Z-Test vs. Chi-Square: When to Use Each Statistical Test
Z-Test compares a sample mean to a known population mean when data are continuous and the variance is known; Chi-Square tests whether two categorical variables are independent by comparing observed versus expected counts.
People confuse them because both evaluate “significance,” but one deals with averages of numbers (heights, dollars) and the other with counts of labels (yes/no, red/blue). Picture a marketer asking, “Did our ad boost average spend?” (Z-Test) versus “Does ad exposure change the proportion of buyers?” (Chi-Square).
Key Differences
Z-Test needs numeric data, known σ, and n ≥ 30 or normal population. Chi-Square works with frequency tables, expects ≥ 5 counts per cell, and is distribution-free. Z outputs a z-score; Chi-Square outputs χ² with degrees of freedom.
Which One Should You Choose?
If your variable is continuous and you have the population standard deviation, pick Z-Test. If you have categorical data arranged in a contingency table, pick Chi-Square. When in doubt, check the data type: numbers → Z, labels → Chi-Square.
Examples and Daily Life
Testing if the average delivery time dropped below 30 minutes? Z-Test. Checking if smartphone brand preference differs by age group? Chi-Square. A café owner uses Z-Test for latte temperature, Chi-Square for syrup flavor choices.
Can I use Chi-Square for small samples?
No; expected cell counts should be ≥ 5. Otherwise, use Fisher’s exact test.
What if population σ is unknown?
Use a t-test instead of a Z-Test; Chi-Square remains unchanged.
Do these tests prove causation?
Neither test proves cause; they only flag association or difference.