DATA1001 Lecture 2: Lecture 2- Controlling Data
Document Summary
Important (not to be ignored) but less reliable: personal testimony, media articles. Design of the study: where did you get your data from, eg. Testing roaccutane on mice: what is roaccutane prescribed for and how does it chemically work, what are known side affects, what other drugs are prescribed for acne, results: mice were less mobile. They interpreted this as "they were depressed: questions: Does mice = people: comparison: always must have a control group and a treatment group within the, control group is given a placebo (a pretend treatment) Randomised controlled trial experimental subjects: potential problems, confounding. When the differences between the control and treatment group affects/complicates the results. Two types of control: historical and contemporeous - historical was before the time of the test group and contemporerous is at the same time. Contemporaneous is preferrable as there are less confounding factors.