random assignment & designs

Two group randomized, pre-test/post-test

can conclude difference between groups over time.

one group, pre-test/post-test

can't infer that it was the intervention that caused the change in score, could be just an effect of time

one group, post-test only

can't make inferences about whether intervention makes any difference

Two group, post-test only

can't make inferences about which treatment is better because we don't know where the groups started

random assignment

method for assigning subjects to groups in an experiment so that subject as an equal probability of being assigned to any group increases internal validity bc groups are more equal at baseline, its more reasonable to assume that changes between groups are a result of the intervention/treatment

random sampling

method of choosing a sample from a population so that each member of the population has an equal probability of being chosen increases external validity bc if the sample is a good representation of the population being studied, then results can be better generalized to the population

stratification or block, then randomization

pre-specify sources of secondary variance & try to control for it (equal amounts of a characteristic in all group) randomization scheme for those that possess the characteristic, and different randomization scheme for those that do not. example: equal amount of males and females in each group, as gender could be a confounding factor and lead to secondary variance

adaptive randomization

take subjects baseline info, computer decides where to put the subject in order to balance the groups allows control of multiple characteristics (gender, months since surgery etc) ensures diverse characteristics in each group