StatisticsInferential Statistics

Matched Pairs Design: Test Cholesterol Before and After a Diet

Keep all nine before-and-after subject pairs together and perform a one-sided paired t-test at the 5% level.

Question

A study was conducted to investigate how effective a new diet was in lowering cholesterol. Results for the randomly selected subjects are shown in the table. The differences have a normal distribution. Are the subjects’ cholesterol levels lower on average after the diet? Test at the 5% level.

Table 10.13: Before-and-after cholesterol by subject

SubjectABCDEFGHI
Before209210205198216217238240222
After199207189209217202211223201

1. Preserve the matched pairs

Each column contains the same subject measured before and after the diet. Define the difference consistently as d=after−befored=\text{after}-\text{before}, so a negative difference indicates a decrease. There are 9 pairs, not 18 independent observations.

SubjectBeforeAfterDifference d
A209199−10
B210207−3
C205189−16
D19820911
E2162171
F217202−15
G238211−27
H240223−17
I222201−21

2. State the hypotheses and model

Let μd\mu_d be the population mean of after-minus-before differences. Test H0:μd=0H_0:\mu_d=0 against Ha:μd<0H_a:\mu_d<0 at α=0.05\alpha=0.05. The broader null is μd≥0\mu_d\ge0; its boundary zero gives the test statistic. This is a left-tailed paired t-test. The problem supplies randomly selected subjects and normally distributed differences; the model also treats different subjects' differences as independent.

3. Calculate from the unrounded differences

∑di=−97,dˉ=−979=−10.7778.\sum d_i=-97,\qquad \bar d=-\frac{97}{9}=-10.7778. sd=∑(di−dˉ)29−1=11.8615,SE=sd9=3.9538.s_d=\sqrt{\frac{\sum(d_i-\bar d)^2}{9-1}}=11.8615,\qquad SE=\frac{s_d}{\sqrt9}=3.9538. t=dˉ−0sd/9=−2.7259,df=9−1=8.t=\frac{\bar d-0}{s_d/\sqrt9}=-2.7259,\qquad df=9-1=8.

4. Decide and conclude

The one-sided probability is P(T8≤−2.7259)≈0.0130P(T_8\le-2.7259)\approx0.0130. Because 0.0130<0.050.0130<0.05, reject the null hypothesis. Under the stated model, there is sufficient evidence that the population mean cholesterol level is lower after the diet. The sample's mean decrease is about 10.78 in the recorded measurement scale. The source does not specify units, so none are added.

This conclusion concerns the average, not every subject: D and E increased. Random selection does not by itself establish a randomized controlled treatment experiment. Without a control group, this before-and-after comparison does not isolate the diet from other changes over time.

Original exercise attribution

Adapted in layout from Barbara Illowsky and Susan Dean, Introductory Statistics (OpenStax College, 2013), CC BY 4.0. Download for free at http://cnx.org/content/col11562/latest/. The exercise data and all subquestions are retained; the worked explanation and plotted dots are provided here.

Evidence boundary

This answer uses Try It 10.11, Table 10.13 in the 2013 OpenStax PDF, preserving all nine subject pairs and the stated normality assumption. Different subjects’ differences are treated as independent. Units are not supplied. Calculations are independently recomputed; the before-and-after result is not proof that the diet alone caused the change and is not medical advice.

Sources

These references support the concepts and methods used in the explanation above.