The relative importance analysis for the study of the family: Accepting the challenge of correlated predictors

Daniela Barni

Research output: Contribution to journalArticlepeer-review

16 Citations (Scopus)

Abstract

Family researchers are often interested in the importance of variables to be included in the prediction of some outcome, and have traditionally used multiple regression analysis (MR) to study variable importance. However, given the non-independence of family data, multicollinearity can be a pervasive phenomenon in this field of research and MR may produce poor or confusing results in the case of highly correlated predictors. In this article I present two alternative techniques of analysis — relative weight analysis (RWA; Johnson, 2000) and dominance analysis (DA; Budescu, 1993) — both of which address questions pertaining to predictor comparisons and are able to provide a plausible assessment of importance among multiple correlated predictors. Examples of application of RWA and DA to address specific questions about family relationships are discussed.
Original languageEnglish
Pages (from-to)235-250
Number of pages16
JournalTPM. TESTING, PSYCHOMETRICS, METHODOLOGY IN APPLIED PSYCHOLOGY
Volume22
DOIs
Publication statusPublished - 2015

Keywords

  • Family relationships
  • Multicollinearity
  • Relative importance

Fingerprint Dive into the research topics of 'The relative importance analysis for the study of the family: Accepting the challenge of correlated predictors'. Together they form a unique fingerprint.

Cite this