In: Statistics and Probability
Does aspirin cure headaches? A study enrolled 100 students who had frequent headaches to particpate in a study to examine the effets of aspirin on their headaches. The students were instructed to take aspirin when they had a headache and to report whether there was a substantial relief from the headache pain within an hour.
a) explain why this study is biased
b) explain how would you change the study to remove the bias
Randomization- : Diagram the aspirin experiment. - draw a diagram that describes the experiement you suggested in part b above. (Intro to Stats- chapter 3, section 3.2 Design of Experiments, 3.19 pg175)
To your question: You have to answer part b as indicated above "b)explain how you would change the study to remove the bias" then draw a diagram that describes the experiment you suggested.
a) ans. The above study is biased because we take decision about the effect of aspirin based on their opinion not any tested opinion by doctors or clinic. So the study is biased. There exist something which is beyond it and that estimates would be unbiased.
b) ans. To remove this biased we should correct our sample from sampling biasedness. We should check the condition of patient after being treated by aspirin. If you have numerical data on headache of some patients , then you can do paired sample t-test of same patients by taking before & after treated by aspirin sample.
Also you do the headache level of two samples one is treated by aspirin and another is not treated by aspirin and then you can do 2-sample t-test.
Now if you have 2 categorical variable. For say type of patients with headache or not headache , that is two type. Another variable suppose say treatment level of Aspirin that also has three category treated by high , medium or low level of Aspirin. Then we can also do chi-square test two see the association of headache & treatment of aspirin. The data of chi-square test might be looking like below :-