= Stata Collapse =

The '''`collapse`''' command creates a dataset of summary statistics.

<<TableOfContents>>

----



== Usage ==

Given a dataset like:

||'''id'''||'''foo'''||'''bar'''||'''baz'''||
||1       ||0        ||0        ||10       ||
||2       ||1        ||0        ||20       ||
||3       ||0        ||0        ||10       ||
||4       ||1        ||0        ||20       ||
||5       ||0        ||1        ||10       ||
||6       ||1        ||1        ||20       ||
||7       ||0        ||1        ||10       ||
||8       ||1        ||1        ||20       ||

Running `collapse (percent) foo (mean) baz, by(bar)` would create a dataset like:

||'''bar'''||'''foo'''||'''baz'''||
||0        ||50       ||15       ||
||1        ||50       ||15       ||

----



== Statistics ==

The `collapse` command is capable of computing any of:

 * `count`: count of non-missing values
 * `first`
 * `firstnm`: first non-missing
 * `iqr`
 * `last`
 * `lastnm`: last non-missing
 * `max`
 * `mean`
 * `median`
 * `min`
 * `pN` where `N` is any integer 1 through 99: `N`th percentile
 * `p50`: alias for `median`
 * `percent`: percent of non-missing values
 * `rawsum`: sum ''ignoring'' weight factors ''except'' when the weight factor is 0
 * `sd`
 * `sebinomial`: standard error as `sd/sqrt(n)`
 * `semean`: standard error as `sqrt(p(1-p)/n)`
 * `sepoisson`: standard error as `sqrt(mean/n)`
 * `sum`

The default statistic to compute is `mean`.

----



== See also ==

[[https://www.stata.com/manuals/dcollapse.pdf|Stata manual for collapse]]



----
CategoryRicottone