Regression vs. ANOVA
    When to use (not experimental vs. correlational):
    Regression
        •   More than three predictors
        •   Have continuous predictors
        •   Need an SEE for prediction
        •   Want to compare standardized coefficients
        •   Want to build models easier (more control)
                               HDFS 7060 - Bubb, R.
              Regression vs. ANOVA
    When to use:
    ANOVA
        •   Have only categorical predictors (don’t restrict the range)
        •   Predictor has multiple levels (post-hoc test options)
        •   Are measuring observations multiple times (RM designs)
        •   Need a measure for equal variance assumption
                                 HDFS 7060 - Bubb, R.
             Regression vs. ANOVA
    Limitations:
    Regression
        •   No test for equal variances
        •   Post hoc tests are more difficult (require dummy coding)
        •   Post hoc comparisons are not alpha controlled 
        •   Easy to misinterpret main effects in moderation
                                HDFS 7060 - Bubb, R.
             Regression vs. ANOVA
    Limitations:
    ANOVA
        •   Doesn’t support continuous predictors
        •   No measure of SEE for individuals
        •   Limited to three predictors
        •   Fewer diagnostic options in SPSS
        •   Interaction post-hocs can be more difficult to calculate
                               HDFS 7060 - Bubb, R.
           Regression vs. ANOVA
    When to use:
    Quality studies will use both
       •  Use multiple tests—weigh the merits and limitations of each
       •  Leads to testing of multiple models
                          HDFS 7060 - Bubb, R.