I85PREP
⌘ K
Psychology study materials
Psychology study resources

Psychology: Research methods and statistics (design, measurement, data analysis, ethics)

Psychology · Work through your exam, one topic at a time.

All guide sections

Research methods and statistics (design, measurement, data analysis, ethics)

This domain tests whether you can judge what a study can and cannot support. Items ask you to identify the design, the threat to validity, or the correct interpretation of a statistic.

Design

  • True experiments manipulate an independent variable with random assignment, supporting causal inference (internal validity). Quasi-experiments use intact groups or preexisting variables and are vulnerable to selection threats.
  • Correlational designs cannot establish directionality or rule out third variables. Longitudinal designs follow the same people; cross-sectional designs compare cohorts and confound age with cohort.
  • Random sampling affects external validity; random assignment affects internal validity. Do not swap them.
  • Controls: placebo, double blinding, counterbalancing for order effects, matching, manipulation checks. Threats: attrition, regression to the mean, demand characteristics, experimenter expectancy, maturation, history, testing effects.

Measurement and ethics

  • Operationalize constructs, then check construct validity, reliability, and floor or ceiling effects.
  • Ethics: IRB approval, informed consent, minimizing harm, confidentiality, debriefing after justified deception, protections for vulnerable populations, humane animal care.

Statistics

  • Descriptive: mean, median, mode; range, variance, SD. In a normal curve about 68% of scores fall within 1 SD, 95% within 2, 99.7% within 3. z = (X minus mean) divided by SD.
  • r ranges from -1 to 1; r-squared is shared variance, so r = .50 means 25% shared variance.
  • Inference: standard error of the mean shrinks as n grows; t-tests compare two means, ANOVA compares three or more and reveals interactions in factorial designs, chi-square tests association among categorical variables.
  • Errors: Type I is rejecting a true null (probability alpha, usually .05); Type II is failing to reject a false null (probability beta). Power = 1 minus beta and rises with larger n, larger effects, and lower alpha (which raises beta).
  • p is the probability of data this extreme if the null is true, not the probability the null is true. Effect sizes (Cohen's d around .2, .5, .8) describe magnitude.

Worked example

A researcher runs 20 independent tests at alpha = .05, all with true nulls. Reasoning: expect about 20 times .05, that is one false positive. A Bonferroni-type correction lowers alpha per test, controlling familywise error while raising Type II risk.

Traps

Treating correlation as causation; reading p = .03 as a 3% chance the null is true; confusing SD (spread of scores) with standard error (spread of sample means); mixing random assignment with random sampling.