Likert Analysis

#1
Hi all,

I'm doing some analysis of survey data for a research project.

On the broadest level, I have data from art teachers about all sorts of things. They were asked if they identify as artists or teachers, and I am doing analysis on items to see what differences there are.

One item I am looking at is "average time spent art making." I have four years of survey data.

I used Mann-Whitney U tests on responses from each survey to compare the artitsts and the teachers.

Now what I want to do is to look longitudinally, essentially wanting to know is there a difference in the amount of time an artist spends making at at t=1 to t=4, and is that difference statistically significant.

I'm unsure on which tests I should be using. One person suggested a one-way ANOVA, but I read that can only be used on ordinal data. Is there something else I should be looking to use? I guess part of it comes from me being unsure in the exact survey design being employed.

Does the study design constitute a repeated measures design, or is it something different?

Many thanks in advance!
 
Last edited:

Karabiner

TS Contributor
#2
One person suggested a one-way ANOVA,
This is inapproriate if you want to analyse longitudinal
(repeated measures) data.
but I read that can only be used on ordinal data.
No. First of all, assumption for ANOVA is that variables are
measured on an interval scale. Second, even if you have
a test which can deal with ordinal data (like the Kruskal-
Wallis H-Test, or the U-test which you have already
performed), such a test can deal with interval scaled
data also.
Is there something else I should be looking to use? I guess part of it comes from me being unsure in the exact survey design being employed.
Seems like you could have a look at repeated-measured
(mixed) ANOVA, with time as within-subjects factor "time"
and a grouping factor teacher/artist.

With kind regards

K.