A quasi-experimental design for studies on the impact of administrative decisions: Applications and extensions of the regression-discontinuity design

In the regression-discontinuity (RD) design, scores on a pretreatment measure (assignment variable) determine assignment to groups such that respondents above a certain score are assigned to one group, whereas respondents below the score are assigned to another group (e.g., candidates for promotion above a cutoff point on a criterion dimension are promoted, whereas candidates below the cutoff are not). Despite the potential for causal inference about treatment effects, few settings have produced conditions appropriate for applications of the design. These conditions, however, can be found in personnel settings when decisions are based on cutoff procedures. The design is therefore suggested for research on the impact of administrative decisions involving cutoff scores.