1/42
Vocabulary flashcards covering Sections 1F and 1G on data collection, experiments, observational studies, investigative questions, sampling techniques, and forms of bias.
Inference
Using a sample to answer questions about a whole population
Population
The whole group you want to know about
Sample
The part of the population you actually collect data from
Parameter
A number that describes a population
Statistic
A number that describes a sample
Census
Collecting data from everyone in the population
Survey
Asking questions of a sample
Sample variability
Different random samples give different results
Experiment
Researchers assign treatments (hands on)
Observational study
Researchers only observe, nothing assigned (hands off)
Retrospective
Observational study using data from the past
Prospective
Observational study collecting data going forward in time
Response variable
The outcome being measured
Explanatory variable
The variable that may explain or predict the response
Factor
A variable the researchers manipulate
Level
A value or setting of a factor
Treatment
One combination of levels applied to a unit
Experimental unit
What the treatment is applied to (humans = subjects)
Extraneous variable
Another variable (not the explanatory one) that may affect the response
Confounding variable
Linked to both the explanatory and response variables, so you can't tell which caused the change
Random assignment
Using chance to decide who gets which treatment; allows cause-and-effect conclusions
Random sampling
Using chance to pick who is in the sample; allows generalizing to the population
Number of treatments
Multiply the number of levels of each factor
3 components
1) Variable(s) 2) Inference and parameter 3) Conclusion (population generalized to, cause-and-effect or not)
Estimation question
What are the plausible values for the [mean/proportion] of [variable] for [population]?
Association question
Is [explanatory] associated with [direction] in [response] compared to [group]?
Causation question
Does [explanatory] cause [direction] in [response] compared to [group]?
Sampling frame
The list of everyone you could choose from
With replacement
Repeats are allowed
Without replacement
No repeats (SRS)
SRS
Every possible group of that size has an equal chance of being chosen
SRS steps
Label, Randomize, Select (ignore repeats)
Stratified sample
Split into groups (strata) that are alike, then take an SRS from every group
Cluster sample
Split into nearby groups, SRS of groups, then survey everyone in the chosen groups
Systematic sample
Random start, then every kth person on the list
Voluntary response
People choose to respond themselves
Convenience sample
Whoever is easiest to reach
Memory trick
Stratified = Some from All groups. Cluster = All from Some groups.
Bias
A method that consistently over- or underestimates the true value
Undercoverage
Some people have no chance (or less chance) of being selected
Nonresponse bias
Chosen people don't respond, and they differ from those who do
Response bias
Inaccurate answers due to wording, who's asking, or a sensitive topic
Writing about bias
Name population and sample, who's missing and how they differ, then over- or underestimate and why