HLTHAGE 2A03 Lecture Notes - Lecture 20: The Strongest, Quasi, Social Desirability Bias
Document Summary
Experiments have small sample sizes than surveys so when someone die it affects data and statistical calculations. Mortality is not just death - it is also moving or dropping out. Control group learns about the stimulus of the other group. This is another reason why the control group is given something else that mirrors the stimulus. Compensation - experiments makes up for conditions lacking in the control group sometimes by being extra caring. The solution is to do a double blind experiment - the experimenter still knows who is being given the control and who the stimulus but the assistant who administers it does not know. If you are addressing threats you will have high internal validity. Data can"t be transferred beyond the experiment if there is low external validity. Has high external validity but there it is not possible to randomly assign people to live in a community for instance so there is low validity.