{smcl} {* *! version 1.1.0 01jul2009}{...} {cmd:help epctile} {hline} {title:Title} {phang} {bf:epctile} {hline 2} Estimate percentiles with standard errors {title:Syntax} {p 8 17 2} {cmd:epctile} {varname} {ifin} {weight} {cmd:, }{cmdab:p:ercentiles(}{it:numlist}{cmd:)} [{it:options}] {synoptset 25 tabbed}{...} {synopthdr} {synoptline} {syntab:Main} {synopt:{opt p:ercentiles(numlist)}}estimate percentiles corresponding to the specified percentages{p_end} {synopt:{opt svy}}respect survey settings{p_end} {synopt:{cmd:subpop([}{varlist}{cmd:] [}{help if}{cmd:])}}restrict estimation to a subpopulation (domain){p_end} {synopt:{opt over(varlist)}}estimate percentiles separately within the groups of values identified by {varlist}{p_end} {synopt:{opt L:evel(#)}}change the confidence level to be used in CI reporting{p_end} {synopt:{opt spec:label}}use levels of the {cmd:over(}{varlist}{cmd:)} groups sto label the estimation results{p_end} {synopt:{opt val:uemask(string)}}use the value mask to label the estimation results{p_end} {synoptline} {p2colreset}{...} {p 4 6 2} {cmd:pweight}s, {cmd:fweight}s and {cmd:iweight}s are allowed; see {help weight}.{p_end} {title:Description} {pstd} {cmd:epctile} estimates percentiles at given levels and computes appropriate standard errors. {help svy} settings are partially supported: the user needs to specify {cmd:svy} option to invoke appropriate variance estimation methods. {title:Options} {dlgtab:Main} {phang} {opt p:ercentiles(numlist)} provides the list of percentages at which the percentiles are to be estimated. The numbers must be strictly between 0 and 100. {phang} {opt over(varlist)} requests to estimate percentiles separately within the groups of values identified by {varlist}. {dlgtab:Survey settings} {phang} {opt svy} instructs to respect survey settings. The point estimates will be computed using the appropriate weights, and the variance estimates will be obtained via a call with {help svy} prefix. If a specific variance estimation method needs to be used, it must be specified with the design using {help svyset}.{p_end} {phang} {cmd:subpop([}{varlist}{cmd:] [}{help if}{cmd:])} restricts estimation to a subpopulation (domain). At least one of {varlist} and {help if} conditions must be specified.{p_end} {dlgtab:Reporting and formatting} {phang} {opt L:evel(#)} changes the confidence level to be used in CI reporting. The default is the system level, which is usually 95%.{p_end} {phang} {opt spec:label} requests to use levels of the {cmd:over(}{varlist}{cmd:)} groups rather than {cmd:_subpop} prefix to label the estimation results{p_end} {phang} {opt val:uemask(string)} requests to use the value mask to label the estimation results. The trailing characters in the mask will be replaced by the actual levels. Thus a good mask will look like {cmd:valuemask(}{it:00000}{cmd:)}.{p_end} {title:Examples} {phang}{cmd:. use adept_2002, clear} {phang}{cmd:. svyset mesto [pw=hhw], strata(stratum)} {phang}{cmd:. epctile income, p( 5 10 25 50 75 90 95 ) over( obraz ) svy} {phang}{cmd:. epctile income, p(5 10 25 50 75 90 95) over(pol obraz) svy speclab valuemask( "00000")} {title:Also see} {psee} {space 2}Francisco, C A, and Fuller, W A (1991). Quantile estimation with a complex survey design. {it:The Annals of Statistics}, {bf:19} (1), 454--469. {browse "http://www.jstor.org/stable/2241867":JSTOR link}. {psee} {space 2}Help: {help pctile}, {help svy} {p_end}