Statistical Inference and Study Design flashcards

0/60

60 cards remaining

Take the chapter practice test

All 60 cards in this deckShow

Sample size and biasPopulations, Samples, and Representative Sampling: More of the wrong sampling method is still the wrong sampling method.
When should I use population and sample?Populations, Samples, and Representative Sampling: Distinguish the target group from observed participants.
What method supports sample size and bias?Populations, Samples, and Representative Sampling: size reduces random variation, not systematic selection bias
Representative selectionPopulations, Samples, and Representative Sampling: Random selection from the right list is more useful than an easy-to-reach crowd.
Population and samplePopulations, Samples, and Representative Sampling: Population is everyone you want to know about; sample is whom you observed.
What must I check in populations, samples, and representative sampling?Populations, Samples, and Representative Sampling: A large sample does not repair a systematically biased selection method
What method supports representative selection?Populations, Samples, and Representative Sampling: random sampling supports population generalization
What method supports population and sample?Populations, Samples, and Representative Sampling: sample ⊆ target population
When should I use sample size and bias?Populations, Samples, and Representative Sampling: Explain why selection method matters even for large samples.
When should I use representative selection?Populations, Samples, and Representative Sampling: Choose a method that gives the target group a fair opportunity for inclusion.
Rescale an estimated proportionEstimating Population Counts, Means, and Proportions: Scale only when the new group matches the population you sampled.
When should I use estimate a population count?Estimating Population Counts, Means, and Proportions: Scale a sample proportion to a known population size.
What method supports rescale an estimated proportion?Estimating Population Counts, Means, and Proportions: estimated affected count=proportion×group size
Estimate a population totalEstimating Population Counts, Means, and Proportions: Estimated average times population count gives an estimated total.
Estimate a population countEstimating Population Counts, Means, and Proportions: Apply the sample share to the full population.
What must I check in estimating population counts, means, and proportions?Estimating Population Counts, Means, and Proportions: Population estimates inherit sample uncertainty; use an appropriate representative sample from the target population
What method supports estimate a population total?Estimating Population Counts, Means, and Proportions: estimated total=N·sample mean
What method supports estimate a population count?Estimating Population Counts, Means, and Proportions: estimated count=N·x/n
When should I use rescale an estimated proportion?Estimating Population Counts, Means, and Proportions: Translate a rate estimate into a comparable group count.
When should I use estimate a population total?Estimating Population Counts, Means, and Proportions: Use a sample mean to estimate a total quantity.
Compare estimates with uncertaintyMargin of Error and Sample Size: Do not treat two estimated percentages as exact facts.
When should I use margin-of-error interval?Margin of Error and Sample Size: Form an interval around a stated estimate.
What method supports compare estimates with uncertainty?Margin of Error and Sample Size: compatible interval endpoint = estimate ± reported margin
Sample size and precisionMargin of Error and Sample Size: More random observations usually narrow sampling uncertainty.
Margin-of-error intervalMargin of Error and Sample Size: Subtract the margin for the low end and add it for the high end.
What must I check in margin of error and sample size?Margin of Error and Sample Size: A reported margin of error describes sampling uncertainty under stated conditions, not every source of error or an individual-outcome range
What method supports sample size and precision?Margin of Error and Sample Size: rough scaling: margin ∝ 1/√n under comparable conditions
What method supports margin-of-error interval?Margin of Error and Sample Size: interval=[estimate−margin, estimate+margin]
When should I use compare estimates with uncertainty?Margin of Error and Sample Size: Use reported intervals to judge whether a claimed difference is supported by the supplied information.
When should I use sample size and precision?Margin of Error and Sample Size: Identify the effect of a larger random sample under similar conditions.
Control and comparisonObservational Studies versus Experiments: A comparison group helps distinguish the treatment from what would happen anyway.
When should I use observational studies?Observational Studies versus Experiments: Recognize when researchers measure existing conditions.
What method supports control and comparison?Observational Studies versus Experiments: treatment effect is assessed by comparing group outcomes
ExperimentsObservational Studies versus Experiments: Researchers choose which condition each participant receives.
Observational studiesObservational Studies versus Experiments: Researchers watch what already happens.
What must I check in observational studies versus experiments?Observational Studies versus Experiments: Observing existing behavior differs from assigning treatments; causal conclusions depend on study design
What method supports experiments?Observational Studies versus Experiments: assigned treatment + randomized groups supports causal comparison
What method supports observational studies?Observational Studies versus Experiments: no assigned treatment ⇒ observational design
When should I use control and comparison?Observational Studies versus Experiments: Identify a comparison group’s role in an experiment.
When should I use experiments?Observational Studies versus Experiments: Recognize assigned explanatory conditions.
Both design featuresRandom Sampling versus Random Assignment: Selection and treatment randomization answer different questions together.
When should I use sampling purpose?Random Sampling versus Random Assignment: Connect random selection to population inference.
What method supports both design features?Random Sampling versus Random Assignment: random sampling + random assignment → potential generalization and causation
Assignment purposeRandom Sampling versus Random Assignment: Assignment chooses what happens to each participant.
Sampling purposeRandom Sampling versus Random Assignment: Sampling chooses who joins the study.
What must I check in random sampling versus random assignment?Random Sampling versus Random Assignment: Random sampling supports generalization; random assignment supports causal treatment comparison; neither substitutes for the other
What method supports assignment purpose?Random Sampling versus Random Assignment: random assignment → causal treatment comparison
What method supports sampling purpose?Random Sampling versus Random Assignment: random sample → population generalization
When should I use both design features?Random Sampling versus Random Assignment: Evaluate a study that uses random sampling and random assignment.
When should I use assignment purpose?Random Sampling versus Random Assignment: Connect randomized treatment allocation to causal evidence.
Bounded conclusionsBias, Generalization, Association, and Causation: Match both the claim type and its population to the evidence.
When should I use selection bias?Bias, Generalization, Association, and Causation: Identify a mismatch between a sample source and target population.
What method supports bounded conclusions?Bias, Generalization, Association, and Causation: claim strength ≤ design support
Confounding and causeBias, Generalization, Association, and Causation: A third factor can move both variables together.
Selection biasBias, Generalization, Association, and Causation: Check who had a real chance to be included.
What must I check in bias, generalization, association, and causation?Bias, Generalization, Association, and Causation: Limit conclusions to the target population and design; a strong association alone does not prove cause
What method supports confounding and cause?Bias, Generalization, Association, and Causation: an alternative shared cause can explain association
What method supports selection bias?Bias, Generalization, Association, and Causation: representative inference requires an appropriate sampling frame
When should I use bounded conclusions?Bias, Generalization, Association, and Causation: Choose a conclusion consistent with sampling and treatment design.
When should I use confounding and cause?Bias, Generalization, Association, and Causation: Explain how a third factor can produce an observed association.