Baseline Characteristics

[Eve Alexander]

Hello,
I'm running a manual DID using an exogenous variation of a hh member's death across two waves of a longitudinal survey.

Consolidated data of both survey waves is given together in long format.

Post takes the value 0 if the obs appeared in survey wave 1 and takes the value 1 if it appeared in wave 2.

I would like to look at the value of my outcome based on whether a particular member was disabled at the baseline ie(in survey 1).

so I constructed the disability variable at the baseline like this

 clonevar walking_mil0=difficulty_walk_mil if post==0

Then I run the regression like this (my treatment is hh_memberdied)

reg  dep i.hh_memberdied##i.post##walking_mil0   i.PERSONID , cl(var)

But here the outcome is getting omitted due to collinearity issues. And I understand its because the walking_mil0 was created by making use of post variable and interacting it with post again will be problematic.

How will I be able to get the outcome wrt HH_member had died by wave 2 if they had been disabled at the baseline.
How will I code this?

	
	
* Example generated by -dataex-. To install: ssc install dataex
clear
input double ID_PERSON float(PERSONID post HH_memberdied walking_mil0 difficulty_walk_mil)
10201002009  9 0 0 . .
10201016004  4 0 0 0 0
10201018005  5 0 0 0 0
10201030102  2 1 1 . .
10201050102  2 1 1 . .
10201160104  4 1 0 . 0
10201170102  2 1 1 . .
10201180105  5 1 . . 0
10201200104  4 1 0 . 0
10202006004  4 0 0 0 0
10202007004  4 0 0 0 0
10202014002  2 0 0 . .
10202015002  2 0 0 0 0
10202017002  2 0 0 . .
10202040102  2 1 1 . .
10202060102  2 1 0 . 1
10202070104  4 0 0 1 1
10202150102  2 1 0 . 1
10202160102  2 1 1 . .
10202170102  2 1 1 . .
10202200102  2 1 1 . .
10203006009  9 0 0 0 0
10203011002  2 0 0 0 0
10203014004  4 0 0 0 0
10203015003  3 0 . 0 0
10203019004  4 0 0 . .
10203050102  2 1 1 . .
10203060109  9 1 0 . 0
10203070102  2 1 1 . .
10203140104  4 1 0 . 1
10203160102  2 1 1 . .
10203190102  2 1 1 . .
10204001002  2 0 0 0 0
10204008004  4 0 0 0 0
10204017011 11 0 0 . .
10204040102  2 1 1 . .
10204090102  2 0 1 . .
10204130302  2 1 1 . .
10204150102  2 1 1 . .
10205007004  4 0 0 0 0
10205010102  2 1 1 . .
10205012003  3 0 1 . .
10205015002  2 0 1 . .
10205020104  4 1 0 . 1
10205040104  4 1 0 . 0
10205070104  4 1 0 . 0
10205120103  3 1 1 . .
10205150102  2 0 1 . .
10205170102  2 1 1 . .
10205180102  2 1 1 . .
10205200104  4 1 0 . 0
10206003002  2 0 0 . .
10206006002  2 0 0 . .
10206012004  4 0 0 0 0
10206015004  4 0 0 0 0
10206030102  2 1 1 . .
10206040102  2 1 1 . .
10206110102  2 1 1 . .
10206120104  4 1 0 . 0
10206150104  4 1 0 . 0
10207001002  2 0 0 0 0
10207002002  2 0 0 0 0
10207005002  2 0 1 . .
10207006004  4 0 0 0 0
10207010102  2 1 0 . 1
10207060104  4 1 0 . 0
10207070102  2 1 1 . .
10207100102  2 1 1 . .
10207110102  2 1 1 . .
10207120102  2 1 1 . .
10208012002  2 0 0 0 0
10208030103  3 0 1 . .
10208060102  2 1 1 . .
10208070102  2 1 1 . .
10208090102  2 1 1 . .
10208120102  2 0 0 0 0
10208130102  2 0 1 . .
10208140102  2 1 1 . .
10301013004  4 0 . 0 0
10301020102  2 0 1 . .
10301040102  2 1 1 . .
10301050103  3 1 0 . 0
10301090104  4 0 0 0 0
10302001004  4 0 . 0 0
10302005004  4 0 0 0 0
10302010106  6 1 . . 0
10302014002  2 0 0 . .
10302020102  2 1 1 . .
10302070102  2 1 1 . .
10302130102  2 1 0 . .
10302140102  2 1 1 . .
10303011004  4 0 0 0 0
10303015004  4 0 0 . .
10303110104  4 1 0 . 0
10303150102  2 0 1 . .
10304009002  2 0 1 . .
10304014002  2 0 0 0 0
10304080102  2 0 1 . .
10304110102  2 0 1 . .
10304120102  2 1 1 . .
end

[George Ford]

this may do it.

egen walking_mil0alt = mean(cond(post==0,difficulty_walk_mil,.)) , by(ID_PERSON)