AP Psychology · Cross-Unit · 2026 Exam

Scientific Foundations

Research methods, study design, sampling, statistics, and APA research ethics — tested across all 5 units and directly in the AAQ and EBQ. Mastering this section multiplies the value of every unit you study.

All Units2026 CED Science PracticesAAQ & EBQ CriticalHigh Yield
M.1

Science Practices Overview

The AP Psychology CED defines four science practices that are assessed throughout all units and especially in the two free-response questions (AAQ and EBQ). Understanding these practices is essential because they define the skills that the exam tests, not just the content.

PracticeWhat It AsksWhere It Appears
Concept ApplicationApply psychological perspectives, theories, concepts, and research findings to a described scenario; explain how cultural norms, cognitive biases, and contextual circumstances shape behavior and mental processesAll MCQ scenarios; FRQ concept application sections; EBQ synthesis and claim
Research Methods and DesignDetermine the type of research design used in a given study; evaluate the appropriate use of research design elements; identify threats to validity; assess sampling methods; identify ethical guidelines in researchFRQ design identification and causal-conclusion questions; MCQs about methodology and ethics (Skill 2.D)
Data InterpretationRead, analyze, and interpret quantitative data from tables, graphs, charts, diagrams, and study descriptions; draw accurate, appropriately qualified conclusions; recognize the limitations of dataMCQs presenting data tables, figures, or study results; identifying what conclusions a study can and cannot support
ArgumentationConstruct or evaluate an evidence-based argument or claim; use evidence from sources to support, qualify, or challenge a psychological claim; provide reasoning that connects evidence to claimPrimarily the EBQ (Evidence-Based Question): make a claim, select supporting evidence from multiple sources, and justify the reasoning that connects evidence to the claim
Why Research Methods Are High-Yield

Research methods questions appear in every unit because any scenario involving a study can be asked about for its design, controls, sampling method, statistical conclusions, or ethical considerations. On the AAQ specifically, Part A almost always asks to identify the research design, and Part B often asks whether the study can support a causal conclusion (requiring understanding of why random assignment matters).

M.2

Experimental Methodology

An experiment is the only research design that allows causal conclusions. The key requirement is random assignment of participants to conditions, which controls for pre-existing differences.

Core Experimental Vocabulary

TermDefinitionExam Tip
Independent Variable (IV)The variable the researcher deliberately manipulates; the proposed causeMust be named specifically: "whether participants received X or Y" not just "the treatment"
Dependent Variable (DV)The variable that is measured; the proposed effect; what changes in response to the IVMust be measurable and operationally defined: "score on the depression inventory" not just "depression"
Operational DefinitionA precise, testable definition of how a variable is measured or manipulated in a specific studyCritical for replication; e.g., "stress" operationalized as "cortisol level in saliva" or "score on the PSS"
Experimental GroupThe group that receives the treatment (the IV)Must be compared to a control group — without comparison, no conclusions can be drawn
Control GroupThe group that does not receive the treatment; provides a baseline for comparisonIdeally identical to experimental group in all ways except the IV
Random AssignmentRandomly placing participants into experimental vs. control conditions so pre-existing differences are equally distributed across groups by chanceThe single most important feature for establishing causation; different from random sampling
Confounding VariableA third variable that correlates with both the IV and DV, potentially explaining the result without the IV being the actual causeControlled by random assignment in experiments; a major threat to validity in non-experimental designs
PlaceboAn inert treatment that appears identical to the real treatment, given to the control groupEnsures any outcome differences are due to the active ingredient, not merely the expectation of treatment
Placebo EffectThe improvement in symptoms caused by the expectation of receiving a beneficial treatment, even when the treatment is inertEspecially large in pain and mood research; reason why control groups must receive a placebo, not nothing
Single-Blind ProcedureParticipants do not know which condition (experimental or control) they are in, but the researchers doControls for participant expectancy and demand characteristics
Double-Blind ProcedureNeither participants nor the researchers administering treatment know which condition each participant is inControls for both participant expectancy AND experimenter bias simultaneously; gold standard for drug trials
Experimenter BiasThe tendency for researchers to unconsciously influence participants or interpret data in ways that confirm their hypothesisControlled by double-blind procedures and using standardized instructions
Social Desirability BiasThe tendency for participants to respond in ways they believe are socially acceptable rather than truthfullyControlled by anonymous surveys, physiological measures, or behavioral observation rather than self-report
The Causation Rule

Random assignment → causal conclusion possible. This is the most tested relationship in research methods. When a study uses random assignment, any difference in the DV between groups can be attributed to the IV. Without random assignment (in quasi-experiments or correlational studies), only association — not causation — can be concluded, regardless of how large the effect is.

MCQ · Experimental Design

Researchers randomly assigned 100 participants to receive either a new anxiety treatment (experimental group) or a placebo pill (control group) for 8 weeks. Neither participants nor the clinicians assessing outcomes knew which condition each participant was in. Results showed significantly lower anxiety scores in the experimental group. Which conclusion is best supported?

  • (A) The new treatment is associated with lower anxiety, but causation cannot be determined
  • (B) The new treatment caused a reduction in anxiety scores compared to placebo
  • (C) The results may be due to the placebo effect, because participants did not know which group they were in
  • (D) Random assignment was not necessary because the groups were compared to each other
Answer: (B) — Random assignment controlled for pre-existing differences between groups. The double-blind procedure controlled for both placebo effects and experimenter bias. Therefore, the difference in anxiety scores between groups can be causally attributed to the treatment. (A) is wrong because random assignment DOES allow causal conclusions. (C) is wrong because the double-blind procedure controls for placebo effects by giving both groups pills — if the placebo effect were responsible, both groups would improve equally.
M.3

Non-Experimental Research Methods

Non-experimental methods describe, predict, or find associations between variables but cannot establish causation because there is no random assignment and no manipulation of variables.

MethodDefinitionStrengthsLimitations
Correlational ResearchMeasuring the statistical relationship between two or more variables as they naturally occur, without manipulationReveals real-world relationships; can handle variables that can't be manipulated; can use existing dataCannot establish causation; third variable (confound) problem; directionality problem (which causes which)
Case StudyAn in-depth investigation of a single individual, group, or event; typically uses interviews, observations, and recordsRich, detailed data; can study rare phenomena; generates hypothesesCannot generalize to larger populations; researcher bias in interpretation; idiosyncratic results
Naturalistic ObservationSystematically observing and recording behavior in natural settings without manipulating variablesHigh ecological validity; captures real behaviorObserver effect (behavior changes when observed); time-consuming; cannot control confounds; no causation
Survey / Self-ReportAsking a sample of participants questions about their attitudes, behaviors, or experiencesEfficient; reaches large samples; low costSocial desirability bias; self-report error; question wording effects; sampling bias
Meta-AnalysisA statistical procedure that combines and analyzes the results of multiple independent studies on the same question to reach an overall quantitative conclusionIncreases statistical power; reveals overall effect size across studies; identifies moderating variables"Garbage in, garbage out" — if individual studies are flawed, results are compromised; publication bias
Correlation ≠ Causation

The most repeatedly tested reasoning error in AP Psychology. Even a perfect correlation (r = 1.0 or r = −1.0) between two variables does not mean one causes the other. Two alternative explanations always exist: (1) reverse causation — maybe B causes A rather than A causing B; (2) third variable problem — some unmeasured variable C may cause both A and B independently. Example: ice cream sales and drowning rates are positively correlated — because both are caused by hot weather, not because ice cream causes drowning.

M.4

Sampling & Study Design

ConceptDefinitionWhy It Matters
PopulationThe entire group the researcher wants to draw conclusions aboutResults are only generalizable to the population that the sample represents
Representative SampleA sample that accurately reflects the characteristics of the population; key requirement for generalizabilityIf the sample is not representative, findings cannot be generalized; the main goal of sampling
Random SamplingEvery member of the population has an equal chance of being selected for the sampleMaximizes representativeness; different from random assignment (which controls for pre-existing differences within the study)
Convenience SamplingSelecting participants who are easily available (e.g., college students, volunteers)Common in psychology research; may not be representative; limits external validity
Sampling BiasWhen the sample systematically differs from the population in a way that distorts resultsExample: WEIRD bias (Western, Educated, Industrialized, Rich, Democratic participants dominate psychology research)
External ValidityThe degree to which results can be generalized beyond the specific study to other people, settings, and timesHigh in naturalistic studies; low in laboratory experiments with convenience samples
Internal ValidityThe degree to which observed differences in the DV can be attributed to the IV rather than to confounding variables or alternative explanations; the confidence that the IV caused the DV changeHighest in well-controlled experiments with random assignment; threatened by confounding variables, selection bias, and design flaws. Note: distinct from construct validity (whether the measure captures the intended concept) and external validity (generalizability)
Random Assignment vs. Random Sampling

These are the most commonly confused terms on the exam. Random sampling = selecting participants from the population (addresses who is in the study; improves generalizability). Random assignment = assigning selected participants to experimental conditions (addresses which condition each person is in; enables causal conclusions). A study can have one, both, or neither.

M.5

Data Interpretation & Statistics

Measures of Central Tendency and Variation (Science Practice 3.B)

Science Practice 3.B requires the ability to calculate and interpret measures of central tendency, variation, and percentile rank. These appear in data interpretation MCQs and in FRQ scenarios presenting research data.

MeasureDefinitionWhen to Use / Exam Tip
MeanThe arithmetic average; sum of all scores divided by the number of scoresMost common measure of central tendency; sensitive to outliers — one extreme score can dramatically pull the mean. If a dataset is skewed, mean ≠ median
MedianThe middle value in a ranked (sorted) distribution; the score that splits the dataset in halfBest measure when distribution is skewed or contains outliers; unaffected by extreme scores. Example: median household income is reported instead of mean because a few billionaires would inflate the mean
ModeThe most frequently occurring score in a distribution; the only measure of central tendency usable with categorical dataUseful for identifying the most common response. A distribution can have no mode, one mode (unimodal), or multiple modes (bimodal, multimodal)
RangeThe difference between the highest and lowest scores; the simplest measure of variabilityQuick indicator of spread; heavily influenced by outliers
Standard DeviationThe average amount by which scores deviate from the mean; the most commonly used measure of variabilityLarge SD = scores spread widely around mean; small SD = scores clustered near mean. In a normal distribution, ~68% of scores fall within 1 SD of the mean, ~95% within 2 SDs
Percentile RankThe percentage of scores in a distribution that fall at or below a given score; indicates where a score falls relative to the rest of the distributionExample: a score at the 84th percentile means 84% of scores in the sample scored at or below that level. In a normal distribution, a score 1 SD above the mean falls at approximately the 84th percentile
Mean, Median, Mode in Skewed Distributions

In a positively skewed distribution (tail extends right; e.g., income): Mode < Median < Mean. The mean is pulled right by high outliers.
In a negatively skewed distribution (tail extends left; e.g., easy test scores): Mean < Median < Mode. The mean is pulled left by low outliers.
In a normal distribution: Mean = Median = Mode.

Descriptive Statistics

ConceptDefinitionApplication
Normal Distribution (Normal Curve)A symmetrical, bell-shaped distribution in which the mean, median, and mode are equal and data cluster around the center; most psychological measurements approximate a normal distributionAllows use of standard statistical tests; describes how most trait scores (IQ, personality) distribute in the population
Standard Deviation (SD)A measure of variability that indicates the average amount by which scores deviate from the mean; large SD = more spread; small SD = more clustered around meanAllows comparison of score distributions; a score 2 SDs above the mean is unusually high regardless of the scale
Regression Toward the MeanThe statistical tendency for extreme scores on a first measurement to be closer to the mean on a second measurement, not because of any treatment but purely due to chance variabilityExplains why "treatments" administered after extremely bad performance appear to work even if ineffective; a major confound in before-after studies without a control group
Positive CorrelationAs one variable increases, the other tends to increase; correlation coefficient between 0 and +1.0r = +0.80: strong positive correlation. Example: hours of study and exam score
Negative CorrelationAs one variable increases, the other tends to decrease; correlation coefficient between 0 and −1.0r = −0.70: strong negative correlation. Example: stress level and immune function
Correlation Coefficient (r)A number from −1.0 to +1.0 that indicates both the direction (sign) and strength (absolute value) of the relationship between two variablesStrength: closer to ±1.0 = stronger. r = 0.00 = no relationship. Direction: positive vs. negative sign

Inferential Statistics (Applied)

Statistical Significance

A result is statistically significant when it is unlikely to have occurred by chance alone. The conventional threshold in psychology is p < .05 (less than 5% probability the result is due to chance). Statistical significance does NOT mean the result is important or large — only that it is reliable enough to not be random noise.

Effect Size

A measure of the magnitude of a result, independent of sample size. Statistical significance can be achieved with tiny effects in large samples. Effect size (e.g., Cohen’s d) tells you whether the result is practically meaningful. Common in meta-analyses. On the AP exam, focus on understanding that significance ≠ importance.

MCQ · Statistics & Correlation

A researcher finds a correlation of r = −0.78 between the number of hours of sleep per night and reported levels of anxiety. Which of the following is the most accurate interpretation?

  • (A) Anxiety causes people to sleep fewer hours, as shown by the negative correlation
  • (B) Fewer hours of sleep causes greater anxiety
  • (C) There is a strong negative relationship between sleep hours and anxiety, but causation cannot be determined
  • (D) A correlation of −0.78 indicates a weak negative relationship
Answer: (C) — r = −0.78 is a strong negative correlation (closer to −1.0 than 0). It means that as sleep hours increase, anxiety tends to decrease (or as sleep hours decrease, anxiety tends to increase). However, this is correlational data — no random assignment, no manipulation of variables — so causation cannot be determined. The relationship could go either direction: maybe anxiety causes poor sleep, or maybe a third variable (stress, caffeine, exercise) causes both. (D) is wrong: 0.78 is a strong correlation, not weak.
M.6

APA Research Ethics

APA ethical guidelines ensure that research participants are treated with respect and dignity. The same four ethical principles that apply to clinical practice (see Unit 5) also guide research conduct, but research ethics adds specific procedural requirements:

Ethical RequirementDefinitionApplication in Research
Informed ConsentParticipants must be told the nature of the study, procedures, risks, and benefits before they agree to participate; participation must be voluntarySigned consent forms; explaining what the study involves; emphasizing that withdrawal at any time is acceptable without penalty
DebriefingAfter the study (especially if deception was used), participants must be fully informed of the true purpose, procedures, and any deceptionEspecially required when deception was used (e.g., Milgram obedience studies); ensures no lasting harm from misinformation
ConfidentialityParticipants' individual data must be kept private; results reported as group aggregatesAnonymizing data; protecting records; IRB oversight requires confidentiality provisions
Protection from Harm & Institutional ReviewResearchers must take all reasonable steps to protect participants from physical or psychological harm; benefits must outweigh risks. Research involving human participants requires prior review and approval by an Institutional Review Board (IRB). Research involving non-human animals also requires institutional review — an Institutional Animal Care and Use Committee (IACUC) evaluates whether the research is justified and whether animal welfare standards are metIRB required for all human research; institutional review required for non-human animal research; stopping criteria if distress develops; ongoing monitoring throughout the study
Voluntary Participation / Right to WithdrawParticipants must be free to withdraw from the study at any time without penalty or loss of benefits already earnedCannot coerce participation; cannot punish withdrawal; cannot withhold earned payment for withdrawal
Deception (Limited Use)Researchers may use deception only when it is necessary for valid research and the research could not be done otherwise; must be followed by full debriefingClassic use: withholding the true hypothesis to prevent demand characteristics; requires IRB approval and full debriefing
Ethics in FRQs — Skill 2.D

Science Practice 2.D assesses the ability to identify an ethical guideline applied in a study and describe how the researchers followed it. This is frequently assessed in source-based written questions in Unit 5. Practice this template: “The ethical guideline is [name]. The researchers applied this by [specific action from the study description].” The two parts must match: identify the specific principle, then connect it to specific evidence from the source.

Practice

Comprehensive Practice Questions

MCQ · Design Identification

A school psychologist wants to determine whether a new anti-bullying program reduces aggression. She selects all 6th-graders in a school district and randomly assigns each student to either receive the program or not. Aggression is measured via teacher ratings before and after the 10-week program. Which research design is this, and what conclusion does it support?

  • (A) Correlational study; can determine that the program is associated with reduced aggression
  • (B) Experiment; can determine that the program caused a reduction in aggression
  • (C) Quasi-experiment; can suggest an association but not causation because random assignment was used
  • (D) Naturalistic observation; can describe aggression patterns as they naturally occur
Answer: (B) — Random assignment is present (students randomly assigned to program vs. control). This is the defining feature of a true experiment. Because random assignment controls for pre-existing differences between groups, any difference in aggression ratings at the end of the 10 weeks can be causally attributed to the anti-bullying program. (C) is wrong: quasi-experiments do NOT use random assignment — this study does use random assignment, making it a true experiment.
MCQ · Sampling & Validity

A researcher studies stress and academic performance by recruiting 200 undergraduate psychology students at a single large university who volunteered to participate. Results show a significant negative correlation between stress and GPA. Which is the primary limitation of this study?

  • (A) The study cannot establish that stress affects academic performance because only correlational data were collected
  • (B) The study lacks internal validity because stress was not randomly assigned
  • (C) The use of a convenience sample limits the generalizability of the findings to other populations
  • (D) The sample size of 200 is too small to yield statistically significant results
Answer: (C) — Using voluntary undergraduate psychology students from one university is a convenience sample. This limits external validity (generalizability) — the findings may not apply to high school students, older adults, students at other institutions, or non-volunteers (who may differ systematically from volunteers). (A) is also true but is not the "primary limitation" asked for — the correlation vs. causation issue is an expected feature of this design, not an additional flaw. (B) is incorrect: internal validity concerns confounds within the study, not sampling.
FRQ-Style · AAQ Design Analysis

Researchers randomly assigned 80 participants with depression to either 12 weeks of cognitive-behavioral therapy (CBT) or a waitlist control condition. Depression was assessed using the Beck Depression Inventory (BDI) at baseline and after 12 weeks. The clinicians who scored the BDI did not know which condition each participant had been assigned to. Results showed significantly lower BDI scores in the CBT group (p = .002). Prior to the study, all participants provided written informed consent and the study was approved by an Institutional Review Board (IRB).

(a) Identify the independent variable and the dependent variable.
(b) Explain why this study can support a causal conclusion about CBT and depression. Refer to a specific design feature.
(c) Identify one potential confounding variable NOT controlled by this design and explain why it is a threat.
(d) Identify one ethical guideline that researchers followed in this study. Describe how they applied it.

(a) IV: Whether the participant received CBT or was placed on the waitlist control condition. DV: Depression score as measured by the Beck Depression Inventory (BDI) at the end of the 12-week period.

(b) The study uses random assignment of participants to conditions. Random assignment distributes pre-existing differences (baseline depression severity, personality traits, coping skills, life circumstances) equally across both conditions by chance. Therefore, when CBT participants show lower BDI scores at the end of 12 weeks, this difference can be causally attributed to the CBT itself rather than to pre-existing differences between the groups.

(c) One potential confound: Therapeutic contact and expectancy effects. Participants in the CBT condition receive regular one-on-one contact with a therapist for 12 weeks. Even if CBT techniques have no specific benefit, the attention, support, and expectation of improvement from regular therapeutic contact could reduce depression. The waitlist control group receives no contact at all. This means any improvement in the CBT group could be attributed to non-specific therapeutic contact rather than the CBT technique itself. To control for this, the study would need an active control condition (e.g., supportive listening therapy providing equivalent contact time) rather than a no-treatment waitlist.

(d) Ethical guideline: Institutional review / protection from harm. Research involving human participants must receive prior approval from an Institutional Review Board (IRB), which evaluates whether potential benefits justify risks, confirms that participant protections are in place, and ensures the study follows APA ethical principles. The researchers applied this by obtaining IRB approval before beginning the study — the IRB reviewed the procedures to confirm they were appropriate for a vulnerable population (people with depression) and that safeguards (monitoring for worsening symptoms, right to withdraw, referral to other services) were in place.

Alternative acceptable answer: Informed consent. All participants provided written informed consent before enrollment, meaning they were told the study’s purpose, that they might be assigned to a waitlist (receiving no treatment for 12 weeks), and that they could withdraw at any time without penalty.
Common Mistakes

High-Frequency Research Methods Errors

AP® PsychologySophriva · sophriva.com