Premium
Intuiting strength of association from a correlation coefficient
Author(s) -
Oakes Mike
Publication year - 1982
Publication title -
british journal of psychology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.536
H-Index - 92
eISSN - 2044-8295
pISSN - 0007-1269
DOI - 10.1111/j.2044-8295.1982.tb01789.x
Subject(s) - bivariate analysis , statistics , correlation , correlation coefficient , psychology , association (psychology) , pearson product moment correlation coefficient , rank correlation , correlation ratio , fisher transformation , mathematics , geometry , psychotherapist
Correlation statistics, although more useful than test statistics, are prone to misinterpretation. In particular, Pearsonian and rank‐order correlation coefficients of moderate size may be derived from bivariate distributions where the strength of association is surprisingly small. Psychologists asked to construct a bivariate distribution exhibiting a correlation of + 0·5 gave systematically inaccurate responses. The median obtained correlation was 0·76. A different sample of psychologists was confronted with a bivariate distribution where the correlation was + 0·5 and gave a median response of 0·25 when asked to estimate the strength of association present. It is concluded that care should be exercised in the interpretation of an obtained value of a correlation coefficient.