* use wapop data for descrips and regression

*******open log file to save all my output
log using "D:\My Documents\529 adv reg course\wapop1.log"	*begins log

set more off		*turns off pause mode

*************opens the wapop data.
use "D:\My Documents\529 adv reg course\wapop08", clear  	


***makes set of indicators for region that I can use by prefacing commands with xi:
xi i.REGION 	

xi: summarize HHINC i.REGION

*********runs regression of household income on region and number of people
xi: regress HHINC i.REGION PEOPL	

*******get predicted income and  summarize
predict inchat
summarize inchat


****what would happen to income if every household had one more person?
replace PEOPL =PEOPL+1

summarize inhat2

replace PEOPL = PEOPL=1


**recode into new home ownership dummy
 recode Q3P2 1=1 2/3=0, generate(owndum)

summarize owndum


****create new income squared***********************
generate hhincsq=HHINC*HHINC

summarize HHINC hhincsq

***logit of home ownership ************

logit owndum PEOPL HHINC hhincsq  _IREGION_2  _IREGION_3  _IREGION_4 _IREGION_5  _IREGION_6 _IREGION_7 _IREGION_8 _IREGION_9 _IREGION_10


*****get predicted probability of home ownership, summarize, and scatterplot
predict predprob1, p

 summarize predprob1 
 
 scatter predprob1 HHINC if HHINC< 100000

*****add $10,000 to income if it was less than 10,000
***and remake income squared to match

replace HHINC=HHINC+10000 if HHINC<10000
replace hhincsq=HHINC*HHINC

***get predicted probb with new HHINC
predict predprob2, p
summarize predprob2
scatter predprob2 HHINC if HHINC< 100000


 
 **calculate marginal effects at median values.
 mfx compute, dydx at(median)
 mfx compute, dydx at(mean)   

*Don't save your data if you changed the values!