Analyzing Likert Scores

Hi - from this thread you'll be able to ascertain my very low skill levels when it comes to statistics! Here is your chance to give me the answer and feel superior!

I have completed a survey using Likert scales, asking respondents to rank their level of competency in 15 different management competencies.

I then asked their managers to rank the same person on the same scale using the same competencies.

I now have two tables of data, one for the self-assessment and one for the manager's assessment. Just from the numbers I can see that in most categories the self-assessment score is higher, but I need to select the right test to prove this statistically.

So basically, I'm looking to establish how significant the differences are between the two sets of likert scores. Can anyone help?


New Member
I do not know what likert scales are, but if you have two mean competence scores for your two samples, you can compare them with a difference in means approach. Do a Google search on "difference in means" and you will find plenty tutorials.

If you are using an ordinal scale of measurement (i.e, rank data) on a likert scale you should look into non parametric tests. I believe the Mann-Whitney U test is appropriate. If you decided to treat the scale of measurement as ratio (i.e, summing up all the scores on each scale) you could employ the regular Student T Test for two independent samples. The Mann-Whitney is basically the non-parametric analog to the Student T Test for two independent samples.

Best of luck.
Last edited: