Experimental Design Flashcards
observational study
simply observe the individuals and measure the results cause and effect cannot be demonstrated.
experimental units
the smallest “unit” on which treatment is imposed
response variable
what is measured after treatment
explanatory variable
the factor that is manipulated
control
things that we know affect the outcome but are not being studied must be taken into account in the design of the experiment
randomization
treatments must be randomly assigned in order for an experiment to be valid. The purpose of this is to even out the impact of factors we cannot control.
replication
many subjects should be included in the study often, experiments are repeated in order to get representative subjects.
blocking ( sometimes )
grouping similar subjects together to reduce the effects of things that cannot be controlled.
single blind
when either the subjects/treatment administrators/technicians or the evaluators if the data do not know what treatments are applied to what subjects.
double blind
when neither the subjects/treatment administrators/nor the evaluators of the data do not know what treatments are applied to what subjects.
Placebo effect
“faux” testament: humans tend to show a response ( typically around 20% of the time ) even when a placebo is administrated
statistically significant
the difference in the response for the control group and the response group is too large to occur by chance.
confounding variables
factors that are linked in the response variable and cannot be separated from them. Factors not in the experiment that might effect the outcome. try to control for
experiment
deliberately impose an experiment
retrospective
look over past records and events
