The table below contains the systolic blood pressure of 16 patients before and after one week treatment with captoril or placebo. The data are from a randomized controlled study of diabetics dependent on insulin with nephropathy (Hommel et al., BMJ, 1986).
Table: Systolic blood pressure of 16 patients before and after treatment with Captoril or Placebo.
|
Captoril
|
Placebo
|
|
Baseline
|
After 1 week
|
Baseline
|
After 1 week
|
|
147
|
137
|
133
|
139
|
|
129
|
120
|
129
|
134
|
|
158
|
141
|
152
|
136
|
|
164
|
137
|
161
|
151
|
|
134
|
140
|
154
|
147
|
|
155
|
144
|
141
|
137
|
|
151
|
134
|
156
|
149
|
|
141
|
123
|
|
|
|
153
|
142
|
|
|
The data are found in the file captopril.sav
The authors made a separate analysis for each treatment by means of a paired t-test. They found a significant decrease of the blood pressure in the Captoril group; however the analysis of the placebo group did not show any significant change. The authors concluded: "Captoril represents a valuable new drug for treating hypertension in diabetics dependent on insulin with nephropathy".
a) Check their findings, by repeating their tests. Show the output of the analysis and describe the two tests in terms of the hypothesis and give the p-values for the tests, the estimated effect sizes (i.e. the differences) and corresponding confidence intervals.
b) Although their conclusion might be correct their analysis has not really shown that Captoril is better than Placebo. Why?
c) Perform a proper analysis. Describe your null and alternative hypothesis, report the p-value. What is your conclusion?