Misc. Stats Topics Discussion > Compare different likert scales
Hi Mimmi! That is a tough one, for sure. To my knowledge, there aren't any really great ways to accomplish this. There are some ways to accomplish this, but they aren't without flaws. The first thing that comes to mind would be to dichotomize all items and then create a count variable from the resulting items.
The procedures you would use to dichotomize the Likert variables would depend on the nature of the items and what cut-points might be conceptually meaningful. For example, if your likert scale is 1 to 5 and 1=never, 2=seldom, 3=sometimes, 4=often, and 5=always, you may choose to make 2 or 3 your cut-point. If you choose 2, you would recode so that (1=0) and (2 thru 5=1).
The SPSS syntax for that recode would be:
Recode XXXX (1=0) (2 thru 5=1) into XXXXr.
EXECUTE.
NOTE: you would need to replace "XXXX" with your variable name.
I hope that is helpful!
Jeremy Taylor
Thanks for your reply!
But if i compute z-scores of all item scores and take the mean of the z scores would that not also be a good measure?
Because otherwise i am afraid to loose so much information...
Cheers!
Mimmi
I would be leery of using z-scores with dichotomous (yes/no) variables, so that wouldn't be my first choice, but you could always try it both ways and see if there are meaningful differences in the outcome. If there are, you can then try to determine why that is and what it could mean. I hope that is helpful!
Jeremy Taylor
Your all wrong
June brown
hey!
I want to make one score for each person from different items. But the answers on the items are given on differen likert scales: 5-point likert scale, 7-point lickert scale and even just a "yes" "no" answer (2 point scale).
How can i do this?
Can i use z-scores?
Thanks in advance!