ParetoAdj | R Documentation |
Estimate shrunken Pareto-optimal solution based on Pareto-optimal shrinkage formulas
ParetoAdj(Ncal, wpred, Rperf_cal, Rrace_cal, prop, sr)
Ncal |
Calibration sample size |
wpred |
Matrix of predictor weights, where each row displays a Pareto-optimal solution and each column displays a weight for each predictor in that solution |
Rperf_cal |
Vector of calibration sample job performance validity |
Rrace_cal |
Vector of calibration sample race bivariate correlation (i.e., correlation between race dummy variable (0-minority, 1-majority) and predictor composite score) |
prop |
Proportion of minority applicants in full applicant pool |
sr |
Selection ratio |
A data frame, each row shows a Pareto-optimal solution prior to and after the adjustment using the Pareto-optimal shrinkage formula
Job performance validity prior to and after adjustment, respectively
Race bivariate correlation prior to and after adjustment, respectively
Adverse impact ratio prior to and after adjustment, respectively
The extent to which, respectively, job performance validity and race bivariate correlation is optimized in a given Pareto-optimal solution
# Example corresponds to Song, Tang, Newman, & Wee (2023), Supplemental Material 4
# Song, Q. C., Tang, C., Newman, D. A., & Wee, S. (2023). Adverse impact reduction
# and job performance optimization via Pareto-optimal weighting: A shrinkage formula
# and regularization technique using machine learning. Journal of Applied
# Psychology. doi:10.1037/apl0001085.
# (1) Calibration sample size
Ncal <- 40
# (2) Predictor weights for each Pareto solution.
# Rows: Pareto solutions; Columns: Predictors
wpred <- matrix(c(0, 0, 0, 0, 1,
0, 0, 0, 0.07, 0.93,
0, 0, 0, 0.13, 0.87,
0, 0, 0, 0.18, 0.82,
0, 0, 0, 0.23, 0.77,
0, 0, 0, 0.28, 0.72,
0, 0, 0, 0.32, 0.68,
0, 0, 0, 0.37, 0.63,
0, 0, 0, 0.41, 0.59,
0, 0.01, 0, 0.45, 0.55,
0, 0.04, 0, 0.44, 0.52,
0, 0.07, 0, 0.43, 0.5,
0, 0.1, 0, 0.42, 0.48,
0, 0.14, 0, 0.41, 0.46,
0, 0.17, 0, 0.4, 0.43,
0, 0.21, 0, 0.38, 0.41,
0, 0.25, 0, 0.37, 0.38,
0, 0.3, 0, 0.35, 0.35,
0, 0.33, 0.03, 0.34, 0.3,
0, 0.37, 0.08, 0.31, 0.25,
0, 0.42, 0.15, 0.27, 0.15), ncol = 5, nrow = 21)
# (3) Vector of calibration sample job performance validity
Rperf_cal = c(.20, .24, .27, .30, .33, .36, .39, .42, .45, .48, .51, .54,
.57, .60, .63, .66, .69, .72, .74, .76, .78)
# (4) Vector of calibration sample race bivariate correlation
# (i.e., correlation between race dummy variable (0-minority, 1-majority)
# and predictor composite score)
Rrace_cal = c(-.12, -.11, -.10, -.10, -.09, -.08, -.07, -.06, -.05, -.03,
-.02, .00, .01, .03, .05, .07, .09, .12, .15, .19, .24)
# (5) proportion of minority
prop <- 1/6
# (6) selection ratio
sr <- .15
# Estimate shrunken Pareto-optimal solution
ParetoAdj(Ncal = Ncal, wpred = wpred, Rperf_cal = Rperf_cal,
Rrace_cal = Rrace_cal, prop = prop, sr = sr)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.