Week 7 Flashcards
What is parsimony?
The amount to which predictors explain unique variance in the criterion.
Is high parsimony good?
Yes, it means the predictors individually account for variance, whereas low parsimony means the predictors have a lot of shared variance, suggesting they were poorly chosen / designed as they’re measuring the same thing.
What is model R^2?
The total amount of variance accounted for by all the predictors in a regression with multiple predictors.
What is model r^2?
The total amount of variance accounted for by a single predictor in a bivariate regression.
What is semi partial correlation (sr^2)?
The amount of variance explained by a predictor individually (sr^2 for predictor 1, sr^2 for predictor 2, etc.)
What is a semi partial correlation (sr^2) recorded as on SPSS?
Part correlation, although this requires squaring to become the semi-partial correlation.
What is partial correlation?
The amount of variance accounted for by an indiv
What is partial correlation (pr^2)?
The amount of variance accounted for by a predictor individually while removing variance that is shared accounted for by another variable.
Does semi-partial correlation include the total variance in its calculation?
Yes, it measures the proportion of variance accounted for individually by a predictor in comparison to the total amount of variance, whereas partial correlation removes variance accounted for by other variables.
What does a Fisher’s Z-test do?
Tests for significant differences in the strength of the individual contribution of predictors ie does one account for significantly more than another?
In a hierarchical multiple regression (HMR), what key statistic is measured at each step or block?
R^2change!
What does R^2change measure?
The variance accounted for by the predictors added in that step or block, ignoring the contributions of prior or latter predictors/blocks. ie R^2change at Block 2
How do you find the total Model R^2 in a HMR?
You add all of the R^2change’s together.
Can a HMR identify interactions between predictors?
No. This can only be done in regression using a moderated multiple regression (MMR).
How does MMR work?
They add an extra predictor for the interaction, which is created by multiplying each participants score for predictor 1 by their score for predictor 2.
What is a plane of best fit?
A 2D representation of a 3D relationship between 3 predictors.
How do we know when a plane of best fit represents the data well?
If a regression model adequately explains the data, the plane of best fit will represent the scores (dots) on the graph well
What are validities?
The relationship between each predictor and the criterion.
What are collinearity?
Inter-correlations between predictors.
Explain the linear composite.
The linear composite is basically the end result of a regression. Any given score in a regression represents the combination (composite) of each predictor, weighted by their influence on the DV.
What is the regression coefficient?
Beta. It represents the amount y changes for a unit increase in x.
What is a coefficient?
A numerical or constant quantity placed before and multiplying the variable in an algebraic expression. In regression, our regression coefficient is the amount y changes for a unit increase in x.
How is a linear composite denoted?
Y^hat
What is the difference between predicted and observed scores?
Error / residual.
How is the overall regression model calculated / conceptualised?
A bivariate correlation between the the DV and the best linear explanation of the predictors (Y^) - the linear composite.
