Maximal correlation between normal and dichotomous variables.

Research output: Contribution to journalArticlepeer-review


Calculates the upper limit of the correlation between a supposedly continuous normally distributed variable and a dichotomous variable. It is argued that an empirically obtained correlation should be evaluated in view of this limit, instead of the usual limit of Pearson correlation. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Original languageEnglish GB
Pages (from-to)259-261
Number of pages3
JournalJournal of Educational Statistics
Issue number4
StatePublished - 1 Dec 1986


  • Confidence Limits (Statistics)
  • Point Biserial Correlation
  • Statistical Variables

Cite this