Paired t test worksheet with answer key

8 problems on paired t test, ready to print for class or homework. The answer key prints on its own page, so hand out everything before it.

Problem 1

For each design, say whether a paired tt procedure or a two-sample tt procedure is appropriate, and give the reason.

(a) Twenty runners each run a timed mile in two shoe models, with the order of the models randomized for each runner. (b) Eighteen tomato plants are randomly assigned, nine to fertilizer A and nine to fertilizer B, and yield is recorded. (c) Twelve pairs of identical twins are recruited; within each pair, one twin is randomly assigned to a tutoring program and the other is not. (d) A random sample of 40 seniors and an independent random sample of 35 juniors report weekly study hours.

Problem 2

Six office workers are randomly selected for a keyboard-shortcut workshop. Their data-entry speed, in records per hour, is measured before and after the workshop.

WorkerBeforeAfter
14247
25558
33846
44749
55157
64545

At α=0.05\alpha = 0.05, is there convincing evidence that the workshop increases mean data-entry speed?

Problem 3

A shoe company tests a new insole. Ten randomly selected runners each run on a force plate twice, once in their usual insole and once in the new one, with the order randomized. For the differences (usual minus new) in peak impact force, in newtons, the company reports dˉ=18\bar{d} = 18 N and sd=21s_d = 21 N. Construct and interpret a 95% confidence interval for the true mean difference, and say what it implies about the new insole.

Problem 4

Eight athletes have their vertical jump measured, in centimeters, before and after a six-week plyometric program.

AthleteBeforeAfter
15255
25861
34651
46163
55558
64953
75758
85055

(a) Run the correct test at α=0.05\alpha = 0.05 for whether the program increases mean vertical jump. (b) A classmate instead treats the 8 before values and the 8 after values as two independent samples and runs a two-sample tt test with the conservative degrees of freedom. Carry out that analysis too. (c) Explain why the two answers differ so sharply and which one is right.

Problem 5

Twelve randomly selected patients have their total cholesterol measured, in mg/dL, before and after eight weeks on a diet. For the differences d=before−afterd = \text{before} - \text{after}, the study reports dˉ=14.5\bar{d} = 14.5 and sd=22.0s_d = 22.0. (a) At α=0.05\alpha = 0.05, test whether the diet lowers mean cholesterol. (b) A second analyst subtracts in the other order, d=after−befored = \text{after} - \text{before}. State exactly what changes and what does not in the hypotheses, the statistic, the p-value, and the 95% confidence interval.

Problem 6

A driving school matches 15 pairs of students on their initial road-test score. Within each pair, one student is randomly assigned to simulator-based lessons and the other to road-based lessons. For the differences (simulator minus road) in final score, the school reports dˉ=2.4\bar{d} = 2.4 points and sd=5.1s_d = 5.1 points. (a) At α=0.05\alpha = 0.05, test whether the two lesson types differ in mean final score. (b) State what the design does and does not allow you to conclude.

Problem 7

A researcher runs a paired tt test on 20 randomly selected participants who complete a mindfulness course, using d=post score−pre scored = \text{post score} - \text{pre score}. Software reports:

QuantityValue
nn20
Mean of dd3.6
StDev of dd8.4
SE Mean1.878
tt1.917
P (one-sided)0.035

(a) State the hypotheses this output tests. (b) Interpret the p-value in context. (c) A student writes: 'There is a 3.5% chance that the course has no effect.' Explain what is wrong with that. (d) Build the 95% confidence interval for μd\mu_d and reconcile it with the reported p-value.

Problem 8

A city measures rush-hour travel time, in minutes, on the same 9 routes before and after installing a new traffic-signal system. The differences (before minus after) are:

2.1, 1.4, 3.0, 0.6, 2.5, 1.9, 8.7, 1.2, 2.02.1,\ 1.4,\ 3.0,\ 0.6,\ 2.5,\ 1.9,\ 8.7,\ 1.2,\ 2.0

(a) Check the conditions for a paired tt procedure and identify the problem. (b) Build the 95% confidence interval from all 9 routes anyway. (c) Rebuild it from the other 8 routes and compare. (d) State what the city can and cannot report.

Paired t test: answer key

  1. (a) Paired: each runner gives both times. (b) Two-sample: different plants in each group, so equal sizes do not create pairing. (c) Paired (matched pairs): 12 twin pairs, df=11df = 11. (d) Two-sample: different individuals and unequal sizes 40 and 35.

  2. Differences (after minus before): 5,3,8,2,6,05, 3, 8, 2, 6, 0. dˉ=4\bar{d} = 4, sd=2.8983s_d = 2.8983, SE=1.1832SE = 1.1832, t=3.38t = 3.38, df=5df = 5, one-sided p-value ≈0.0098\approx 0.0098 (table bracket 0.0050.005 to 0.010.01). Reject H0H_0; convincing evidence of a higher mean speed, though the design cannot rule out practice effects.

  3. SE=6.6408SE = 6.6408 N, df=9df = 9, t∗=2.262t^* = 2.262, ME=15.02ME = 15.02 N. The 95% interval for μd\mu_d is (2.98, 33.02)(2.98,\ 33.02) N. It excludes 0, so the new insole reduces mean peak impact force, but the interval is too wide to say by how much with any precision.

  4. (a) Differences 3,3,5,2,3,4,1,53, 3, 5, 2, 3, 4, 1, 5; dˉ=3.25\bar{d} = 3.25, sd=1.3887s_d = 1.3887, SE=0.4910SE = 0.4910, t=6.62t = 6.62, df=7df = 7, one-sided p-value ≈0.00015\approx 0.00015. Reject H0H_0. (b) Two-sample: SE=2.2971SE = 2.2971, t=1.41t = 1.41, conservative df=7df = 7, one-sided p-value ≈0.100\approx 0.100, fail to reject. (c) The paired procedure is correct; ignoring the pairing pushes between-athlete variation into the standard error, inflating it about 4.7 times and turning an overwhelming result into a null one.

  5. (a) SE=6.3509SE = 6.3509, t=2.28t = 2.28, df=11df = 11, one-sided p-value ≈0.0216\approx 0.0216 (table bracket 0.010.01 to 0.0250.025). Reject H0H_0; convincing evidence of a drop in mean cholesterol over the eight weeks, though with no control group the drop cannot be credited to the diet. (b) dˉ\bar{d} and tt flip sign and HaH_a flips to μd<0\mu_d < 0; sds_d, SESE, dfdf, ∣t∣|t|, the p-value, and the conclusion are unchanged, and the interval (0.52, 28.48)(0.52,\ 28.48) becomes (−28.48, −0.52)(-28.48,\ -0.52).

  6. (a) SE=1.3168SE = 1.3168, t=1.82t = 1.82, df=14df = 14, two-sided p-value ≈0.0898\approx 0.0898 (table bracket 0.050.05 to 0.100.10). Fail to reject H0H_0; not convincing evidence of a difference in true mean final score. (b) Random assignment within pairs would support a cause-and-effect claim, but with no random selection the result does not generalize beyond these students.

  7. (a) H0:μd=0H_0: \mu_d = 0 versus Ha:μd>0H_a: \mu_d > 0 with df=19df = 19. (b) If the course had no effect, about 3.5% of samples of 20 would show a mean gain of 3.6 points or more. (c) A p-value is the probability of data this extreme given H0H_0, not the probability that H0H_0 is true. (d) t∗=2.093t^* = 2.093, ME=3.931ME = 3.931, interval (−0.331, 7.531)(-0.331,\ 7.531); it contains 0 because it matches the two-sided p-value of 0.0700.070, not the one-sided 0.0350.035.

  8. (a) The random condition is not met (these are not a random sample of routes), and the differences contain an outlier at 8.78.7, above the upper fence of 4.9254.925, so the shape condition fails. (b) dˉ=2.6\bar{d} = 2.6, sd=2.3948s_d = 2.3948, SE=0.7983SE = 0.7983, df=8df = 8, t∗=2.306t^* = 2.306, interval (0.759, 4.441)(0.759,\ 4.441) minutes. (c) Without that route, dˉ=1.8375\bar{d} = 1.8375, sd=0.7577s_d = 0.7577, df=7df = 7, t∗=2.365t^* = 2.365, interval (1.204, 2.471)(1.204,\ 2.471) minutes. (d) Both exclude 0, so the direction is stable, but the size is not; report both, flag the outlier, and claim no causal effect.

The key gives final answers only. The full worked solution for each problem is on the matching practice set, and every other sheet is on the worksheets page.

More worksheets from this unit