I use for each case?

• Does the order matter? Is $1$ less (or more) agreement than $6$? Or are the $-6$ labels purely categorical?
– Dave
May 19, 2021 at 17:32
• That Wilcoxon Mann-Whitney U test will do what you want if you have two groups. If you have three or more groups, the analogue would be the Kruskal-Wallis test. However, the KW test will be saying if the distribution of agreement values differs in each of your $3+$ groups. Is that what you want? (I say that this is equivalent to there being a relationship between the two variables, but you know your problem.)
– Dave
May 19, 2021 at 17:41
• You are right, Wilcoxon Mann-Whitney U test won't do since there are 6 groups. What I want to know is if there is a difference between each of my categories and the level of agreement. For example, does the level of experience matter with the agreement with one particular question? The level of experience is in 4 time intervals. May 19, 2021 at 17:43
• So both variables have ordering to them?
– Dave
May 19, 2021 at 17:46
• It sounds like the problem is somewhat more complex than the original post indicates. Perhaps you can edit the question to include the complexities by describing the data and the problem(s) you want to address.
– Dave
May 19, 2021 at 17:51