ANOVA+Constrasts

ANOVA (1-Way?) Constrasts (ANOVA Analysis) (GWU EMSE-271)
Index | Topics (Logical Lectures) | Lectures | Problems | Readings | Nomenclature | Concepts

Major Points About Contrasts: Assumptions and Checks:
 * Constrats have a single degree of freedom
 * Must be orthogonal (make them independent) (required assumption)
 * "Constrast coefficient must be chosen before running the experiment and prior to examining the data." - EMSE 271, Fall 2009, Slide 304
 * Assuming that ANOVA 1-Way (Contrasts) is a special case of regression, then then normal regression assumptions apply
 * Error terms are normally distirbuted with a mean 0 and a variance sigma
 * Need to run same residual checks as for a simple regression analysis
 * As additionally ANOVA 1-Way (Contrasts), model assumes equal variances; variances are constant across error terms
 * Need to test for equal variances by plotting, using Barttlet's or Levene's Tests for Equal Variance

Sample Conclusions:
 * Slide 305


 * Sources:**
 * EMSE 271, Fall 2009