Perceiving correlations from scatterplots

Joachim Meyer*, David Shinar

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

Abstract

This study investigated the effects of statistical training and various perceptual characteristics of scatterplot displays on intuitive estimates of correlations. University professors and first year undergraduate students estimated correlations from scatterplots (1) with three different levels of correlations in the data, (2) with or without a regression line, and (3) with three different types of dispersion of the data point clouds. The faculty generally used higher and a wider range of values, but both groups perceived a higher correlation when a regression line was present and both groups were equally influenced by the different types of dispersion of the point cloud. These findings indicate that the estimation of correlations from scatterplots is a perceptually based process, which is largely independent from formal statistical training.

Original languageEnglish
Pages (from-to)1537-1540
Number of pages4
JournalProceedings of the Human Factors Society
Volume2
DOIs
StatePublished - 1991
Externally publishedYes
EventProceedings of the Human Factors Society 35th Annual Meeting Volume 2 (of 2) - San Francisco, CA, USA
Duration: 2 Sep 19916 Sep 1991

Fingerprint

Dive into the research topics of 'Perceiving correlations from scatterplots'. Together they form a unique fingerprint.

Cite this