_

Brief

Podium Assessment Systems advised a professional services firm looking to increase the accuracy of its selection system, enhance the candidate experience by shortening assessments that were too long, and increase diversity.

Available data included demographic data, source of the candidate, scores on the situational judgment test, and job performance data on the people that were hired. There was a large amount of data available on candidates; several thousand participants in total.

To address diversity, Podium looked at the applicant sources and identified those that showed the highest diversity. Applications that came from one recruiter had a greater proportion of minority applicants. A recommendation was made to use this firm more frequently as it had greater access to universities with diverse demographic populations.

Solution

Podium also looked at the relationship between the situational judgment test scores and job performance. It compared two methods for predictive accuracy, one using simple correlations and a second that used a machine learning method called gradient boosting.

The advanced gradient boosting converged on the conclusions of the simpler correlational analysis in this case. Podium was able to identify and retain the most predictive questions and drop the least predictive, shortening the assessment.

Once the system was operational, higher performing candidates with a stronger representation of minority candidates began to be selected.