Pooled Variance t-Test
What you'll learn: How to compare two group means when both populations have equal variances using a pooled variance approach.
What Is the Pooled Variance t-Test?
When comparing two independent groups (like treatment vs. control, or method A vs. method B), you often need to test whether their means differ significantly. The pooled variance t-test is specifically designed for situations where you can reasonably assume both populations have the same variance (even if their means differ).
Instead of calculating separate variance estimates for each group, this test "pools" (combines) the variance information from both samples into a single, more precise estimate. Think of it like combining data from two thermometers measuring the same phenomenon—you get a more reliable reading.
Why Pool the Variance?
By combining variance estimates, you gain statistical power. You're essentially saying: "Both groups vary the same amount around their respective means, so let's use all available data to estimate that common variability." This pooled estimate uses more information (both samples combined), giving you more degrees of freedom and a more stable test statistic.
When to Use It
- Your samples are independent (not paired or matched)
- You have reason to believe both populations have equal variances
- You're comparing the means of exactly two groups
The pooled approach is what you learned about in "Pooled vs Unpooled Variance Approaches"—it's the practical application when the equal-variance assumption holds. If variances differ substantially, you'd use Welch's t-test (the unpooled version) instead.
Key Takeaway: The pooled variance t-test combines variance information from both samples to create a more powerful test for comparing two means when populations have equal variance.