A study is designed to investigate potential adverse effects of drug X. A table
ID: 3301185 • Letter: A
Question
A study is designed to investigate potential adverse effects of drug X. A table given below summarizes the first look of the outcomes. The re- searcher randomly assigns 204 from a group of 409 participants to take drug X. The remaining 205 subjects received placebo. Row 1 of the data table gives the numbers of participants who report experiencing adverse effects.
(a) Consider two binary variables: Y : having adverse reaction or not; and X : being assigned to the drug X group or not. Which variable would you consider to be the response, and the predictor? Please briefly explain why.
(b) Now focus on the placebo group, please provide a model that you believe to be suitable to study this subset of data – make sure you describe the parameter(s) present in the model.
(c) In this type of studies, you can consider the numbers of partici- pants vs the proportion of participants that suffer from the side effects. Both of them are statistics obtained from the data. One argues that the proportion is a better statistic to focus on. Do you agree with this assessment, why or why not?
Drug X Placebo
Adverse 30 21
No Effect 174 184
Total 204 205
Explanation / Answer
a) "Y : having adverse reaction or not" will be considered the response variable whereas X (drug or not) will be considered the experiment or independent variable or control variable.
b) this placebo can be seen as proportion that incorporates both adverse and no effect proportion i.e. 21/205 and 184/205.. The model parameter is true propotion.
c) Yes proportion is better to consider for the study since it considers the fact that how many samples are present in actual experiment and reflects the weights as per samples. Example
CONSIDER Drug X case:
it says that adverse case for drug x is 30 in number as opposed to 174 in case of no-effect. That means drug X bears very high number of 'no-effect' case compared to adverse case. But in proportion drug x bears 30/51= 59% population of adverse case and 174/358=48% of no-effect case which is less than the case of adverse.
Related Questions
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.