Experiment terminology Flashcards
experimental unit
person or object that an experiment is performed on
explanatory variable /// factor
the variable being studied to see if it causes an effect or response.
Sometimes jont effects of more than one factors are studied.
levels
a specific value of a factor
For example, if the explanatory variable is dosage of meds, the levels might be low, medium, high, none.
response variable
the variable being measured to determine whether it was affected by the explanatory variable.
For example when factor is drug dose, response is degree of relief.
Treatments
Conditions specifically imposed on a group (related to levels of explanatory variable, or combinations of levels of more than one explanatory variable)
Lurking variables
Variables not considered in the study which may have an influence on the response variable. If the subjects in one group have a certain characteristic that the other group does not, then the characteristic (variable) can be confounded with the explanatory variBle. Therefore, randomization is used in order to distribute the effects of these evenly into the control and treatment groups.
Basic principles of experimental design
RANDOMIZATION
to equalize groups so that they are as similar as possible before treatment. (try to distribute possible lurking variables)
Basic principles of experimental design
CONTROL
control aspects of the experiment that we know may have an effect on the reponse, but that are not factors being studied
Basic principles of experimental design
REPLICATION
repeat enough times so that chance variation can be ruled out. This is to obtainstatistical significance.
Basic principles of experimental design
BLINDNESS
the subjects do not know which treatment they are recieving
double-blindness: the people evaluating the response do not know which treatment the subject has been givem
triple-blindness: we, the statisticians, have no idea of what we’re doing… (not really)
Basic principles of experimental design
A CONTROL GROUP
control group does not recieve treatment so can be used for comparison to eliminate attributing to the treatment changes that are really fro other influences such as passage of time, or psychological benefits of being treated. Therefore we use the control group as a means of comparison to rhe experimental treatments.
Basic principles of experimental design
PLACEBO EFFECT
Tendency of many human subjects to show a response physically just because they believe they are taking a medication.
PLACEBO
A null treatment administered for comparison to help determine whether the observed effect is due to experimental treatment or just to placebo effect
Completely Randomized experiment
Randomly allocate subjects into groups
Assign treatment
Evaluate response variable, compare results of the treatment groups
Randomized block experiment
If you determine that there is some important characteristic pf the experimemtal units that could obscure results, first break into subgroups or blocks of homogeneous units
Within each block, randomize and assign treatment
Compare response to other treatment groups within the same block
