Thanks as ever to KitBaum a new package quantilesets is now available from SSC.
The package follows from the update to Stata 19.50 on 12 November 2025 announced here https://www.statalist.org/forums/for...9-and-stata-19 so that it will work for you if and only if you have Stata updated to that date (or one later, as you read this).
quantilesets joins a small family of packages each with a command of the same name,
cisets
pctilesets
momentsets
lmomentsets
posted to SSC at various dates. My talk at the London meeting is September accessible via https://www.stata.com/meeting/uk25/ is perhaps the best way in to understand what is being offered here. If your system can't understand PowerPoint then the Statalist announcements should give enough of a story to decide whether you care,
https://www.statalist.org/forums/for...-interval-sets
https://www.statalist.org/forums/for...ercentile-sets
https://www.statalist.org/forums/for...lable-from-ssc
Turning to quantilesets the essence of the story is this.
quantilesets computes percentile or quantile sets. Typically the user will specify one or more probability levels between 0 and 1 inclusive using the required option prob().
quantilesets differs from the author's pctilesets as follows.
* quantilesets is a wrapper for Mata function quantile() while pctilesets is a wrapper for summarize.
* pctilesets offers no choice of estimation method, while quantilesets offers several methods. That could be important
if you have a strong preference in principle for one method; or if you wish to match a choice made in other software;
or if you wish to compare results obtained with different methods. Often differences will be trivial, but not always,
as with small datasets or those with gaps, spikes or multiple modes.
* pctilesets as a wrapper for summarize offers only selected percentiles, including the minimum and maximum.
quantilesets allows any probability level between 0 and 1 (inclusive) to be specified.
* quantilesets does not support weights, while pctilesets does support aweights and fweights.
The publication of this package has triggered updates to the other packages in the family, with small updates to code to improve consistency of presentation and updates to the help files to improve cross-referencting.
The package follows from the update to Stata 19.50 on 12 November 2025 announced here https://www.statalist.org/forums/for...9-and-stata-19 so that it will work for you if and only if you have Stata updated to that date (or one later, as you read this).
quantilesets joins a small family of packages each with a command of the same name,
cisets
pctilesets
momentsets
lmomentsets
posted to SSC at various dates. My talk at the London meeting is September accessible via https://www.stata.com/meeting/uk25/ is perhaps the best way in to understand what is being offered here. If your system can't understand PowerPoint then the Statalist announcements should give enough of a story to decide whether you care,
https://www.statalist.org/forums/for...-interval-sets
https://www.statalist.org/forums/for...ercentile-sets
https://www.statalist.org/forums/for...lable-from-ssc
Turning to quantilesets the essence of the story is this.
quantilesets computes percentile or quantile sets. Typically the user will specify one or more probability levels between 0 and 1 inclusive using the required option prob().
quantilesets differs from the author's pctilesets as follows.
* quantilesets is a wrapper for Mata function quantile() while pctilesets is a wrapper for summarize.
* pctilesets offers no choice of estimation method, while quantilesets offers several methods. That could be important
if you have a strong preference in principle for one method; or if you wish to match a choice made in other software;
or if you wish to compare results obtained with different methods. Often differences will be trivial, but not always,
as with small datasets or those with gaps, spikes or multiple modes.
* pctilesets as a wrapper for summarize offers only selected percentiles, including the minimum and maximum.
quantilesets allows any probability level between 0 and 1 (inclusive) to be specified.
* quantilesets does not support weights, while pctilesets does support aweights and fweights.
The publication of this package has triggered updates to the other packages in the family, with small updates to code to improve consistency of presentation and updates to the help files to improve cross-referencting.
Comment