Two Sample Independent t Tests
So far, we have tested claims about a single population. In many real-world situations, however, we need to compare two different groups.
The two-sample independent t test allows us to determine whether the means of two independent populations are significantly different.
What Does “Independent Samples” Mean?
Two samples are independent if:
- The samples are taken from different groups
- No individual appears in both samples
- One sample does not influence the other
Examples include:
- Test scores of students from two different schools
- Sales figures from two different regions
- Blood pressure of two different patient groups
When Do We Use a Two-Sample t Test?
A two-sample independent t test is used when:
- We compare the means of two independent populations
- Population standard deviations are unknown
- Samples are random and independent
- Sample sizes are reasonably large or data is approximately normal
Setting Up the Hypotheses
Let μ₁ and μ₂ represent the population means of group 1 and group 2.
| Hypothesis | Statement |
|---|---|
| H₀ | μ₁ = μ₂ (no difference) |
| H₁ | μ₁ ≠ μ₂ (difference exists) |
Depending on the context, the alternative hypothesis may also be:
- μ₁ > μ₂
- μ₁ < μ₂
The t Test Statistic
The t statistic measures the difference between sample means relative to the variability in the data.
The general idea is:
t = (Difference in sample means) ÷ (Standard error of the difference)
Deep Numerical Example (Step-by-Step)
A company compares productivity between two teams.
- Team A: n₁ = 30, x̄₁ = 75, s₁ = 8
- Team B: n₂ = 28, x̄₂ = 70, s₂ = 7
- Significance level α = 0.05
Step 1: State the Hypotheses
H₀: μ₁ = μ₂
H₁: μ₁ ≠ μ₂
Step 2: Compute the Difference in Means
x̄₁ − x̄₂ = 75 − 70 = 5
Step 3: Compute the Standard Error
Standard Error =
√( s₁²/n₁ + s₂²/n₂ )
= √( 8²/30 + 7²/28 )
= √( 64/30 + 49/28 )
≈ √(2.13 + 1.75) = √3.88 ≈ 1.97
Step 4: Calculate the t Statistic
t = 5 ÷ 1.97 ≈ 2.54
Step 5: Make the Decision
At α = 0.05 (two-tailed), the critical t value is approximately ±2.00.
Since |2.54| > 2.00:
Decision: Reject the null hypothesis
Interpretation in Plain English
There is statistically significant evidence that the average productivity differs between the two teams.
Why We Use t Instead of Z
In real-world problems, population standard deviations are rarely known.
The t distribution:
- Accounts for extra uncertainty
- Depends on sample size (degrees of freedom)
- Approaches the normal distribution as sample size increases
Common Mistakes to Avoid
- Using paired data with an independent t test
- Ignoring unequal sample sizes
- Confusing statistical significance with importance
- Forgetting to check assumptions
Quick Check
When should a two-sample independent t test be used?
Practice Quiz
Question 1:
What does independence between samples mean?
Question 2:
Why is the t distribution used instead of the normal distribution?
Question 3:
What does rejecting H₀ indicate?
Mini Practice
Two different teaching methods are tested.
- Method A: x̄ = 82, s = 6, n = 25
- Method B: x̄ = 78, s = 5, n = 24
At α = 0.05, test whether the average scores differ.
What’s Next
In the next lesson, we will study Paired t Tests, which are used when observations are naturally linked.