Dear Statalisters,

For my thesis I am working on a dataset in which empowerment is measured from three perspectives: personal (indicator: GHQ-12 to measure psychological well-being), relational (decision-making power of women within household decisions) and environmental (social support and stereotypes statements). For all, I would like to generate separate indices (not an aggregated one, because I think I will then loose much of the explanatory value). Before creating the indices, I measured the Cronbach's Alpha score of the several variables and as all had a score of +0,7 I started to create the following indices for the personal and environmental indicators:

Code:
gen psychwellbeingindex = psychwellbeing_1 + psychwellbeing_2 + psychwellbeing_3 + psychwellbeing_4 + psychwellbeing_5 + psychwellbeing_6 + psychwellbeing2_1 + psychwellbeing2_2 + psychwellbeing2_3 + psychwellbeing2_4 + psychwellbeing2_5 + psychwellbeing2_6

gen socialsupportindex = socialsupport_1 + socialsupport_2 + socialsupport_3 + socialsupport_4

gen stereotypesindex= stereotypes_1 + stereotypes_2 + stereotypes_3 + stereotypes_4
Question 1: the above are not indices from 0 - 1 yet. I can imagine that is simply done for the socialsupportindex as well as the stereotypeindex by / 4 but I am not too sure about the psychwellbeingindex as this one is based upon a three-point likert scale. The max score is 36 (great psychwellbeing) and the lowest 21.

Question 2: Then, I am not too sure how to go about the decision-making index. For this index, I have 8 separate variables which I coded as followed: 1 = someone else in the family makes the decision, 2 = Jointly decision between me and my husband, 3 = myself. If I would use the same approach as above, I would score ''myself'' higher as ''jointly'', which is not correct. Therefore, I want to recode ''myself'' as 2 as well without losing the label ''myself''.

I hope this makes sense!

Best,

Linda