{smcl}
{* 15jun2015}{...}
{cmd:help erepost}
{hline}
{title:Title}
{p 4 4 2}{hi:erepost} {hline 2} Repost the estimation results
{title:Syntax}
{p 8 16 2}{cmd:erepost} [{cmd:b =} {it:b}] [{cmd:V =} {it:V}] [{cmd:,}
{cmd:cmd(}{it:string}{cmd:)}
{cmdab:ren:ame}
[{ul:{cmd:no}}]{cmdab:e:sample:}[{cmd:(}{it:varname}{cmd:)}]
{it:ereturn_post_opts}
]
{p 4 4 2}
where {it:b} is a 1 x p coefficient vector
(matrix) and {it:V} is a p x p covariance matrix.
{title:Description}
{p 4 4 2}
{cmd:erepost} changes the {cmd:b} or {cmd:V} matrix of the current estimation results
or changes the declared estimation sample. {cmd:erepost} is similar to
{helpb ereturn repost}. However, {cmd:erepost} is allowed
after estimation commands that do not post their results
using {cmd:ereturn post} (e.g. {cmd:logit}) and {cmd:erepost}
can be used outside of {cmd:eclass} programs (see help {helpb program}).
{p 4 4 2}
Technical note: After applying {cmd:erepost} the original command (or
{cmd:estimates replay}) may not be able to replay the output.
{title:Options}
{p 4 8 2}
{cmd:cmd(}{it:string}{cmd:)} sets the {cmd:e(cmd)} macro.
{p 4 8 2}
{cmd:rename} causes the names from the coefficient vector to be used
as the labels for both the coefficient vector and the covariance matrix
in case of name conflicts.
{p 4 8 2}
{cmd:esample(}{it:varname}{cmd:)} gives the name of the 0/1 variable
indicating the observations involved in the estimation. The variable is
removed from the data but is available for use as {hi:e(sample)}. Specify {cmd:noesample}
if you want to remove {cmd:e(sample)} from the estimation results.
{p 4 8 2}
{it:ereturn_post_opts} are any other options allowed with {cmd:ereturn post}
(see help {helpb ereturn}).
{title:Examples}
{p 4 4 2}Say, you are estimating a regression model that contains all interactions between the
regressors and a binary variable and you want to display the main effects and the
interaction effects in two columns side by side. One approach is to modify the names
in {cmd:e(b)} and {cmd:e(V)} so
that there is a "main equation" and an "interaction equation" and then tabulate
the model using {cmd:estout} or {cmd:esttab} with the {cmd:unstack} option. Example
(you will need the {cmd:estout} package to run this example; type
{net "describe http://fmwww.bc.edu/repec/bocode/e/estout.pkg":ssc describe estout}):
{com}. sysuse auto
{txt}(1978 Automobile Data)
{com}. generate foreign_mpg = foreign*mpg
{txt}
{com}. generate foreign_weight = foreign*weight
{txt}
{com}. quietly regress price foreign_mpg foreign_weight foreign mpg weight
{txt}
{com}. matrix b = e(b)
{txt}
{com}. matrix coleq b = foreign foreign foreign main main main
{txt}
{com}. matrix colname b = mpg weight _cons mpg weight _cons
{txt}
{com}. erepost b=b, rename
{txt}
{com}. esttab , unstack order(main: foreign:)
{res}
{txt}{hline 44}
{txt} (1)
{txt} price
{txt} main foreign
{txt}{hline 44}
{txt}mpg {res} 237.7 -257.5 {txt}
{res} {ralign 12:{txt:(}1.90{txt:)}} {ralign 12:{txt:(}-1.66{txt:)}} {txt}
{txt}weight {res} 4.415*** 0.741 {txt}
{res} {ralign 12:{txt:(}5.18{txt:)}} {ralign 12:{txt:(}0.45{txt:)}} {txt}
{txt}_cons {res} -13285.4* 8219.6 {txt}
{res} {ralign 12:{txt:(}-2.58{txt:)}} {ralign 12:{txt:(}1.13{txt:)}} {txt}
{txt}{hline 44}
{txt}N {res} 74 {txt}
{txt}{hline 44}
{txt}t statistics in parentheses
{txt}* p<0.05, ** p<0.01, *** p<0.001
{title:Author}
{p 4 4 2}Ben Jann, University of Bern, jann@soz.unibe.ch
{p 4 4 2}Thanks for citing this software as follows:
{p 8 8 2}Jann, B. (2007). erepost: Stata module to repost the estimation results. Available from
{browse "http://ideas.repec.org/c/boc/bocode/s456850.html"}.
{title:Also see}
{psee}
Online: {helpb ereturn}, {help estcom},
{helpb estimates}