Comparison of deprivation across sex

[Piku Sarkhel]

Dear All:
I am coming up with a conceptual doubt regarding a measure of food deprivation across male and female in the same household. I have frequency of food intake (say regular and irregular) for both male and female. I have constructed a nutrition index for both the groups that ranges from 0 to 1 where higher values indicate better nutrition. Now I intend to see the gap between male and female in the same household. The gap index that I have in mind is (male nutrition_index -female nutrition_index)/male nutrition_index. Is this a statistically valid measure of gap ?
Any help would be greatly appreciated.

[Carlo Lazzaro]

Piku:
I am not I got what you're after, but can't you simply interact -nutrition index- and -sex-?

[Piku Sarkhel]

Hello Carlo ,
Thanks a lot for responding. Actually here I intend to keep deprivation as my dependent variable and explore what factors influence deprivation. What I am worried about is whether difference in male and female nutrition index normalized by the male index does capture the deprivation or not. I have constructed nutrition index separately for both male and female. So interacting nutrition index by sex wouldn't probably be feasible if nutrition is my dependent variable.

[Carlo Lazzaro]

Piku:
thanks for clarifying.
With no data example, I find difficult to reply (more) positively.
That said, if male is your reference category, I would feel more comfortable with (female nutrition_index-male nutrition_index)/male nutrition_index.

[Piku Sarkhel]

Hello Carlo
Thanks again. I was wondering what would this signify. So ideally if female index=male index then the value of the deprivation would be zero . If female index is greater than male index then the gap should be negative , right ? So for example if I regress gap on say female employment and the coefficient of female employment is negative it should show that the gap falls in employed females . Ami I right in thinking it through ?

[Carlo Lazzaro]

Piku:
I follow and share your point up to the question mark.
Again, a data excerpt could help enormously.