Intro to Statistics at IU Indy - Two Sample t-Tests: Independent-Samples
Overview
Andy Baker explains the independent-samples t-test for comparing the means of two unrelated groups, covering hypotheses, standard error, degrees of freedom, critical values, and confidence intervals. Worked examples compare stress scores by work hours, where t = 5.75 leads to rejecting equal means, and a second pair of groups, where t = -0.92 and an interval containing zero leads to failing to reject.
Key takeaways
- For two independent groups, the standard error of the mean difference is calculated as √(s₁²/n₁ + s₂²/n₂), combining each group's variance divided by its sample size.
- The lesson calculates degrees of freedom as n₁ + n₂ - 2, then uses the two-tailed α = 0.05 column of a Student's t table to find critical values.
- In the stress example, the 20-or-more-hours group averaged 7.56 versus 3.25 for the under-20-hours group; t ≈ 5.75 exceeds the 2.101 critical value.
- In the second example, the mean difference of -1.5 and t ≈ -0.92 do not cross the ±2.145 critical values, and the approximate interval [-4.76, 1.76] contains zero.
- The confidence-interval calculations in the lesson use a simplified margin of approximately 2 standard errors; an exact 95% t interval uses the relevant t critical value multiplied by the standard error.
- Failing to reject the null means the sample does not provide clear evidence of a difference; it does not prove that the population means are equal.
Chapters
- The test asks whether the mean of one independent group differs statistically from the mean of another.
- Example comparisons include group versus solo studying, college graduates versus non-graduates, and first-year versus senior students.
- The null hypothesis states that the population means are equal; the two-tailed alternative states that they differ.
- A 95% confidence interval for the mean difference supports the decision: if it includes zero, fail to reject the null; if not, reject it.
- The first example compares stress scores for people working 20 or more hours, n = 9, mean = 7.56, SD = 1.33, with those working under 20 hours, n = 12, mean = 3.25, SD = 2.09.
- The difference in sample means is 7.56 - 3.25 = 4.31.
- The independent-groups standard error is calculated as the square root of s₁²/n₁ + s₂²/n₂, giving approximately 0.749.
- Degrees of freedom are calculated as n₁ + n₂ - 2 = 19; the two-tailed critical value at α = 0.05 is approximately ±2.101.
- Dividing the mean difference, 4.31, by the standard error, 0.749, gives t ≈ 5.75, beyond the positive critical value of 2.101.
- The result supports rejecting the null hypothesis: the two work-hour groups have statistically different sample stress means at the 5% significance level.
- Using the lesson's approximate 95% interval method of mean difference ± 2 standard errors gives about [2.812, 5.808], which excludes zero.
- The higher sample mean, 7.56 versus 3.25, is consistent with greater reported stress among those working 20 or more hours.
- The second example compares groups with means 12 and 13.5, both with n = 8; their standard deviations are 3 and 3.5.
- The mean difference is 12 - 13.5 = -1.5, and the standard error is approximately 1.630.
- With 14 degrees of freedom, the two-tailed critical values are approximately ±2.145; t ≈ -0.92 falls between them, so the test fails to reject the null.
- The approximate interval, [-4.76, 1.76], includes zero, providing consistent evidence that the sample does not establish a difference between the group means.
Transcript excerpt
0:00 Last week we looked at the one sample t-test. This week we're going to look at the two sample t-test. And so the key thing is only two. There's two different types. The first one we're going to look at is differences amongst two different groups, independent groups of each other. So examples, do students who study in groups score higher on exams than students who study alone? Is the average income different for people with a college degree versus those without a college degree? Same job. See if this difference in terms of men, how happy are they at their job versus women, how happy are they at their job? Or is the average number of hours spent on social media different for first-year
0:48 students versus senior students? So these are examples of where we have two different groups and now we can test to see if there's any differences between them. And so that's the whole idea here. The whole goal of this particular t-test is to try see is there is the mean of one group meaningfully or statistically different than the mean of another group? So in the case of the independent samples t-test, the null hypothesis then would be that the mean of the one is going to be the same as the mean of the other. So that would be the null. And so mu, mean of one,
1:35 is going to be same as the mean of the other sample. That means the null hypothesis or our research hypothesis or experimental hypothesis will then be the means are not the same. And so what's going to happen is just like we did with the one sample t-test, we're going to get our critical value and then we're going to run a t-test and compare that t-test to
This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.
Related notes
Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, Andy Baker.