0
$\begingroup$

When I listen / read manuals on how to measure Cronbach's alpha, none of them mention how to deal with independent and dependent variables ( variables Role in SPSS). Could someone advise if it is correct to run SPSS analysis simultaneously on both types of data. Say I have 40 items that represent 8 constructs ( independent variables) and 5 items that represent one construct - dependent variable. According to what i already heard, I just need to go to " Reliability Analysis " and run it for all 45 items without bothering if they are independent / depended. Is it correct? Thanks!

$\endgroup$
3
  • 2
    $\begingroup$ Asking for how to do this in SPSS is off-topic here. I think you have a legitimate statistical question, though. Can you clarify it & make it software neutral? It is fine to mention SPSS, & someone might provide some tips about it, but the part that is on-topic here is the non-software related portion. $\endgroup$ May 12, 2015 at 13:07
  • $\begingroup$ Many thanks! My question is if it is ok to calculate Cronbach's alpha for independent and dependent variables simultaneously? $\endgroup$
    – Oleg Cohan
    May 12, 2015 at 13:17
  • 2
    $\begingroup$ I see what you are asking about. Your question is entirely about SPSS and is not statistical. The so called "variable role" attribute bothers you. It is irrelevant to Reliability procedure in SPSS. So input there variables disregarding their predefined "role". I even would recommend you to switch off menus auto completions based on "roles" (go to Edit - Options). $\endgroup$
    – ttnphns
    May 12, 2015 at 14:08

1 Answer 1

3
$\begingroup$

Cronbach's alpha is not related to the concepts of independent and dependent variables.

Cronbach's alpha measures the internal consistency of scales that are made up of multiple items. You would have to calculate the alpha of each of the 8 each contstructs using (I assume) 5 items each, and one for the 5 items that will form your dependent variable.

A high alpha indicates that the separate items are correlated. This is an indication that the items might measure the same construct, and you can add the scores of the individual items to a total score per construct to create a scale score.

If the items would not be correlated, adding the items would mean adding noise, which would give unreliable scores.

After this verification you could regress the dependent total score on the other 8 total scores.

Some extra information: https://explorable.com/cronbachs-alpha

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.