Decomposition analysis

[Ishrat Jahan]

I want to run a decomposition analysis between groups. My outcome is binary, with multiple independent and profounding variables. Which type of decomposition should I run in the STATA? Moreover, this is survey data showing a comparison in years. Can anyone guide me with proper instructions?

[Dung Le]

Hi Ishrat,
If you have two groups, you then may want to use oaxaca command (from SSC, written by Ben Jann) which does Oaxaca-Blinder decomposition. You also may want to specific logit option to account for the fact that your outcome is a binary.

Hope this helps.

[Elsie Akwara]

Following this conversation, can someone please help me with the STATA code for doing the Oaxaca decomposition analysis?
I am using the demographic and health survey Data (household survey data) with a binary outcome "contraceptive use 'cuse'" 0,1 and I have three independent variables: wealth (3 categories: poor, middle, rich), residence (2 categories: urban, rural) and age (2 categories: 15-19, 20-24). I want to do a decomposition analysis showing comparisons between age groups. What would be the STATA code? Many thanks in advance!

[George Ford]

ssc install oaxaca
help oaxaca

there's a logit option.

use omega option for a two-way decomposition.

age must be 2 categories (0/1).

Also see:

https://www.statalist.org/forums/forum/general-stata-discussion/general/1739797-twofold-oaxaca-blinder-decomposition-interpretation

[Elsie Akwara]

Many thanks, George! I shall try this and come back if I hit a roadblock.