Premium
Assessing the impact of influential observations on multiple regression analysis in human resource research
Author(s) -
Bates Reid A.,
Holton Elwood F.,
Burnett Michael F.
Publication year - 1999
Publication title -
human resource development quarterly
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.756
H-Index - 65
eISSN - 1532-1096
pISSN - 1044-8004
DOI - 10.1002/hrdq.3920100406
Subject(s) - outlier , regression analysis , regression , human resources , resource (disambiguation) , psychology , econometrics , statistics , data science , computer science , economics , management , mathematics , computer network
An overlooked aspect of multiple regression research using employee‐provided data is the potential for microlevel bias that can influence regression results. This article demonstrates the potential impact of influential observations using a case study and outlines an improved procedure for identifying which influential observations to delete. Because human resource research frequently asks employees to provide information about themselves or others, the potential for microlevel bias is almost always present. Increased attention to evaluating influential observations and outliers is suggested.