• Skip to primary navigation
  • Skip to main content
  • Skip to primary sidebar
The Analysis Factor

The Analysis Factor

Statistical Consulting, Resources, and Statistics Workshops for Researchers

  • Home
  • About
    • Our Programs
    • Our Team
    • Our Core Values
    • Our Privacy Policy
    • Employment
    • Guest Instructors
  • Membership
    • Statistically Speaking Membership Program
    • Login
  • Workshops
    • Online Workshops
    • Login
  • Consulting
    • Statistical Consulting Services
    • Login
  • Free Webinars
  • Contact
  • Login

Chi-Square Test of Independence Rule of Thumb: n > 5

by Audrey Schnell Leave a Comment

We all want rules of thumb even though we know they can be wrong, misleading or misinterpreted.

Rules of Thumb are like Urban Myths or like a bad game of ‘Telephone’.  The actual message gets totally distorted over time.
For example, you may have heard this one: “The Chi-Square test is invalid if we have fewer than 5 observations in a cell”.

I frequently hear this mis-understood and incorrect statement.
The correct statement is not about the observed in each cell. Those can be less than 5. It’s the EXPECTED count that needs to be >5 per cell*.

Why is this important?

The Chi-square statistic follows a chi-square distribution asymptotically with df=n-1.  That means we can use the chi-square distribution to calculate an accurate p-value only for large samples. (That’s where the asymptotically comes in). For small samples, it doesn’t work.

How large of a sample is needed?

One that is large enough that the expected value for each cell is at least 5. Those expected values come from the total sample size, and the corresponding total frequencies of each row and column. So if any row or column totals in your contingency table are small, or together are relatively small, you’ll have an expected value that’s too low.

Software packages typically supply a warning when this occurs. For example, it’s the criterion SAS uses.

Look at the table below, which shows observed counts between two categorical variables, A and B. The observed counts are the actual data. You can see that out of a total sample size of 48, 28 are in the B1 category and 20 are in the B2 category.

Likewise, 33 are in the A1 category and 15 are in the A2 category. Inside the box are the individual cells, which give the counts for each combination of the two A categories and two B categories.

The Expected counts come from the row totals, column totals and the overall total, 48. For example, in the A2, B1 cell, we expect a count of 8.75. It is an easy calculation: (Row Total * Column Total)/Total. So (28*15)/48.

The more different the observed and expected counts are from each other, the larger the chi-square statistic.

Notice in the Observed Data there is a cell with a count of 3. But the expected counts are all >5. If the expected counts are less than 5 then a different test should be used (e.g. Fisher’s Exact Test).

But is 5 the true minimum?

There are other suggested guidelines too.

According to Cochran (1952, 1954), all expected counts should be 10 or greater. If < 10, but >=5, Yates’ Correction for continuity should be applied.

More recent standards for a 2 x 2 Table (Campbell 2007) say Fisher’s Exact and Yates Correction are too conservative and proposes alternative tests depending on the study design.

For tables larger than 2 x 2 Yates, Moore & McCabe (1999), state “No more than 20% of the expected counts should be less than 5 and all individual expected counts should be greater or equal to 1. Some expected counts can be <5, provided none <1, and 80% of the expected counts should be equal to or greater than 5.”

The Minitab manual criteria are: If either variable has only 2 or 3 categories, then either
— all cells must have expected counts of at least 3 or
— all cells must have expected counts of at least 2 and 50% or fewer have expected counts below 5
If both variables have 4 to 6 levels then either:
— all cells have expected counts of at least 2, or
— all cells have expected counts of at least 1 and 50% or fewer cells have expected counts of < 5

In summary, different sources have different criteria for the Chi-square test to be valid. All criteria refer to the expected cell counts and not the observed data.

There are alternative tests for small cell counts such as Fisher’s Exact Test and Yates correction. Lastly, sometimes it might be necessary to collapse across categories to obtain adequate cell counts.

*The ‘5’ probably came from Fisher.

Standard Non-Deviation: The Steps to Running Any Statistical Model
Get the road map for your data analysis before you begin. Learn how to make any statistical modeling – ANOVA, Linear Regression, Poisson Regression, Multilevel Model – straightforward and more efficient.

Tagged With: chi-square test, fisher exact test, rules of thumb, sample size, Yates correction

Related Posts

  • Member Training: Seven Fundamental Tests for Categorical Data
  • Member Training: Statistical Rules of Thumb: Essential Practices or Urban Myths?
  • Why Statistics Terminology is Especially Confusing
  • How Big of a Sample Size do you need for Factor Analysis?

Reader Interactions

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Please note that, due to the large number of comments submitted, any questions on problems related to a personal study/project will not be answered. We suggest joining Statistically Speaking, where you have access to a private forum and more resources 24/7.

Primary Sidebar

This Month’s Statistically Speaking Live Training

  • April Member Training: Statistical Contrasts

Upcoming Workshops

  • Logistic Regression for Binary, Ordinal, and Multinomial Outcomes (May 2021)
  • Introduction to Generalized Linear Mixed Models (May 2021)

Read Our Book



Data Analysis with SPSS
(4th Edition)

by Stephen Sweet and
Karen Grace-Martin

Statistical Resources by Topic

  • Fundamental Statistics
  • Effect Size Statistics, Power, and Sample Size Calculations
  • Analysis of Variance and Covariance
  • Linear Regression
  • Complex Surveys & Sampling
  • Count Regression Models
  • Logistic Regression
  • Missing Data
  • Mixed and Multilevel Models
  • Principal Component Analysis and Factor Analysis
  • Structural Equation Modeling
  • Survival Analysis and Event History Analysis
  • Data Analysis Practice and Skills
  • R
  • SPSS
  • Stata

Copyright © 2008–2021 The Analysis Factor, LLC. All rights reserved.
877-272-8096   Contact Us

The Analysis Factor uses cookies to ensure that we give you the best experience of our website. If you continue we assume that you consent to receive cookies on all websites from The Analysis Factor.
Continue Privacy Policy
Privacy & Cookies Policy

Privacy Overview

This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary
Always Enabled

Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.

Non-necessary

Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.

SAVE & ACCEPT