1/52
Looks like no tags are added yet.
What is a population?
The entire group we want information about.
What is a sample?
A subset of the population that is actually studied.
What is an SRS?
A sample where every group of the same size has an equal chance of selection.
What are the steps for an SRS?
Label individuals, randomize with an RNG, then select.
What should you do with repeated RNG numbers?
Ignore repeats and continue until the sample is complete.
What is a voluntary response sample?
People choose themselves to participate in the sample.
Why can voluntary response samples be biased?
People with strong opinions are more likely to respond.
What is a convenience sample?
A sample made from individuals who are easiest to reach.
Why can convenience samples be biased?
The easiest people to reach may not represent the population.
What is a stratified random sample?
Split the population into homogeneous strata, then take an SRS from each.
“Sample some from ALL groups”
What does "strata" mean?
Groups within a population that share a characteristic.
What is the key idea for stratified sampling?
Sample some from all groups.
Why use stratified sampling?
It can reduce variability and produce more precise estimates.
What does homogeneous mean?
Individuals within a group are similar in a relevant characteristic.
What is a cluster random sample?
Split into heterogeneous groups, randomly choose groups, and sample everyone in them.
“All from SOME groups”
What is the key idea for cluster sampling?
Sample all from some groups.
What does heterogeneous mean?
Each group contains a mix of different types of individuals.
What is systematic random sampling?
Choose a random starting point, then select using equal intervals.
How does systematic sampling begin?
Choose a random starting point.
What happens after the starting point in systematic sampling?
Select individuals at equal intervals.
Stratified vs. cluster: how are groups different?
Strata are homogeneous; clusters are heterogeneous.
Stratified vs. cluster: how are samples different?
Stratified: some from all groups. Cluster: all from some groups.
What is μ (mu)?
The population mean.
What is x̄ (x-bar)?
The sample mean.
What makes a sampling method produce a good estimate?
Low bias and low variability.
What does low bias mean?
The method's estimates are centered close to the true population value.
What does low variability mean?
Repeated samples give estimates that are close together.
In the fan example, what method was best for average enjoyment?
Stratifying by row, because estimates had low variability.
Why might stratifying by row improve the fan estimate?
Each row is represented, and estimates are less variable.
Why might choosing only one row be poor?
It may not represent the entire population if rows differ.
Why can choosing from every row improve representation?
It includes individuals from different rows.
What is sampling without replacement?
Once selected, an individual cannot be selected again.
If selecting 100 unique students, replacement or no replacement?
Without replacement, because each student should be selected only once.
Advantage of stratified sampling?
Guarantees representation from every stratum and can reduce variability.
Advantage of cluster sampling?
It can be easier and faster to collect data.
Advantage of systematic sampling?
It spreads the sample across the population using equal intervals.
What is undercoverage?
Some groups in the population are inadequately represented or left out.
Example of undercoverage?
Calling only landlines misses people who don't use landlines.
What is nonresponse?
Selected individuals cannot be reached or refuse to respond.
Example of nonresponse?
A selected person does not answer the phone or hangs up.
What is response bias?
Responses are systematically affected by the data-collection process.
What can cause response bias?
Lying, sensitive questions, interviewer effects, or biased wording.
Example of response bias from wording?
A leading question may influence people toward a particular answer.
Observational study
The participants receive no treatment
Experiment
Treatment imposed - allows us to show causation
Experimental units
Who/what the treatment gets imposed on
Treatment
What is done (or not done) to experimental units
Well designed experiment
Comparison - 2 or more treatments
Random assignment
Replicate-more than 1 in each treatment group
Control -keep other variables constant
Random assignment
Label
Randomized
Assign
Shows causation.
Placebo effect
When a false treatment (placebo) works
Statistically Significant
When a result of an experiment is unlikely (less than 5%) to happen purely by chance.
If it’s statistically significant, we have convincing evidence that treatment caused the difference.
A random sample allows us to..
Generalize our conclusions to the population.
Random Assignment allows us to…
Conclude a treatment causes change in the response variable.