Unit 1: Designing an Experiment or Study Flashcards
Why do we randomize in a study/experiment?
To reduce bias
Cluster Sample
Sample that is taken by doing an SRS on clusters and selecting a few entire groups.
Bias
When a survey over- or under-estimates some characteristic of the population.
Steps of a simulation
- Assign digits to outcomes
- Describe random selection/method
- Describe trials and response variable
- State conclusion
Population
The entire group of individuals about whom we hope to learn.
Sample
A (representative) subset of a population.
Sample survey
A study that asks questions of a sample drawn from some population.
Randomization
The best defense against bias.
Sample size
The number of individuals in a sample.
Census
A sample that consists of the entire population.
Best used on a small population.
Population parameter
A numerically valued attribute of a model for a population.
Statistic, sample statistic
Values calculated for sampled data.
Simple Random Sample (SRS)
A simple random sample of a sample size, n, which has an equal chance of selection.
Stratified random sample
A sampling design in which the population is divided into strata, random samples are then drawn from each stratum.
Cluster sample
A sampling design in which entire groups are chosen at random.
Systematic sample
A sample drawn by selecting individuals systematically from a sampling frame.
Voluntary response bias
Bias introduced to a sample when individuals can choose on their own whether to participate in the sample.
Convenience sample
A sample that consists of the individuals who are conveniently available
Undercoverage
A sampling scheme that biases the sample in a way that gives a part of the population less representation.
Nonresponse bias
Bias introduced when a large fraction of those sampled fails to respond.
Response bias
Anything in a survey design that influences responses. Normally found in the wording of the survey questions.
Random assignment
To be valid, an experiment must assign experimental units to treatment groups at random. MOST IMPORTANT!!
Principles of Experimental Design
Control, Randomize, Replicate, Block
Placebo
A treatment known to have no effect
Completely Randomized
No blocks or pairs.
Blocked
Has blocking/ Strata. Reduces variability.
Matched
Before/ After. Comparing results for some group of participants.
Blind
Participants do not know the treatment . Used to reduce bias.
Double-blind
participants and administrators do not know treatments.
Retrospective
Any study done by gathering data that already exists ( In the past).
Prospective
Any study done by gathering data in the present ( no treatments).
Treatments
Total number of groupings in an experiment