{smcl}
{* 10oct2006}{...}
{hi:help aggsum}
{hline}

{title:Title} 
 
{p 4 24 2}{cmd:aggsum} {hline 2} Summarize aggregated data

{title:Syntax}

{p 8 17 2}
{cmdab:aggsum} [{it:function-list}] {ifin}, {opth by(varlist)} [ {opt f:ormat} ] 

{p 4 4 2}Where {it:function-list} is one or more functions (including parameters) recognized by {helpb egen}.

{title:Description}

{pstd}Let's say, to pick a random example out of thin air, that you wanted to know the distribution of hospital-median time-to-ekg, and you have patient level data. {cmd:aggsum} will give you that distribution:

{pstd}{cmd:. aggsum median(t_to_ekg), by(hospital)}

{pstd}Instead of, or in addition to, {cmd:median()}, any {cmd:egen} function that works with the {cmd:by} prefix could be used. Specifically:

{p 8 8 2}{opth count(exp)}{break}
{opth iqr(exp)}{break}
{opth kurt(varname)}{break}
{opth mad(exp)}{break}
{opth max(exp)}{break}
{opth mdev(exp)}{break}
{opth mean(exp)}{break}
{opth median(exp)}{break}
{opth min(exp)}{break}
{opth mode(varname)}{break}
{opth pc(exp)}{break}
{opth pctile(exp)}{break}
{opth rank(exp)}{break}
{opth sd(exp)}{break}
{opth skew(varname)}{break}
{opth total(exp)}{break}

{pstd}The output of {cmd:aggsum} is identical to that of {cmd:summarize, detail}. The {opt f:ormat} option is also the same.

{pstd}If the optional {it:function-list} is omitted, {opt count()} of the first {opt by()} variable is assumed. In other words, 

{pmore}{cmd:aggsum, by(ID)}

{pstd}summarizes the number of records per ID.

{title:Remarks}

{pstd}The syntax is very non-standard, but, it was easy to do this way. Also, duplicating the non-detail version of {cmd:summarize} would have required more work.

{title:Examples}

{pstd}{cmd:. aggsum mean(height) median(height) mode(height), by(city) format}

{pstd}{cmd:. aggsum mean(age) mean(heartrate) mean(bloodpressure), by(provider)}

{pstd}{cmd:. aggsum, by(id)}

{title:Also see}

{pstd}Online: help for {help summarize}, {help egen}

{pstd}Contact: {browse "mailto:elliott.lowy@va.gov":Elliott Lowy}