Partial least squares regression in the social sciences

Megan L. Sawatsky, Matthew Clyde, Fiona Meek
2015 The Quantitative Methods for Psychology  
Partial least square regression (PLSR) is a statistical modeling technique that extracts latent factors to explain both predictor and response variation. PLSR is particularly useful as a data exploration technique because it is highly flexible (e.g., there are few assumptions, variables can be highly collinear). While gaining importance across a diverse number of fields, its application in the social sciences has been limited. Here, we provide a brief introduction to PLSR, directed towards a
more » ... ice audience with limited exposure to the technique; demonstrate its utility as an alternative to more classic approaches (multiple linear regression, principal component regression); and apply the technique to a hypothetical dataset using JMP statistical software (with references to SAS software).
doi:10.20982/tqmp.11.2.p052 fatcat:oije6mdfkbcxzmq7bsy6domlfu