Researchers from the University of Minnesota’s Carlson School of Management are using machine learning to help make hiring predictions about teaching applicants’ job performance and potential turnover.

By combining psychological and economic theory, historical data and machine learning, the team attempted to create an automated process to assist human resource professionals and school administrators screen job candidates — a process that is largely based on trial and error.

According to the research team, candidates use different ways to describe past work experiences in job applications and resumes. However, how those words are interpreted by hiring personnel might change from day to day or from applicant to applicant. As such, researchers believe that machine learning could potentially be used to offer an objective, automated and auditable way for translating the hiring candidate’s words into quantifiable outcomes that could be used to make predictions about an applicant’s estimated length of tenure and future performance.

To demonstrate, the team examined data from more than 16,000 candidates who had applied for teaching positions within the Minneapolis Public School District between 2001 and 2013. The research team paid particular attention to the assortment of retention and effectiveness measures for the 2,225 candidates hired during that period, looking at how well the skills, abilities and knowledge matched a candidate’s previous job using Department of Labor (DOL) data. The team also examined a candidate’s tenure history and any reasons the candidate had listed for leaving their former positions.

Based on that data, researchers determined that candidates with more relevant work experience were likely more effective teachers. Likewise, those candidates with shorter tenures at their previous positions reportedly performed more poorly on the job and generally left their positions sooner. Meanwhile, those that had listed that they were leaving their former position to seek out a better position as their reason for leaving their previous employer tended to perform better in all categories, according to the research team.

The team believes that applied to hiring decisions in the education sector, the machine learning benefits will effectively predict which candidates will be better performers and will last longer.

“Most machine learning emphasizes pure automated prediction from a mass of data,” said John Kammeyer-Mueller, Carlson School professor and director of the Center for Human Resources and Labor Studies. “This theoretical approach means it’s hard to know why some people are better hires. Because we started from a model of experience and motivation developed in prior research, we were able to use the power of big data in a way that organizational leaders can easily interpret and understand.”

The team’s findings appear in the Journal of Applied Psychology.