Can the reliability be used using Cronbach's alpha when the base of the respondent for each question is different?

[George_Shin]

For example, in the survey on job satisfaction, questions A1 to A10 consist of items related to work-life balance. At this time, the survey was conducted for all respondents in A1~A5, only white-collar in A6~A8, and only blue-collar in A9~A10. At this time, is it not unreasonable to measure the reliability of the A1 to A10 responses of work-life balance with the Cronbach alpha value? Or are there other alternatives for measuring reliability?

 

[spunky]

For example, in the survey on job satisfaction, questions A1 to A10 consist of items related to work-life balance. At this time, the survey was conducted for all respondents in A1~A5, only white-collar in A6~A8, and only blue-collar in A9~A10. At this time, is it not unreasonable to measure the reliability of the A1 to A10 responses of work-life balance with the Cronbach alpha value? Or are there other alternatives for measuring reliability?

If you expect group differences in reliability, your safest bet is to fit a multiple-group Confirmatory Factor Analysis to your data and calculate the reliability from the loading estimates and error variances.