# Should I take a mean of answers given by an individual on a Likert-Style Scale?

I want to use three established scales which have a Likert Scale as a response - for instance the 10-item Need for Belonging scale (Leary et al., 2013). This scale is a 1-5 response alongside another scale and the third scale is a 1-7 response. I would like to ask that when a participant answers the scales, can I compute a mean score from their answers and thus have a per participant score like this for each scale i.e. three scores, one for each scale per participant? Some papers I read that employ these scales say "To analyze the motivational scales, mean scores were computed across the relevant items to form indices" or "We averaged the items in each scale and used the average scores in the data analyses." but I was not sure how to interpret them.

• You do not say anything about the overall objective of your study. You have described one of three 'scales' in detail, but said nothing about the other two, how the three scales relate to each other, or how you propose to analyze your data. // Because Likert scores are ordinal, but not numerical it is unclear whether a mean of answers on several questions (say 10 as for 'Belong') is meaningful. (Is the difference btw Likert 4 and 5 as meaningful as difference btw 2 and 3?) It may be better to use the median score Commented Mar 22, 2021 at 20:23

Comment continued: Also, if your analysis will involve any comparison of Likert-5 data with Likert-7 data, you may need to transform Likert-7 for compatibility with Likert-5. Maybe something like $$L.5 = 2(L.7 - 4)/3 + 3 = 2L.7/3 + 1/3$$ might work.

Consider Likert-7 data:

L.7 = sample(1:7, 100, rep=T)
summary(L.7)
Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
1.00    2.00    4.00    4.05    6.00    7.00


'Transformed' to L.5 scale:

L.5 = 2*L.7/3 + 1/3
summary(L.5)
Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
1.000   1.667   3.000   3.033   4.333   5.000


Note: There is more than one way to do this. You can google something like 'transform Likert 7 to Likert 5' to see links with lots of approaches and lots of ads.