Chi-square test in SPSS: a guide for your questionnaire
Run a chi-square test of independence in SPSS on your questionnaire data: assumptions, expected counts, Cramer's V, interpretation and how to report it.
Data Académique
10/13/20263 min read
"Is there a link between sex and social media use?", "Does education level influence knowledge of the disease?" As soon as you cross two categorical variables from your questionnaire, the most widely used test is the chi-square test of independence (χ²). Here is how to run it in SPSS and, above all, how to interpret and report the result.
In short: the chi-square test checks whether there is a link between two categorical variables (for example sex and product choice). In SPSS: Analyze, Descriptive Statistics, Crosstabs, then tick Chi-square under Statistics. If the asymptotic significance is below 0.05, the link is significant, provided the expected counts are large enough.
When should you use the chi-square test?
The chi-square test checks whether there is an association between two categorical variables. Examples:
sex (male/female) and purchase intention (yes/no);
age group and preferred social network;
length of service (under 5 years / 5 years or more) and knowledge level (low / medium / good).
If one of the variables is numeric (age in years, a score out of 20), you need another test (t-test, ANOVA, correlation…).
The hypotheses of the test
H0 (null hypothesis): the two variables are independent; there is no link.
H1: the two variables are related.
H0 is rejected if the p-value (significance) is below the chosen threshold, usually 0.05.
Running the chi-square test in SPSS
Go to Analyze > Descriptive Statistics > Crosstabs.
Put one variable in Row(s) and the other in Column(s). By convention, the explanatory variable goes in rows.
Click Statistics, tick Chi-square. To measure the strength of the link, also tick Phi and Cramer's V. Continue.
Click Cells and tick Observed, Expected and Row percentages. Continue.
Click OK.
Check the expected count condition
The chi-square test is only reliable if the expected counts are not too small. The usual rule: at least 80% of cells must have an expected count of 5 or more, and no cell below 1.
SPSS shows it below the "Chi-Square Tests" table: "a. x cells (y%) have expected count less than 5". If y exceeds 20%:
for a 2×2 table, read Fisher's exact test (SPSS gives it automatically);
for a larger table, merge similar categories (for example "46-55" and "over 55") and run the test again.
Reading the results
In the "Chi-Square Tests" table, look at the Pearson Chi-Square row:
Value: the χ² statistic;
df: degrees of freedom, (rows − 1) × (columns − 1);
Asymptotic Significance (2-sided): this is the p-value.
If p < 0.05, there is a statistically significant association. Then look at Cramer's V for the strength of the link: around 0.1 weak, 0.3 moderate, 0.5 and above strong (Cohen's benchmarks, to adapt to the degrees of freedom).
Finally, go back to the crosstab percentages to say in which direction the link goes: that is what really interests the jury.
How to report the result
An example wording (fictitious data):
"The chi-square test shows a significant association between sex and purchase intention, χ²(1, N = 210) = 8.42, p = 0.004, Cramer's V = 0.20. Women report purchase intention more often (62%) than men (42%)."
If the result is not significant: "No significant association was observed between length of service and knowledge level, χ²(2, N = 150) = 3.11, p = 0.211."
Mistakes to avoid
Using chi-square with a continuous numeric variable without grouping it into classes.
Forgetting to check expected counts.
Writing "sex causes purchase intention": chi-square shows an association, not causality.
Pasting the raw SPSS table into the thesis. Make a clean, readable table instead.
The complete data analysis path
For a questionnaire-based thesis, the steps follow this order:
cross categorical variables with the chi-square test;
test the effect of several variables with linear regression;
for models with latent variables: SmartPLS and PLS-SEM.
Official software documentation: IBM SPSS Statistics.
Frequently asked questions
What sample size is needed for a chi-square test?
There is no absolute minimum, but the smaller the sample, the more likely expected counts fall below 5. See our article on sample size.
Chi-square or Fisher's test?
Fisher's exact test is preferable for a 2×2 table with small expected counts. Otherwise, Pearson's chi-square is the reference.
Chi-square or regression: which should I choose?
Chi-square relates two categorical variables. To explain a quantitative variable (or a score) with several variables, use linear regression.
Put it into practice
Our SPSS course has you practice chi-square, t-tests, ANOVA and correlations on thesis data. Need help with your own analysis? Discover our thesis support or message us on WhatsApp.
