cd "Q:\My Drive\HEAL_LAB\STATA Training\data_sets"


**import the data **
import excel using birth_data.xlsx, first sh("data") clear

** this estimates the first model **
reg birthweight gestation

** create a scatter plot for the first model **
scatter birthweight gestation, mcolor(gray) mfcolor(red) 

** create a scatter plot with best fit line **
twoway (scatter birthweight gestation, msymbol(plus) mcolor(emerald)) (lfit birthweight gestation, lcolor(forestgreen))


** create table **
sum birthweight if boy==0
sum birthweight if boy==1
ttest birthweight, by(boy)


*** estiamte the model with months ***

*option 1 - create dummy variables for each month *
tab birth_month, gen(month)
reg birthweight month2-month12

*option 2 - use fixed effects *
reg birthweight i.birth_month

* option "i will lose in a stata death match" *
gen feb=0
replace feb=1 if birth_month==2
gen mar=0
replace mar=1 if birth_month==3
//I'm not going to finish this because i don't hate myself//


*** estimate the 5 models into one table ***
*first, need to create a dummy variable for smoking*
gen smoke=0
replace smoke=1 if cigar>0



* model 1*
reg birthweight smoke
outreg2 using application_table.xls, se bracket replace

*model 2 *
reg birthweigh smoke gestation 
outreg2 using application_table.xls, se bracket append


*model 3 *
reg birthweigh smoke gestation weight_gain 
outreg2 using application_table.xls, se bracket append

*model 4 *
*need to create dummy for married *
gen married=0
replace married=1 if mother_married=="yes"

reg birthweigh smoke gestation weight_gain married 
outreg2 using application_table.xls, se bracket append

*model 5 *
reg birthweigh smoke gestation weight_gain married i.birth_month
outreg2 using application_table.xls, se bracket append