Compute Cronbach's alpha for dichotomous items

amcr729 — Tue, 09 Mar 2021 15:48:53 GMT

Hello all,

I have a simple problem. I am trying to compute cronbach's alpha for a group of dichotomous items (0/1) -- see code below -- but am getting error message that the first variable name is not valid. There is nothing wrong with the variable name -- when I run a proc freq for it, I have no problem. I wonder if the problem is trying to compute cronbach's alpha for dichotomous items? I know the Kuder-Richardson Formula 20 (KR-20) was developed to assess reliability for dichotomous items, but my understanding was that it is also acceptable to compute cronbach's alpha for dichotomous items. It is my preference to use alpha because it is more familiar to readers. If someone can help me understand the problem I would be grateful!

proc corr data=si nomiss alpha;

var bl_sw_stigma7 bl_sw_stigma8 bl_sw_stigma9 ;

run;

ERROR: " BL_SW_STIGMA7" is not a valid name.

Re: Compute Cronbach's alpha for dichotomous items

FreelanceReinh — Tue, 09 Mar 2021 16:47:25 GMT

Hello @amcr729,

Your code as posted runs without problems with dummy data I've just created. The code with the double "blank" after the VAR keyword, the error message and the "blank" it contains after the first quotation mark suggest that in your real code there is an invisible character (e.g., a non-breakable space 'A0'x) between "var" and "bl_sw_stigma7" which the compiler sees as an invalid character in the first variable name (like "var §bl_sw_stigma7"). So, deleting that character (and possibly other irregular "blank" characters, if any) should solve the problem.

topic Compute Cronbach's alpha for dichotomous items in Statistical Procedures

Compute Cronbach's alpha for dichotomous items

Re: Compute Cronbach's alpha for dichotomous items