{smcl}
{* 13dec2004}{...}
{hline}
help for {hi:nodup}{right: El }
{hline}

{title:Show discrepancies within sets of records}

{p 8 17 2}
{cmdab:nodup} [{it:varlist}] [{cmd:if} {it:exp}] {cmd:, dup(}{it:varlist}{cmd:)} [{cmd:over(}{it:integer}{cmd:)}]

{title:Description}

{p 4 4 2}Be sure to see {bf:Remarks}, below.

{p 4 4 2}{cmd:nodup} lists differences among similar records. The dataset is divided into {bf:set}s based on the mandatory {cmd:dup()} option, the records within each set are compared, and the differences reported.

{p 4 4 2}For each variable in {it:varlist}, the number of {bf:set}s that are {it:not} identical is reported (twice, actually: with and without counting {it:missing} as a distinct value). 

{title:Options}

{p 8 8 2}{cmd:dup(}{it:varlist}{cmd:)} determines what constitutes a {bf:set} - specifically, records that are identical over {it:varlist}.

{p 8 8 2}{cmd:over()} specifies another level of reporting: In addition to reporting the number of sets with more than a single value per variable,
{cmd:nodup} includes a column that reports the number of sets having more than {it:integer} values per variable (2 by default).

{title:Remarks}

{p 4 4 2}Tooth-achingly difficult to describe in the abstract, this command is very easy to actually use. For example, consider a patient database including SSN, name, and birth date. The following command:

{p 8 8 2}{cmd:. nodup name bdate, dup(ssn)}

{p 4 4 2} will output:

{p 8 11 2}1. the number of distinct patients based on SSN

{p 8 11 2}2. the number of patients (based on SSN) with more than one name in the dataset.

{p 8 11 2}3. the number of patients (based on SSN) with more than one birth date in the dataset.

{p 4 4 2} Specifying {cmd:over(5)} would cause the output to include the number of patients with more than 5 names (and the number with more than 5 birthdates).

{title:Also see}

{pstd}Online: help for {help duplicates} 

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