Building a Model to Predict Caseworker and Supervisor Turnover Using a Neural Network and Logistic Regression

Abstract

Human service professionals are increasingly pressured to use sophisticated data analysis tools to support service decisions. However, the application of these tools often involves assumptions and nuances that are difficult for the practitioner to evaluate without specialized information. This article helps the practitioner evaluate two different quantitative methods, a logistic regression and a neural network. Both were used on the same data set to develop a model for predicting employee turnover in a regional child protective services agency. The different steps of building and enhancing the model were discussed. Ultimately, the neural network was able to predict turnover more accurately than a logistic regression by only 1%. The article provides advice to practitioners on comparing, evaluating, and interpreting logistic and neural network tools.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 30 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.