* Closes a log named read_ad_rev if it is already open.
* capture suppresses any error message if the log does not exist.

capture log close read_ad_rev 

* Removes any dataset currently in memory.
clear

* Prevents Stata from pausing output one screen at a time.
set more off

* Sets the output width to 255 characters.
set linesize 255

* Opens the combined NSDUH Stata dataset containing the individual data from 2002 to 2012.

 use "F:\Heroin\short0212.dta", clear 

* Produces a frequency table showing the number of records for each survey year.
 tab1 year 

* Number of at-risk individuals for heroin initiation
 tab1 year if ELIGHER2==1
 
* Shows the unweighted count of respondents classified as past-year heroin initiates.
 table year if RECHER2==1

* Creates a log file.
* Replace overwrites an existing file of the same name.
* All subsequent output is written to the log.

log using "F:\heroin\read_ad_rev.log", name (read_ad_rev) replace

* Estimate total heroin initiates (svy total) by year from the individal yearly 
* data files that are extracted from combined 11 years of data files

* The foreach loop goes through survey years 2002 to 2012.
* For each year, reload  short0212.dta.
* keep if (year==`x') retains only observations from that year (e.g., 2002).
* compress optimizes variable storage types to reduce file size.
* save "F:\Heroin\short_nsduh_`x'.dta", replaces saves the annual dataset (e.g., short_nsduh_2002.dta).
* use "F:\Heroin\short_nsduh_`x'.dta", clear loads the newly created single-year dataset.
* svyset specifies the complex survey design.
* verep is the variance estimation replicate or PSU variable.
* [pweight=analwt] is the analysis weight.
* strata (vestr) is the stratification variable.
* dof(900) uses 900 degrees of freedom for variance estimation.
* singleunit(missing) handles strata with only one PSU by treating variance as missing rather than using certainty or scaled adjustments.
* di %~59s "NSDUH:" `x' prints a heading such as:NSDUH: 2002 before the survey estimate.
* svy: total RECHER2 computes a survey-weighted population total of those who initiated heroin use during the past year (RECHER2=1).
* The output includes:Estimated total, Standard error, t statistic, Confidence interval with proper adjustment for the complex survey design.
* est store total`x' stores each year's results in memory under names such as:total2002... total2012.
* The // significance ...part is commented out, preventing execution.
* log close read_ad_rev closes the log file and writes output to disk.
* exit terminates the Stata session.

foreach x of numlist 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 {  
  use "F:\Heroin\short0212.dta", clear   
  
  keep if (year==`x')
  compress
  save "F:\Heroin\short_nsduh_`x'.dta", replace
  use "F:\Heroin\short_nsduh_`x'.dta", clear
  svyset verep [pweight=analwt],strata (vestr) dof(900) singleunit(missing)
  di %~59s "NSDUH:" `x'
  svy: total RECHER2
  est store total`x'
		   	}
	
	//significance tests not performed and suppression rules not performed
	
log close read_ad_rev 
exit