Sampling
People who volunteer to be in a sample. This is biased.
Voluntary Response Sample
Individuals selected for a sample cannot be reached/refuse to participate.
Nonresponse
The group of people who are given a placebo or given the treatment of "no treatment"
Control group
All experimental units are assigned to all of the treatments at random.
Completely randomized
Choosing individuals who are easy to reach.
Convenience Sampling
A group of the population is left out of the process when choosing a sample.
Undercoverage
A dummy treatment used to see if something works. Looks the same as the real treatment.
Placebo
What is being measured and used for comparison.
Response variable
Divide population into homogeneous groups, perform a simple random sample in each group.
Stratified Random Sample
The respondent or interviewer cause the question answers to be dishonest.
Response bias
When neither the person in the experiment nor the person measuring the results is aware which treatment group they are in
Double Blind
Only two treatments done, each group gets both treatments
Matched pair design
Use the "hat method" or a RNG to randomly select individuals from the population. Each individual is equally likely to be selected.
Simple Random Sample
Asking a question in such a way that the preferred answer is obvious: a form of response bias
Leading question
Individuals are broken into blocks then randomly assigned to treatments separately.
Block Design
Applies a treatment to obtain a result. Observes the subject response when the treatment is administered.
Experiment
Randomly choose the first person, then choose more people from the population at a regular interval, ex: Every 6th person.
Systematic Sample
The type of unidentified variable that could be at play in observational studies, such that you cannot determine cause and effect.
Confounding Variable
Specific values of a treatment
Level
Observes people and measures the variables of interest without imposing treatments
Observational study