Propensity Score Matching for Education Data: Worked Examples

Abstract

Randomized controlled trials are not always feasible in educational research, so researchers must use alternative methods to study treatment effects. Propensity score matching is one such method for observational studies that has shown considerable growth in popularity since it was first introduced in the early 1980s. This paper outlines the concept of propensity scores by explaining their theoretical principles and providing two examples of their usefulness within the realm of educational research. Through worked examples, we highlight the effectiveness of propensity scores as a method for reducing bias and increasing the balance between treatment and comparison groups. To aid in the understanding and future use of propensity scores, we provide R syntax for all our analyses.

Keywords:

• Propensity scores
• matching
• observational study

Notes

Notes

1 Technically, this gives a standardized version of the average treatment effect. In nonexperimental situations, this becomes an intention to treat estimate, so represents the effect of treatment assignment, not whether the participant actually received treatment (West et al., 2014).

2 In the PS literature, calipers are often provided in the logit metric, which is$\mathrm{l}\mathrm{n}\left(\frac{PS}{1-PS}\right)$, where ln is the natural logarithm function.

3 Thoemmes and West (2011) is an excellent resource for using propensity score techniques with nested data.