On the beta transformation family

Research output: Contribution to journalArticle

16 Scopus citations

Abstract

Several flexible, one-parameter classes of transformations have been proposed for random variables taking values in [0, 11. This article studies a transformation family based on the incomplete beta function that, unlike other families proposed, includes the arcsin squareroot transformation, the logit transformation, and the identity transformation. Applications to regression, correlation analysis, and response-surface analysis are presented. A new method of comparing members of different transformation families is derived, and connections with generalized linear models are discussed. The article concentrates on continuous data rather than binomial data.

Original languageEnglish (US)
Pages (from-to)72-81
Number of pages10
JournalTechnometrics
Volume35
Issue number1
DOIs
StatePublished - 1993

Keywords

  • Incomplete beta function
  • Normalizing transformation
  • Variance-stabilizing transformation

ASJC Scopus subject areas

  • Modeling and Simulation
  • Applied Mathematics
  • Statistics and Probability

Fingerprint Dive into the research topics of 'On the beta transformation family'. Together they form a unique fingerprint.

  • Cite this