STAT 3011 Fall 2026: Hypothesis Testing Final
Overview
Prof Kazeem Adepoju explains how to choose and carry out hypothesis tests for population proportions and means, including how the research question determines a left-tailed, right-tailed, or two-tailed alternative. He distinguishes proportion Z tests, mean Z tests when population standard deviation is known, and mean t tests when it is unknown, then works through examples using p-values, significance levels, and contextual conclusions.
Key takeaways
- For a population mean, use a Z test when population standard deviation σ is known and a t test when σ is unknown and must be estimated with sample standard deviation s.
- For a one-sample t test, degrees of freedom equal n − 1; a sample of 35 therefore uses 34 degrees of freedom.
- For a population proportion, use a Z test and check the null-based counts np₀ and n(1 − p₀); Adepoju uses 15 as the minimum for each condition.
- Translate wording into the alternative hypothesis: “lower” means left-tailed, “higher” or “exceed” means right-tailed, and “differs” means two-tailed.
- Match the p-value tail to the alternative: use one tail for directional tests and double the appropriate tail probability for a two-tailed test.
- When p-value < α, reject H₀ and describe evidence in the alternative’s direction; when p-value > α, fail to reject H₀ rather than claiming H₀ has been proven.
Chapters
- Adepoju contrasts a mean Z test, which uses known population standard deviation σ, with a t test, which uses sample standard deviation s when σ is unknown.
- The t statistic uses the sample mean, hypothesized population mean, sample standard deviation, and sample size: (x̄ − μ₀)/(s/√n).
- A t test assumes random, independent observations; small samples require an approximately normal population, while samples of at least 30 can use the Central Limit Theorem.
- A sample of 35 observations is used to test whether mean calories differ from 130, with a two-sided alternative.
- The test statistic is t = 1.392 and the degrees of freedom are n − 1 = 34.
- Using a two-tailed p-value of 0.173 and α = 0.05, the decision is to fail to reject H₀; the sample does not provide sufficient evidence that the mean differs from 130.
- Adepoju introduces a Canvas worksheet with examples of hypothesis testing and emphasizes translating a research question into hypotheses before calculating a test statistic.
- He notes that hypothesis testing and confidence intervals are major topics for Exam 2 and encourages students to identify the test from the question rather than expect it to be named.
- A prior report gives a food-insecurity proportion of 28%; a random sample of 400 University of Minnesota students has 92 reports.
- Because the research claim is that the proportion is lower, the hypotheses use p₀ = 0.28 and a left-tailed alternative, p < 0.28.
- The proportion test uses a Z statistic with the null-based standard error; the stated condition checks are np₀ = 112 and n(1 − p₀) = 288, both at least 15.
- At α = 0.05, the example rejects H₀ and concludes there is evidence the student proportion is below 28%; the estimated difference is about 5 percentage points, whose practical importance is separate from statistical significance.
- A survey benchmark of 45% is compared with a sample of 350 students, of whom 189 regularly use AI tools for coursework.
- The claim that the student proportion is higher sets the alternative to p > 0.45, making this a right-tailed proportion Z test.
- The example uses α = 0.01 and compares the right-tail p-value with that threshold; the stated conclusion is evidence that more than 45% regularly use AI tools.
- A public-transportation example compares a benchmark of 40% with 225 of 500 students using transit at least three times weekly.
- The word “differs” signals a two-tailed alternative, p ≠ 0.40; “lower” and “higher” instead indicate left- and right-tailed tests.
- Adepoju summarizes that proportion tests use Z tests and stresses writing hypotheses with population parameters such as p, not sample statistics such as p̂.
- For sleep duration, a historical mean of 7 hours and known σ = 1.4 are compared with a sample of 100 students averaging 6.65 hours.
- The sleep claim is left-tailed; the Z statistic is −2.5, and the example rejects H₀ at α = 0.05 in favor of a mean below 7 hours.
- For tutoring scores, the benchmark is 72, known σ = 12, and n = 64 with sample mean 75.5; the claim that scores are greater than 72 sets a right-tailed Z test.
- The tutoring-score example calculates Z = 2.333 and uses a right-tail p-value for the claim that the population mean exceeds 72.
- A smartphone-use example compares a historical mean of 4.5 hours and known σ = 1.8 with 144 students averaging 4.9 hours.
- Because the screen-time question asks whether the mean differs from 4.5, its alternative is two-sided and the p-value is twice the probability beyond the absolute Z statistic.
- A hospital waiting-time example compares a 50-minute benchmark with a random sample of 36 visits averaging 45.2 minutes; only a sample standard deviation is supplied.
- Unknown population σ makes this a t test, with statistic (x̄ − μ₀)/(s/√n) and degrees of freedom 36 − 1 = 35.
- For the claim that waiting time is less than 50 minutes, the example uses a left-tail p-value from the t distribution and reports rejecting H₀ at α = 0.05.
- A physical-activity example asks whether students exceed a recommendation of 150 minutes per week, so the alternative is greater than 150 and the test is right-tailed.
- A grocery-spending example uses a benchmark mean of $250 per month, a sample of 40 students, and unknown population standard deviation.
- Because the grocery question asks whether mean spending differs from $250, it calls for a two-tailed t test; Adepoju directs students to practice all three tails using the Canvas worksheet and course resources.
Transcript excerpt
0:01 Okay. So, uh I want to welcome you to another week. Uh we are gradually moving to another exam. I've already posted um exam two. Um I think I'll pass exam two question the one for spring. Uh that will give you opportunity uh to see um the way exam 2 is um going to look like. And uh for those uh who miss time one and um and for those who want to have opportunity of improving on their score, I think I I gave an opportunity of a makeup. Uh initially the deadline was supposed to be
0:49 yesterday but I extended that to today. I think today is November 9. So if you have not um made use of opportunity of uh retaking or you um you you miss um the m you know it's an opportunity uh to do that and if you score less than what you have in the meter one of course I'm not going to use this I'm going to use the m one. So the idea behind this is just to provide our student opportunity to improve on their grade. So um there's not going to be another opportunity the moment this is due tonight and that's why I want you to make use of this opportunity.
1:37 Okay? There's not going to be any opportunity to maybe for makeup or to retake the exam. So if it is uh the moment it is due tonight. Okay. So um we started a poty testing the other time. Uh today um we're going into what I call a poty test for a population mean when the population standard deviation is unknown. I remember the one we did the other time. It was um you know when the population standard deviation is known but what of in a situation
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, Prof Kazeem Adepoju.