Inference Reference

Sampling, Surveys, and Experiments

Reliable statistical conclusions start with a representative sample and a design that matches the claim being made.

Solved Examples

Jump directly to a worked example by study-design decision.

Fact Table

IdeaMeaningWhat It Supports
PopulationEntire group of interest.The group a conclusion aims to describe.
Random sampleSelection process gives population members a fair chance to be chosen.Generalizing from sample to population.
Observational studyResearchers measure without assigning treatments.Association, not causation.
Randomized experimentResearchers assign treatments randomly.Evidence for a causal effect within the study scope.

Content Formulas

Sample Proportion
$$\hat p=\frac{\text{successes}}{\text{sample size}}$$
Margin of Error Idea
$$\text{estimate}\pm\text{margin of error}$$
Design Rule
$$\text{random sample}\Rightarrow\text{supports population inference}$$
Random assignment and random sampling answer different questions. Assignment supports causation; sampling supports generalization.

Classic Examples

Identify Population and Sample

A school randomly selects $50$ of its $800$ students to ask about lunch. Identify the population and sample.

Solution Steps
  1. Identify everyone the study wants to describe.
  2. Identify the students actually surveyed.
  3. State both groups.
$$\text{target group: all }800\text{ students}$$$$\text{surveyed group: }50\text{ selected students}$$$$\boxed{\text{population }=800,\quad\text{sample }=50}$$

Find Sampling Bias

A principal surveys only students in the chess club about school lunch. Explain the bias.

Solution Steps
  1. Identify who is excluded from the selection process.
  2. Explain why the group may differ from the full school.
  3. State the sampling problem.
$$\text{only chess-club students are sampled}$$$$\text{their lunch opinions may not represent all students}$$$$\boxed{\text{The sample is biased and not representative.}}$$

Classify a Study

Researchers randomly assign patients to receive a new medicine or a placebo. What type of study is this?

Solution Steps
  1. Identify that researchers assign treatments.
  2. Notice that the assignment is random.
  3. Classify the study and its causal strength.
$$\text{medicine or placebo is assigned}$$$$\text{assignment is random}$$$$\boxed{\text{randomized experiment; it can support causal inference}}$$

Evaluate a Claim

A poll of $200$ randomly selected voters finds $120$ support a proposal. Estimate the support proportion.

Solution Steps
  1. Use the number supporting the proposal as successes.
  2. Divide by the sample size.
  3. State the estimate as a sample result, not a certainty.
$$\hat p=\frac{120}{200}$$$$\hat p=0.60$$$$\boxed{\text{Estimated support is }60\%\text{ for the sampled population.}}$$