HR Staff Survey

#1
Hi

I've recently completed a staff survey which involved asking a number of questions which required employees to tick a box on a 5 point scale ranging from Strongly Disagree to Strongly Agree (Likert Scale).

For each of the different offices, I've calculated the average for each question.

I've noticed that there are differences in the average responses for our 5 offices. For instance, the average response to one question for our Slough office was 4.3 but, for the same question, for our Oxford office it was 3.9.

I want to know if this difference is significant. What test would I use?

Is there an Excel command that would calculate this?

Thanks in advance.

Rich
 

Karabiner

TS Contributor
#2
a 5 point scale ranging from Strongly Disagree to Strongly Agree (Likert Scale).
Well, no, actually this is just a single Likert-type item. A Likert scale consists of several of such items,
which are meant to jointly measure the same construct and which are summed up.
For each of the different offices, I've calculated the average for each question.
Strictly speaking, Likert-type items are ordinal scaled, so means would not be appropriate.

I want to know if this difference is significant. What test would I use?
Generally speaking, it is problematic to eyeball onre's data and pick out from many possible
comparisons those few which seem to be notewothy. This can distort the p values.

You did not mention number of items and sample sizes, but perhaps you could consider a series
of Kruskal-Wallis H tests (as global tests). I.e., for each item, you perform a H-test with 5 groups
(offices). If such a H-test turns out to be statistically significant, you can perform pairwise comparisons
between offices, using Mann-Whitney U-tests.

With kind regards

Karabiner