Statistics and Experimental Design: Key Sampling and Study Concepts

Simple Random Sample (SRS)

A sample in which every group of size n has an equal chance to be chosen.

Stratified Random Sample

Population divided into homogeneous subgroups (strata), and then a random sample is taken from each subgroup to create the sample

Cluster Sample

Population divided into random heterogeneous clusters, and everyone in all clusters is sampled.

Systematic Sample

Select every “xth” individual from a list after a random start.

Convenience Sample

Individuals easiest to reach are chosen; produces bias.

Voluntary Response Sample

People choose to participate, often producing extreme opinions; not representative.

Population

the entire group you want information about.

Sample

The group of the population actually studied.

Bias

A systematic error in a study that favors certain outcomes, leading to an inaccurate representation of the population or an incorrect conclusion.

Undercoverage

when parts of the population are left out of the sampling frame.

Nonresponse

Chosen individuals do not respond.

Response Bias

Pattern of inaccurate answers due to wording, interviewer influence, or dishonesty.

Sampling Frame

List of individuals from which the sample is chosen.

Experiment

A study in which a treatment is imposed on individuals to measure its effect.

Observational Study

Observes individuals and measures variables of interest but does not attempt to influence responses.

Experimental Unit

Object studied in an experiment; called a subject when human.

Treatment

A specific condition applied to the experimental units.

Factor

An explanatory variable in an experiment.

Level

The specific value of a factor. Example: drug dosage of 50 mg, 100 mg, or 150 mg.

Response Variable

Measures the outcome of a study.

Control

Keep other variables constant; compare treatment and control groups.

Randomization

Assign treatments randomly to reduce bias.

Replication

Use enough subjects to reduce variability in results.

Comparison

Compare two or more treatments or treatment vs control.

Placebo

Inactive treatment used for comparison.

Placebo Effect

Subjects respond to belief in treatment rather than actual treatment.

Blinding

Subjects or evaluators don't know who received the treatment.

Double-Blind

Both subjects and evaluators are unaware of treatment assignment.

Confounding Variable

A variable related to both the explanatory and response variables, making it unclear which caused the effect.

Lurking Variable

Hidden variable influencing both variables under study, creating a false association.

Completely Randomized Design

All experimental units are assigned to treatments entirely by chance.

Randomized Block Design

Experimental units grouped into blocks based on similar traits, then randomly assigned to treatments within each block.

Matched Pairs Design

Each subject receives both treatments in random order, or two very similar subjects are paired.

Scope of Inference

Determines whether we can generalize results to a population or claim causation.

Random Selection

Allows inference about the population.

Random Assignment

Allows inference about cause and effect.

Institutional Review Board (IRB)

Group that reviews studies to protect rights and welfare of human subjects.

Informed Consent

Subjects agree in writing to participate after being informed of risks.

Confidentiality

Researchers keep participant data private.

Factor vs Level

Factor is the explanatory variable; levels are the specific values of that factor.

CRRC (Design Principles)

Control, Randomize, Replicate, Compare — the 4 key principles of experimental design.