Concise answer
Design determines what a study can claim: only an experiment — a manipulated variable with participants randomly assigned to experimental and control groups — supports cause-and-effect conclusions, while correlational research reports the strength and direction of a relationship via r and is always exposed to confounding variables. Separately, every measure must be judged on reliability (consistency, in three standard forms) and validity (accuracy).
Definitions
- Random assignment
- Placing participants into experimental and control groups so each has an equal chance of either group, spreading pre-existing differences across conditions.
- Correlation coefficient (r)
- A number from -1 to +1 whose sign gives the direction of a relationship between variables and whose distance from 0 gives its strength.
- Confounding variable
- An outside factor that drives both variables of interest, creating a relationship between them that is not causal.
- Reliability
- The consistency of a measure — across raters (inter-rater), across its own items (internal consistency), and across occasions (test-retest).
- Validity
- The extent to which an instrument measures what it is supposed to measure; a measure can be reliable without being valid.
Intuition
Random assignment is the whole causal trick: by giving every participant an equal chance of each group, it spreads unmeasured differences — including any would-be confound — evenly across conditions, so a difference in outcomes has only the manipulation left to explain it. Correlational designs have no such eraser, which is why their conclusions stop at association.
Reliability and validity answer different questions about a measure: a bathroom scale that always reads four kilograms heavy is perfectly consistent (reliable) and consistently wrong about your actual weight (invalid). Consistency is necessary for a good measure but never sufficient.
Concept walkthrough
Research approaches trade control for realism. Naturalistic observation records behavior unobtrusively in real settings, earning high ecological validity and better generalization to the real world, but surrenders control; surveys, case studies, archival research, and longitudinal or cross-sectional designs make their own trades among cost, control, and generalizability. The experiment sits at the control extreme: the researcher manipulates a variable, holds conditions constant, and compares an experimental group against a control group.
The experiment's causal license comes from random assignment: because every participant has an equal chance of either group, pre-existing differences are distributed across conditions instead of piling up in one, and outcome differences can be attributed to the manipulation. Correlational research instead measures variables as they occur and summarizes the relationship with the correlation coefficient r, a number from -1 to +1 whose sign is direction and whose distance from zero is strength. However strong, r cannot settle causation: the direction of influence is undetermined, and a confounding variable — like temperature driving both ice cream sales and crime rates — may be producing the whole relationship.
Whatever the design, its measurements must themselves be trustworthy. Reliability is consistency, standardly split three ways: inter-rater reliability (do observers agree?), internal consistency (do a scale's items agree with each other?), and test-retest reliability (does the score hold across occasions?). Validity is accuracy — whether the instrument measures the construct it claims to. GRE Psychology's Measurement/Methodology/Other questions lean on exactly these discriminations: which design was used, what it can conclude, and whether the measure deserves trust.
After this page, you should be able to
- Classify a described study as experimental or correlational and state what conclusions that design licenses.
- Interpret a correlation coefficient's sign and magnitude, and explain why correlation does not establish causation.
- Propose a plausible confounding variable for a reported correlation and say how random assignment addresses it.
- Distinguish the three standard forms of reliability from validity when evaluating a psychological measure.
Formulas and assumptions
Correlation coefficient range
Variables
- r: correlation coefficient between two measured variables
- sign: positive (variables move together) or negative (they move oppositely)
- |r|: strength — closer to 1 means a more predictable relationship
Assumptions
- r summarizes a linear relationship between measured variables; it never by itself supports a causal claim.
Causal-inference checklist
manipulated variable + random assignment + control group => cause-and-effect conclusion permitted
Variables
- manipulated variable: the factor the researcher changes
- random assignment: equal chance of each group for every participant
- control group: the comparison condition without the manipulation
Assumptions
- Schematic of the experimental-design requirements described in OpenStax Psychology 2e Section 2.3, not a quantitative formula.
- If any element is missing, the design is observational or correlational and causal answer options are wrong.
Reliability type map
raters agree = inter-rater; items agree = internal consistency; occasions agree = test-retest
Variables
- raters: independent observers scoring the same behavior
- items: questions within a single instrument
- occasions: administrations of the same measure at different times
Assumptions
- All three are forms of consistency; none of them establishes that the measure is valid.
Worked example
What a planner-GPA correlation can and cannot claim
A campus study reports that students who use paper planners have higher GPAs, with r = +0.4. A headline concludes that planners raise grades. Evaluate the conclusion and design a study that could actually test it.
- 1Classify the design: students were measured as found, with no manipulation and no random assignment — this is correlational.
- 2Read the coefficient: r = +0.4 is a positive, moderate relationship — planner use and GPA move together, with predictive but not causal meaning.
- 3Attack the causal leap: a confounding variable such as conscientiousness could drive both planner-buying and studying; reverse influence (strong students adopting planners) is also uneliminated.
- 4Design the experiment: recruit a sample, randomly assign students to a planner group or a no-planner control group, run the semester, and compare mean GPA between groups.
- 5State the licensed conclusion: only with the manipulation plus random assignment would a group difference in GPA support the headline's causal claim.
The headline overreaches: the correlational design supports only 'planner use and GPA are moderately positively related.' A randomized experiment assigning students to planner and control groups is the design that could test whether planners raise grades.
Common traps
- Accepting a causal answer option for a correlational study — no manipulation plus no random assignment means no cause-and-effect conclusion.
- Reading a negative r as a weak relationship; the sign is direction, and r = -0.8 is stronger than r = +0.4.
- Confusing random assignment (to groups, for causal inference) with random sampling (from the population, for generalization).
- Treating a reliable measure as automatically valid; consistent scores can be consistently measuring the wrong thing.
Related pages and practice
Question depth and domain coverage vary by exam. Practice answers are checked after submission.
Sources
- GRE Subject Test Content and Structure — ETS. Accessed 2026-07-06. Use as a cited source for exam facts; do not imply affiliation or reproduce protected test material.
- Psychology 2e, Section 2.2: Approaches to Research — OpenStax. Accessed 2026-08-03. OpenStax textbook content is CC BY-NC-SA 4.0; attribute and avoid verbatim reuse beyond short cited references.
- Psychology 2e, Section 2.3: Analyzing Findings — OpenStax. Accessed 2026-08-03. OpenStax textbook content is CC BY-NC-SA 4.0; attribute and avoid verbatim reuse beyond short cited references.
Review and maintenance
Source-checked by publisher- Publisher record
- Keiko Study editorial owner
- Publisher placeholder; no individual credential claim is made yet.
- Reviewer record
- Technical reviewer pending
- Placeholder only; this content is not labeled as reviewed by a named specialist.
- Last source check
- 2026-08-03
- Next scheduled review
- 2026-11-03
Recheck the ETS content-structure page and the OpenStax Psychology 2e research chapter sections before each major GRE Psychology preparation cycle.