Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies.

Austin PC. · 2011

Read Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. by Austin PC. (2011) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3120982.txt.

Description

Peter Austin settled a practical question for health researchers in a 2011 paper in Pharmaceutical Statistics. When treatments are not assigned by chance, researchers pair treated and untreated people with similar predicted chances of treatment, so groups compare fairly. How close those paired scores must be, the matching width, had little guidance behind it.

Austin ran wide-ranging computer simulations for both whole-number and yes-or-no outcomes. A width set at one fifth of the spread of the transformed score worked best or near best in most settings. It removed nearly all of the starting bias and gave error ranges with the right coverage. When all pairing factors were yes-or-no traits, the exact width mattered far less. His rule gives analysts a firm default.

Catalog record

CREATORAustin PC.
YEAR2011
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3120982

Browse the full library · About the Tower of Babel