*Stata Assignment I Do-file

log using "C:\Documents and Settings\Ryan Goodstein\My Documents\UNC\courses\econ400\Stata1.smcl", replace
use "C:\Documents and Settings\Ryan Goodstein\My Documents\UNC\courses\econ400\CPSdata.dta", clear

drop recnum

rename interest_income intinc

label variable intinc "Interest Income"

gen annhrs = hours*wkslyr

gen wage = earnings / annhrs

gen agegrp = 1
replace agegrp = 2 if age >= 35 & age <= 44
replace agegrp = 3 if age >= 45 & age <= 54

summarize
*min and max value for age is 25 and 54 respectively

display (99-37.7406)/10.18125
*max value for hours is 99, is an outlier (z score = 6.017)

su earnings, detail
*mean value for earnings is 30141.1
*median value is 26026
*90th percentile is 52000

histogram earnings, title("Earnings")
*"close" to bell shape but distribution is skewed to the right 

graph twoway scatter earnings hours, title("Earnings vs Hours")
correl earnings hours
*both suggest there is a positive relationship between earnings and hours

tabulate margrp, su(wage)
*married people have highest wage

graph bar (mean) wage, over(educgrp) blabel(bar)
*bar chart shows that mean wage is increasing in education group

graph bar (median) wage, over(educgrp) blabel(bar)

table agegrp, contents(mean intinc)
*older age groups have more interest income on average

su intinc if agegrp == 3, detail
*median interest income is much smaller than mean interest income
*suggests that distribution of interest income is skewed to the right

histogram intinc if agegrp == 3


log close