*USP 2/2017
*Course: FLS6183
*Author: Lorena Barberia
*Lab 3 - What is the difference between the population regression function and the sample regression functions? 
* Do these differences depending on how we select our sample?

*In order to examine this question, we will use the data provided by Gujarati in his textbook. 

* We are also going to learn how to summarize the information we have gathered in a regression table using estout.
* ssc install estout


*Loading data
cd "C:\Data\", clear

*Let's see how our data looks like:
sum consumption income

*This is our population model:
reg consumption income
est sto m1
twoway (scatter consumption income) (lfit consumption income)

* Note that there are 10 income categories.

*Now, let's draw our random samples and run the models:

*Sample A: n=10
use "gujarti", clear
set seed 1245
sample 10, count
sum consumption income
reg consumption income
est sto m2
twoway (scatter consumption income) (lfit consumption income)


*Sample B: n=10
use "gujarti", clear
set seed 1245
sample 10, count
sum consumption income
reg consumption income
est sto m3
twoway (scatter consumption income) (lfit consumption income)


esttab m1 m2 m3,  ci  

*Sample C: n=30
use "gujarti", clear
set seed 1245
sample 30, count
sum consumption income
reg consumption income
est sto m4
twoway (scatter consumption income) (lfit consumption income)

*Sample D: n=30
use "gujarti", clear
set seed 1245
sample 30, count
sum consumption income
reg consumption income
est sto m5
twoway (scatter consumption income) (lfit consumption income)

esttab m1 m4 m5,  ci 
** Now let's stratify the sample by income and random sample from each group (this is similar to Gujarati)

** First we will need to install two user-written packages.

*ssc install moremata
*ssc install gsample

 
use "gujarti", clear
gsample 10, percent strata(income)
reg consumption income
twoway (scatter consumption income) (lfit consumption income)
est sto m6


esttab m1 m2 m6,  ci  


* This command draws a 10% sample. 
* For stratified sampling, a 10-percent sample is drawn from each stratum, thus maintaining the proportion of each stratum.