Unit 4 Flashcards
Population
UNIT 4: The ENTIRE group of people/units
Sample
UNIT 4: A small PART of population
“Representative”
UNIT 4: Has similar characteristics to the population
Selection/under coverage Bias
UNIT 4: Some of the population is excluded or underrepresented in selection
Nonresponse Bias
UNIT 4: No data is collected from selected individuals
Response Bias
UNIT 4: Process distorts responses
Judgement Sample
UNIT 4: :(
Individuals chosen based on your judgement
Voluntary Sample
UNIT 4: :(
Individuals volunteer to be a part of your sample
Convenience Sample
UNIT 4: :(
Choose your sample based on what is convenient
Simple Random Sampling (SRS)
UNIT 4: :)
-Pick names, put them back in hat
- MATH, PRB, randInt (min, max, #)
Systematic Sampling
UNIT 4: :)
- Select an SRS from a List
- Every kth individual is selected where k (the “hop”) is a predetermined interval
*Use hat for start (when given a list)
Stratified Random Sampling
UNIT 4: :)
- Divide the population into groups (strata)
- Strata are similar ins one way that’s important for response variable
- Choose an SRS from each strata and combine
*Each strata is a different hat
** Population is NOT homogeneous
Cluster Sampling
UNIT 4: :)
- Divide the population into groups (clusters)
- Randomly selected 2+ of the clusters. ALL individuals in selected clusters are sampled
Useful is clusters already exist
**Put clusters IN hats
**Effective when HOMOGENEOUS
Confounding Variables
UNIT 4: Undesired 3rd variable that influences your treatment and responses
How do you control confounding variables?
UNIT 4: 1. Keep consistent (ideal)
2. Randomize (realistic)
Common confounding variables
UNIT 4: - Placebo effect
- Experimenter bias
What is the Placebo effect?
UNIT 4: Many subjects respond to ANY treatment, even fake ones
How do you control for the Placebo effect?
UNIT 4: - Control group -> a group of patients who receive no treatment
- Blinding -> patients do not know which group they are in
What is experimenter bias?
UNIT 4: Many experimenters WANT their hypothesis to be confirmed
How do you control for experimenter bias?
UNIT 4: - Double blinding -> neither patients or experimenters know which group is getting which treatment
Why should you randomize?
UNIT 4: For confounding variables you do NOT know, reduce confounding effects
When do you use blocking?
UNIT 4: For confounding variables you DO know:
- You REASONABLY suspect a confounding variable
- You didn’t use stratified sampling
What is blocking?
UNIT 4: Everything has that characteristic equally, then randomize within your “block(s)’
What is matched-pair blocking?
UNIT 4: Most common form of blocking, each pair can be blocked off for same age/sex/income/etc.
Define TREATMENT
UNIT 4: Doing something to the experimental units in experiment
Define EXPERIMENTAL UNIT
UNIT 4: “victims” in experiment
Define SUBJECT
UNIT 4: Used when the experimental unit is a human
Define FACTOR
UNIT 4: Explanatory variable in experiment
Define LEVEL
UNIT 4: Treatment formed by combining a specific value of each factors, sub-factors
Factors * Treatments
