Final Exam

Question 1

Regression

Answer 1

Statistical model predicting outcome Y from predictors X.

Question 2

Multiple Regression

Answer 2

Uses two or more predictors to explain variance in Y.

Question 3

Unstandardised Coefficient (b)

Answer 3

Change in Y (in its units) per one unit of X.

Question 4

Standardised Coefficient (β)

Answer 4

Change in Y (in SD units) per one SD of X.

Question 5

Intercept (b■)

Answer 5

Expected value of Y when all Xs = 0.

Question 6

Residual (ε)

Answer 6

Difference between observed and predicted Y.

Question 7

Assumptions

Answer 7

LINE — Linearity, Independence, Normality, Equality of variance.

Question 8

Violation Effects

Answer 8

Reduce precision, widen confidence intervals.

Question 9

T-tests and Regression

Answer 9

t = b / SE, t-tests are special cases of regression.

Question 10

Model equation

Answer 10

Y = b■ + b■X■ + b■X■ + … + ε

Question 11

R²

Answer 11

Proportion of variance in Y explained by predictors.

Question 12

Goal of R²

Answer 12

Identify unique contribution of each predictor to Y.

Question 13

ANOVA

Answer 13

Compares means across groups; extension of regression for categorical predictors.

Question 14

Sums of Squares SST:

Answer 14

SST: Total variance in data.

Question 15

Sums of Squares SSM (Model):

Answer 15

SSM (Model): Variance explained by group means.

Question 16

Sums of Squares SSR (Residual)

Answer 16

SSR (Residual): Variance not explained (error).

Question 17

R²

Answer 17

SSM / SST; variance explained by predictors.

Question 18

F-Test

Answer 18

Ratio of model variance to residual variance; tests overall model fit.

Question 19

η² (Eta Squared)

Answer 19

Proportion of variance in Y explained by group membership.

Question 20

df■ (Between Groups)

Answer 20

k – 1

Question 21

p-value

Answer 21

Probability of data given null hypothesis (H■).

Question 22

Null Hypothesis

Answer 22

Assumes no group difference or association.

Question 23

NHST
Null Hypothesis Significance testing

Answer 23

Combines Fisher’s evidence and Neyman–Pearson decision traditions.

Question 24

Statistical Significance

Answer 24

Arbitrary threshold (usually α = .05).

Question 25

Type I Error

Answer 25

False positive (rejecting true null).

Question 26

Type II Error

Answer 26

False negative (failing to reject false null).

Question 27

Effect Size

Answer 27

Quantifies magnitude of effect (independent of sample size).

Question 28

Confidence Interval

Answer 28

Range of plausible values for a parameter.

Question 29

Key Idea

Answer 29

Interpret p-values with effect size, CIs, and context.

Question 30

Causal Criteria

Answer 30

Covariance, Temporal Precedence, Internal Validity.

Question 31

Cross-sectional Design

Answer 31

Measures variables once; cannot infer causality.

Question 32

Longitudinal Design

Answer 32

Measures same variables repeatedly over time.

Question 33

Cross-lag Correlation

Answer 33

Tests if earlier X predicts later Y.

Question 34

Autocorrelation

Answer 34

Correlation of variable with itself across time.

Question 35

Statistical Control

Answer 35

Using multiple regression to adjust for third variables.

Question 36

Mediation

Answer 36

X affects Y through mediator M (explains why).

Question 37

Moderation

Answer 37

Effect of X on Y depends on moderator Z (explains when/for whom).

Question 38

Confound

Answer 38

Variable influencing both X and Y, distorting true effect.

Question 39

Path Diagram

Answer 39

Visual map of hypothesised causal relations.

Question 40

Purpose of correlational approaches to causal claims

Answer 40

Illustrate theory, test mechanisms, clarify assumptions.

Question 41

Path Diagrams (DAGs)

Answer 41

Depict causal assumptions; plan before data collection.

Question 42

Bailey et al. (2024) Four Common Problems

Answer 42

1. Vague Research Questions – unclear cause/effect. 2. Missing Confounds – unmeasured variables affecting X and Y. 3. Over-Control – controlling for mediators or colliders 4. Fat-Handed Interventions – manipulations changing multiple factors.

Question 43

Collider

Answer 43

Variable influenced by both X and Y; controlling for it induces spurious correlation.

Question 44

Over-Control Problem

Answer 44

Removing causal pathway of interest.

Question 45

Under-Control Problem

Answer 45

Failing to account for confounds.

Question 46

Research Specificity

Answer 46

Clear cause, effect, population, and timing.

Question 47

Responsible Causal Claims

Answer 47

Transparent assumptions, measured confounds.

Question 48

True Experiment

Answer 48

Manipulation of IV, random assignment, and control group.

Question 49

Random Assignment

Answer 49

Equalises groups, reduces confounds.

Question 50

Manipulation

Answer 50

Establishes temporal precedence.

Question 51

Control Group

Answer 51

Provides baseline for comparison.

Question 52

Between-Subjects Design

Answer 52

Different participants per condition.

Question 53

Within-Subjects Design

Answer 53

Same participants in all conditions (counterbalanced).

Question 54

Research Design Notation (R,X,O)

Answer 54

R = Random assignment, X = Treatment, O = Observation. EX. R: O■ X O■ / R: O■ – O■

Question 55

Threats to Internal Validity

Answer 55

Selection Bias: Pre-existing group differences. Maturation: Natural change over time. History: Events outside study influence results. Testing Effects: Measurement alters behaviour.

Question 56

Prevention of Threats to internal validity

Answer 56

Random assignment and control groups.

Question 57

Regression to the Mean (RTM)

Answer 57

Extreme scores move toward the mean on retest.

Question 58

Mechanism for RTM

Answer 58

Observed score = True score + Random error.

Question 59

Threats to RTM

Answer 59

Creates illusion of treatment effect after selecting extreme cases.

Question 60

Prevention of threats to RTM

Answer 60

Use control groups equally extreme at baseline.

Question 61

Control Group Purpose for RTM

Answer 61

Distinguishes real change from statistical artifact.

Question 62

Key Message of RTM

Answer 62

RTM is universal; interpret changes cautiously.

Question 63

Quasi-Experiment

Answer 63

No random assignment; still aims for causal inference.

Question 64

Quasi-IV

Answer 64

Naturally occurring variable (e.g., policy, group status).

Question 65

Common Designs of Quasi-IV

Answer 65

1. Posttest-only Nonequivalent Groups: X O / – O 2. Pretest–Posttest Nonequivalent Groups: O X O / O – O 3. Interrupted Time Series: O O O X O O O

Question 66

Threats to Quasi-IV

Answer 66

Selection effects, history, testing, maturation.

Question 67

Small-N Designs

Answer 67

Intensive designs with few participants.

Question 68

Stable Baseline

Answer 68

Repeated baseline before treatment.

Question 69

Multiple Baseline

Answer 69

Staggered interventions across people/behaviours.

Question 70

Reversal (ABA)

Answer 70

Apply and remove treatment to observe pattern.

Question 71

Internal Validity

Answer 71

Strong via repeated measures, control, replication.

Question 72

External Validity

Answer 72

Context-dependent; triangulate with other studies.

Question 73

Epistemic Humility

Answer 73

Evidence is conditional; causal inference has limits.

Question 74

Responsible Research

Answer 74

Transparent, ethical, open, replicable.