Study Notes on Differences Between Two Groups
Differences Between Two Groups
Overview
Compare two conditions/groups using inferential statistics.
Conditions can involve:
Same group tested in different conditions (A and B)
Different groups in conditions A or B
Tests Used
Parametric Tests:
Independent groups t-test
Paired groups t-test
Non-parametric Equivalents:
Mann–Whitney U test (independent groups)
Wilcoxon signed rank test (repeated measures)
Other metrics include confidence intervals and effect sizes.
Parametric T-tests
Assumptions: normal distribution and similar variances.
Violations may not significantly affect results.
Adjustments available for t-test formulas when variances differ.
Bootstrapping method for skewed data.
Independent Samples T-test
Determines differences between group means.
Hypotheses:
H0: No difference in means
H1: Difference exists
Levene’s test checks variances:
Use results based on variance equality.
Interpret p-values to accept or reject H0.
Dependent Samples T-test (Paired)
Compares scores from the same participants across conditions.
Use when scores are correlated for higher sensitivity.
Evaluate with similar hypotheses as independent t-test.
Effect Sizes
Cohen’s d quantifies the difference magnitude; facilitates comparison across studies.
Z-score guidelines for effect strength:
0.2: weak
0.5: moderate
0.8: strong
1.5: very strong
Non-parametric Tests
Used with ranked data; resistant to outliers.
Mann–Whitney U test for independent groups; Wilcoxon for related samples.
Interpreting Results
Statistical significance determines if group means differ.
Confidence intervals assess the potential overlap of means.
Non-significant results imply no meaningful differences between conditions or groups.
Summary
T-tests are typical for comparing groups.
Descriptive stats: means and standard deviations.
Confidence intervals aid generalization to populations.
Non-parametric tests serve as alternatives for skewed data.