Week 4. MANOVA.2024
Multivariate Statistical Analyses Research Methods II
Course: OCCT 6386
Institution: University of Texas Medical Branch
Department: Occupational Therapy
Instructor: Claudia L. Hilton, PhD, MBA, OTR, FAOTA
Understanding Regression
Types of Regression Questions
Linear Regression Example: Predicting outcomes based on a continuous variable.
Logistic Regression Example: Determining the probability of an event occurring (binary outcome).
Necessary Sample Size for Regression
Minimum number of subjects needed to ensure statistical power during regression analysis.
Entry Methods for Regression
Stepwise Regression: A method of selecting predictors based on statistical criteria.
Key Definitions
Residual: The difference between the observed value and the predicted value in regression analysis.
Standard Error of the Estimate: Represents the average distance that the observed values fall from the regression line.
Coefficients: In the linear regression equation, b represents the slope (rate of change for the dependent variable) and a represents the y-intercept.
Correlation Coefficient vs Coefficient of Determination
Correlation Coefficient (r): Measures the strength and direction of a linear relationship between two variables (range -1 to +1).
Coefficient of Determination (R²): Represents the proportion of variance explained by the independent variable(s) in the regression model.
Statistical Analysis Objectives
Describe various statistical analyses:
ANCOVA (Analysis of Covariance)
MANOVA (Multivariate Analysis of Variance)
MANCOVA (Multivariate Analysis of Covariance)
Principal Component Analysis
Factor Analysis
Cluster Analysis
Rasch Analysis
Comparing Statistical Methods
Correlation, ANOVA, and Regression
Correlation: Examines the relationship between two variables.
ANOVA: Used to identify differences between three or more groups.
Regression: Used for predicting values of a dependent variable based on independent variables.
Covariance
Definition: Occurs when two sets of scores vary together in similar patterns.
Covariance vs Correlation
Covariance results often hard to interpret; correlated variables are standardized to create a correlation coefficient valued between -1 and +1.
Examples of Covary and Correlate
Covary: Two variables that change together without implying causation (e.g., height and weight).
Correlate: Implies a statistical relationship with predictability.
ANOVA (Analysis of Variance)
Purpose of ANOVA
Compares means of three or more groups.
Example Applications
One-way ANOVA: Compares means based on one independent variable (e.g., group differences based on education level).
Two-way ANOVA: Examines interaction effects among two independent variables.
Types of Tests
Parametric and Non-Parametric Tests
Independent T-test: Compares means between two independent groups.
Mann Whitney U: Non-parametric test for two independent groups that does not assume normal distribution.
Paired T-test: Compares means in related groups.
Wilcoxon Signed-ranks Test: Non-parametric test for related samples.
H Tests: Kruskal-Wallis for comparing more than two groups non-parametrically, Friedman’s for repeated measures.
ANCOVA, MANOVA, MANCOVA
ANCOVA
Adjusts for covariates in one-way ANOVA models.
Useful in research where randomization is not possible.
MANOVA
Examines multiple dependent variables simultaneously.
Useful for understanding relationships between variables.
Examples of ANCOVA and MANOVA Applications
ANCOVA Example: Testing the effect of hours studied on test scores while controlling for prior academic performance.
MANCOVA Example: Investigating the impact of educational level on test scores adjusted for study habits.
Factor Analysis Concepts
Techniques and Analysis Types
Principal Component Analysis (PCA): Data reduction technique aiming to explain variance with fewer factors.
Exploratory Factor Analysis (EFA): Testing theories through variable relationships.
Factor Rotation Methods
Orthogonal Rotation: No correlation among factors (e.g., Varimax).
Oblique Rotation: Allows for correlations among factors.
Factor Loadings
Measures the strength of relationship between individual items and their factors. Higher values indicate better item-factor association.
Rasch Analysis
A statistical technique that transforms ordinal data for interval scaling. Useful in occupational therapy assessments.
Recent Studies and Applications in Occupational Therapy
Out-of-School Participation in Children with Autism
Study examining activity participation differences between typically developing children and those with HFASD.
Results highlighted significant differences in participation, indicating the need for targeted interventions.