Term
| Explain stepwise regression. |
|
Definition
| An old, nonsystematic, and flawed approach to figuring out a set of predictors to be added. We have no theory or interest in whether a certain set of variables predicts a criterion over and above another set (e.g., demographic variables)—we simply want a regression equation that predicts. Out of a group of predictors, one variable is selected at each stage that has the largest semi-partial correlation and therefore makes the largest contribution to R-squared. When no predictor makes a statistically significant contribution, predictors are no longer permitted. Thus a posteriori order is defined based solely on a statistical consideration (the statistical significance of semi-partial correlations), which makes it a variable search procedure. |
|
|
Term
| Explain hierarchical regression |
|
Definition
| Predictors are entered cumulatively according to some pre-specified order (based on a theoretical model) that is dictated in advance by the purpose and logic of the research. We’re systematically building up an argument. In a first equation we might have Age and Gender as predictors, but in a second equation we might have Age, Gender, AND Education and Training, and we are interested in seeing how the addition of Education and Training increases the proportion of variance accounted for in the criterion. |
|
|
Term
| Suppose you have a categorical variable with four groups, e.g. four regions of the country. You wish to use it as a predictor in a regression analysis. What is the general strategy for employing a categorical variable as a predictor in a regression analysis? |
|
Definition
| • Because geographical regions do not have an underlying order, we can’t use them as predictors. Instead, we would code each variable into a set of code predictors, such that each category (region) has a unique combination of codes that do not imply any order. There are numerous coding schemes, including = dummy codes, effect codes, and contrast codes. |
|
|
Term
| In any coding scheme for g groups, how many codes are required to characterize the g groups? |
|
Definition
| In any coding scheme for g groups, (g-1) codes are required, corresponding to the (g-1) degrees of freedom for any factor with g levels ANOVA |
|
|
Term
| For three groups, be able to create the following coding schemes for the data set: a dummy variable coding scheme with group 3 as the baseline group. |
|
Definition
If experimental (E), placebo (P), and untreated/control (U), would need D1 and D2 such that all participants in E would be D1 = 1, D2 = 0; all in P would be D1 = 0, D2 = 1; all in U would be D1 = 0, D2 = 0. (U is therefore the base group.) Y-hat = b1D1 + b2D2 + b0
In general, dummy coding assigns a 1 to one group and a 0 to all other groups for each code. One group is coded 0 on all dummy codes. Very useful for experimental groups vs. control group, because the control group is a clear base group in which we are theoretically interested. |
|
|
