Computer-assisted recorded interviewing (CARI) is a boon to researchers in monitoring interviewer performance, but the needed verification and coding can be labor intensive and slow the process down considerably. Can machine learning (ML) help? New research published in Survey Practice, Applying Machine Learning to the Evaluation of Interviewer Performance, notes the answer. Westat’s Hanyu Sun, PhD, a Principal Statistical Associate, and Ting Yan, PhD, a Vice President and Associate Director, coauthored the paper.
The researchers studied a sample of interview recordings to identify in a timely manner falsifications and undesirable interviewer behaviors. They concluded the CARI ML pipeline as robust in effective interviewer monitoring. Plus, while conventional CARI validation methods typically only provide checks for a sample of recordings, 100% of recordings can be processed in real time with ML.