Use Likert scale mean to calculate a ranking of several answers?

[Lehmann]

Hello again,

As my last question was very successfully answered, which I am very thankful for, I do it again


I have 6 Likert type questions (very important - not important at all). I know that it is not recommended to interpret the mean.

But what I want to do is, I want to rank the 6 answers in terms of: this is the most important factor, second most important factor etc. For that purpose, the mean seems to be useful, isn't it?

Or how can I analyse which factor is the most important one?

Thanks!

 

[Karabiner]

Since it is (at least) an ordinal scale, you can use the measure of cenral
tendency for ordinal scales.

With kind regards

K.

 

[Lehmann]

Thanks. The problem is, those measures can lead to different contradicting results. Assume the following example (5="very important" ... 1="not important at all"):

View attachment 3101

Using the mode would rank factor 2 as the most important factor. The median would rank factor 1 as most important. Intuitively, the median makes more sense for the purpose of ranking, however, in this particular example, the result is either way difficult to interpret... I'm just wondering how to deal with such cases...

 

[noetsi]

I think the answer to that depends on the substantive question you are answering. A median tells you the middle value of your distribution obviously, the mode the most common value. Which is germane, and why, to what you actually want to know. Just as importantly, what does it say for that question (or questions) that the most common value and middle value differ?