click below
click below
Normal Size Small Size show me how
Multiple Regression
| Question | Answer |
|---|---|
| Predicting a DV with 2 or more variables Major Benefits | you can have as many IV's predicting your DV as you want. Helps predict DV with great accuracy. You can see if 2 or more IV are redundant in predicting DV. Allow you to predict your DV with great accurracy and with fewest numbers of IV's |
| Strength of Regression How can you tell if you have a "strong" predictor? | Calculate R'2; shows how variance is explained by each predictor. Calculation: R*R=R'2 Interpretation if SAT scores R=.5(R'2= .25) SAT scores explain 25% of college success (.5*.5=.25) |
| Multicollinearity | When the variance that is explained in the DV is overlapped by 2 or more predictors. The IV that wins ends up being a significant predictor. It explains variance that the other IV cannot |
| Hierarchical Regression | Running IV one at a time or in set. BENEFITS: Allows for testing more complex theories. Allows for even better exploratory analysis Let us test complex hypothesis |
| Testing Mediation | Run the IV (number of siblings) as a single predictor of the DV (college GPA) |
| Exploratory Analyses | Stepwise regression SPSS decides how to enter or take away IV's in hierarchical analyses |
| Forward Stepwise Regression | SPSS tests all IV individually and determines which one has the largest R'2 SPSS stops when it no longer finds significant IV |
| Backward stepwise regression | SPSS tests all IV in multiple regression SPSS then repeats the multiple regression analysis wirh the remaining IV SPSS stops when all remaining IV are significant |
| Forward Stepwise Benefits | SPSS gives you a model where only the most powerful IV are selected (survival of the fittest). Good for finding out which IV really matter. |
| Backward stepwise Regression | Gives you a model where there is the greatest number of possible predictors. SPSS Gives you the most predictive model possible. |
| Regression | Is often measured in exploratory studies. Measures your DV along with anything you think may be related to your DV. Example childhood obesity possible predictors diet exercise. |
| Dangers of step wise regression | Should only be used for exploratory analyses. Stepwise regression is guided by statistics not theory so chance of type one error is high. they are not normally PUBLISHABLE Results |
| Multiple regression | Let us predict one DV with multole IV |