ATLASResearch
methods
Quantitative/ Analysis

Chi-square tests

Test categorical counts against an expected pattern

Pearson chi-square procedures compare observed counts with expected counts under independence or a specified distribution. Tests of association use contingency tables; goodness-of-fit tests compare one categorical distribution with a target. Independent sampling and adequate expected counts underpin the usual approximation. Matched binary observations require a paired method such as McNemar’s test.

WHEN IT FITS

Use to examine association between categorical variables or fit to a prespecified distribution. Retain counts and denominators, and establish whether respondents, records or events are independent.

Strengths

  • Works directly with categorical counts
  • Accommodates tables larger than two by two

Limitations

  • Sparse expected cells weaken the asymptotic approximation
  • Significance does not convey association magnitude or direction

Know the boundary

The source’s word qualitative means categorical variables, not interpretive qualitative analysis.

USED ACROSS
Banking & financeBusiness & MBAPsychologyComputer scienceEducation