Statistics & Research Design – Vocabulary Flashcards

A comprehensive set of vocabulary flashcards covering key terms and concepts from the Statistics & Research Design glossary.

Alpha (Level of Significance)

Probability of rejecting a true null hypothesis (Type I error); set by the researcher (commonly .05 or .01).

ANCOVA (Analysis of Covariance)

ANOVA that adjusts each participant’s DV score for an extraneous covariate to increase statistical efficiency.

Between-Groups Designs

Experimental designs that give each group a different level or combination of IVs to compare group means.

Central Limit Theorem

States that the sampling distribution of the mean becomes normal as sample size grows, has a mean equal to the population mean, and a standard error equal to σ/√N.

Chi-Square Test

Nonparametric test for frequency (nominal) data; single-sample version for one variable, multiple-sample version for two or more variables.

Cluster Analysis

Multivariate technique that groups cases into mutually exclusive, exhaustive clusters with high within-group similarity and high between-group difference.

Correlation Coefficient

Numeric index of relationship strength and direction (Pearson r for interval/ratio, Spearman rho for ranks, point biserial for true dichotomy + continuous, biserial for artificial dichotomy + continuous, eta for nonlinear continuous).

Cross-Validation / Shrinkage

Testing a correlation (e.g., validity coefficient) on a new sample; coefficient usually decreases (shrinks), especially with small original N and many predictors.

Discriminant Function Analysis

Uses two or more continuous predictors to classify cases into a single nominal criterion group.

Effect Size

Quantifies magnitude of IV–DV relationship (e.g., Cohen’s d in SD units, eta-squared as % variance explained).

Experimental Research (True vs. Quasi)

Empirical testing of IV–DV relationships; true experiments use random assignment, quasi-experiments do not and provide less control.

Experimentwise (Familywise) Error Rate

Overall probability of a Type I error across multiple statistical tests; increases with number of comparisons.

External Validity

Extent results generalize; threatened by pretest sensitization, reactivity, and multiple-treatment interference (controlled by counterbalancing).

Factorial ANOVA

ANOVA used when a study has two or more IVs and one interval/ratio DV; yields main and interaction effects (e.g., two-way ANOVA).

Factorial Design

Research design with ≥2 IVs allowing analysis of main effects of each IV and their interaction effects on the DV.

Independent & Dependent Variables

IV is manipulated to examine its effect; DV is the outcome measured. Each IV must have at least two levels.

Internal Validity

Confidence that IV, not extraneous factors, caused DV change; threatened by maturation, history, statistical regression, and selection biases.

Interval Recording

Behavior sampling that divides time into intervals and records if behavior occurs in each; suited to behaviors without clear start/stop.

Event Sampling

Behavior sampling that records every occurrence of a rare or permanent-product behavior during predefined events.

LISREL

Structural equation modeling program that tests causal models including latent variables, bidirectional paths, and measurement error.

MANOVA

Multivariate ANOVA for one or more IVs and two or more interval/ratio DVs; lowers experimentwise error and raises power.

Measures of Central Tendency

Mean (arithmetic average, interval/ratio data), median (middle score, ordinal or skewed data), mode (most frequent score, nominal data).

Mixed (Split-Plot) ANOVA

Factorial ANOVA combining at least one between-groups IV and one within-subjects IV.

Mixed Designs

Factorial designs containing both between-groups and within-subjects IVs.

Moderator Variable

Variable that changes the strength or direction of an IV–DV relationship (e.g., treatment more effective for men than women).

Mediator Variable

Variable that explains how/why an IV affects a DV (e.g., self-efficacy mediates parenting → achievement).

Multiple Regression

Predicts continuous criterion from multiple predictors; yields multiple R and regression equation; multicollinearity = high inter-predictor correlations.

Normal Curve

Symmetric bell distribution: ~68% within ±1 SD, 95% within ±2 SDs, 99% within ±3 SDs of the mean.

Null Hypothesis

Statement that IV has no effect on DV; tested against the alternative hypothesis which posits an effect.

One-Way ANOVA / F-Ratio

Parametric test comparing ≥2 group means on one IV; F = (treatment + error) / error; F > 1 suggests treatment effect.

Parametric Tests

Inferential tests for interval/ratio data with normality & homoscedasticity assumptions; more powerful (e.g., t-tests, ANOVA).

Nonparametric Tests

Tests for nominal/ordinal data or when parametric assumptions unmet (e.g., chi-square, Mann-Whitney U, Wilcoxon).

Path Analysis

Structural equation method that tests causal models using observed variables and path coefficients in a diagram.

Probability Sampling

Sampling where every population element has known selection chance (simple random, stratified, cluster).

Random Assignment

Randomly placing participants into treatment groups; cornerstone of true experiments to equalize extraneous factors.

Random Error

Unpredictable error due to sampling or measurement variability.

Randomized Block ANOVA

ANOVA analyzing main and interaction effects after blocking an extraneous variable treated as an IV.

Regression Analysis

Predicts criterion from predictor using regression line located by the least squares criterion to minimize prediction error.

Rejection Region

Area of sampling distribution where sample values are unlikely under H0; if obtained value falls here, reject H0 (size = alpha).

Sampling Distribution of the Mean

Distribution of all possible sample means; mean equals population mean and SD equals standard error (σ/√N).

Scales of Measurement

Nominal (categories), ordinal (rank order), interval (equal units, no true zero), ratio (true zero).

Shared Variability (r²)

Squared correlation showing proportion of variance two variables share (e.g., r=.50 → 25% shared).

Single-Subject Designs

AB, reversal (ABA/ABAB), or multiple-baseline designs using repeated DV measures to evaluate treatment for an individual.

Skewed Distributions

Asymmetrical distributions; positive skew → mean > median > mode; negative skew → mode > median > mean.

Standard Deviation

Square root of variance; expresses average dispersion of scores around the mean.

Statistical Power

Probability of rejecting a false H0; enhanced by large N, strong IV effects, larger alpha, and less error.

Student’s t-Test

Parametric test comparing two means: single-sample, independent samples, or correlated (paired) samples versions.

Systematic Error / Extraneous Variables

Predictable error from confounding variables that systematically affect IV–DV relationship.

Trend Analysis

ANOVA variant assessing linear or higher-order trends when IV is quantitative.

Type I Error

Rejecting a true null hypothesis; probability equals alpha.

Type II Error

Retaining a false null hypothesis; probability equals beta (usually unknown).

Within-Subjects Designs

Designs where each participant receives all IV levels at different times; comparisons made within participants (e.g., time-series).