Summary
In HR Analytics research, hypothesis testing and prediction differ in purpose and procedure, yet the two are often mixed together. This paper explains how the two approaches differ from a statistical point of view and how the method of analysis should change with the purpose, and shows the difference with real data.
The case data are Assessment Center results from Company A, a Korean firm. We analysed the overall assessment rating (OAR) and individual performance of 202 sales manager candidates who took part in the 2015 manager promotion qualification program, using an ordinal logistic regression model. For hypothesis testing, we examined the relationships among variables with a model fixed before the analysis; for prediction, we compared eight candidate models by cross-validation, splitting the data into training data (162 people) and validation data (40 people).
The case showed that a hypothesis-testing model has to be set before the analysis whereas a prediction model is selected during the analysis, and that predictive power cannot be known from the p-value alone. This summary is based on the accepted manuscript.
Citation
이중학, Kim, S., 송지훈, 채충일 (2020). HR Analytics 연구 및 활용에서의 가설 검정과 예측의 차이점: Assessment Center 사례를 중심으로. 조직과 인사관리연구, 44(2), 103–123. https://doi.org/10.36459/jom.2020.44.2.103
Indexed in: KCI
Citation
@online{2020,
author = {},
title = {HR {Analytics} {연구} {및} {활용에서의} {가설} {검정과}
{예측의} {차이점:} {Assessment} {Center} {사례를} {중심으로}},
date = {2020-05-30},
url = {https://chadchae.github.io/posts_publication/2020-05-30-hypothesis-testing-vs-prediction-hr-analytics/hypothesis-testing-vs-prediction-hr-analytics.html},
langid = {en}
}