Propensity-Rating Matching Is the Bedrock of Causal Inference | by Ari Joury, PhD | Dec, 2024

OPINION

And find out how to get began with it utilizing Python

Can coaching packages trigger extra financial prosperity? Picture generated with Leonardo AI

A lot of latest information science solutions the query “What’s happening?” At my agency, for instance, we frequently attempt to spot how properly an organization is performing, and the way one efficiency indicator is tied to a different by way of correlations.

A extra highly effective query value answering can be “Why is that this occurring?” For instance, if we detect a major correlation between the presence of ladies in administration and an organization’s revenues, what’s trigger and what’s impact right here? Or, if individuals bear a coaching program, will this trigger their efficiency to enhance? Or would better-performing individuals need to bear a coaching program, and therefore we solely see an impact as a consequence of choice bias?

A number of approaches exist to pinpoint causal relationships in information science. Propensity-Rating Matching (PSM) is likely one of the older ones, having emerged round 40 years in the past. Different strategies like Structural Equation Modeling arose on the identical time. Approaches like Instrumental Variables arose a number of many years earlier than. Causal statistics remains to be a really lively area, with many new strategies being developed.

A key benefit of PSM is that it permits researchers to work with real-world information. Particularly, it…