Adv Correlat Methods Quiz 5 (Start Of Exam 2 Material)

Front 1

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

By how much does college GPA increase for every I-point increase in SAT scores?

Back 1

use unstandardized B

(because original unit of measure)

Front 2

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

Controlling for high school GPA, what proportion of the total variance in college GPA is explained by
SAT scores?

Back 2

use semipartial correlation coeficient squared

(because question is asking for the proportion of the total variance).

Front 3

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

Is high school GPA or SAT scores a stronger predictor of college GPA in the model?

Back 3

Standardized B, sr^(2), or pr^(2)

Front 4

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

What proportion of variance in college GPA is predicted by the optimal linear combination of high
school GPA and SAT scores?

Back 4

multiple R^(2)

Front 5

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

After removing variance due to high school GPA, what proportion of the remaining variance in college
GPA can be explained by SAT scores?

Back 5

pr^(2)

Front 6

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

By how much does college GPA increase for every 1 standard deviation increase in SAT scores?

Back 6

standardized b

Front 7

Multiple R^(2) is always __ and can range from __ to __.

Back 7

Multiple R^(2) is always positive and can range from 0 to 1.

Front 8

Multiple R^(2)reflects the relationship between...

Back 8

y and predicted y

Front 9

The most prestigious universities have the highest average faculty salaries, but professors who get offers
from two universities at once almost always find that the more prestigious university offers them the lower
salary. ("Surely, salary is not the only compensation. You'll have wonderful facilities and colleagues.")
Suppose prestigious universities find that this argument works and they can generally pay less that other
universities for a person with the same qualifications. What is the sign of the simple correlation between
prestige and salary? What is the sign of the partial correlation between prestige and salary, after removing variance associated with academic ranking?

Back 9

1. the sign of the simple correlation between
prestige and salary is positive
2.the sign of the partial correlation between prestige and salary is negative

Prestigious universities pay more, but within universities of the same rank, more prestigious universities pay less.

Front 10

represents the proportion of the remaining variance in y after we remove what is accounted for by the other predictor.

Back 10

pr^(2)

Front 11

represents the proportion of total variance in y that is uniquely associated with x1.

Back 11

sr^(2)

Front 12

If we remove the variace associated with the other predictor, this term represents how much you can account for out of the remaining variance in y.

Back 12

pr ^(2)

Front 13

Which is almost always higher, pr^(2) or sr^(2)?

Back 13

pr^(2)

Front 14

Interpret the following:
pr^(2) = .25

Back 14

After removing the variance associated with x2 (in this example), 25% of the remaining variance is associated with x1.

Front 15

We take out all of the variance associated with x2. If we then ask "of the remaining variance, what proportion of the variance is accounted for by x1, what is the term to which we are referring?

Back 15

pr^(2)

Front 16

What part of the variance is uniquely predicted by x1?

Back 16

semipartial correlation coeficient squared.

Front 17

This term refers to the proportion of variance accounted for by the regression equation.

Back 17

multiple R^(2)

Front 18

Say that multiple R^(2) = .67. What does this mean?

Back 18

67% of the variance in recovery scores can be accounted for by the optimal linear combination of optimisim and severity (ie, the two IVs).

Front 19

Describe how to interpret sr^(2).

Back 19

sr^(2) represents the increase in multiple R^(2) when this predictor is added to the regression equation.

Front 20

This represents the proportion of total variance in y that is uniquely accounted for by that predictor(it doesn't include overlapping variance).

Back 20

sr^(2)

Front 21

Describe how to interpret pr^(2).

Back 21

The proportion of variance in Y not associated with X2 that is associated with X1.

Asks: How much of Y that is not predicted by X2 does X1 predict?

Front 22

What is sr^(2)?

Back 22

The squared semipartial corelation coeficient.

This tells us the increase in the proportion of explained Y variance when X1 is added to the regression analysis.

Front 23

What is pr^(2)

Back 23

This is the squared partial correlation coeficient.

This tells us the proportion of variance in Y, not explained by other predictors, that is explained by the predictor in question.

Front 24

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

What is the correlation between Y and Y-hat?

Back 24

multiple R^(2)

Front 25

You have a large data set you are using to predict college GPA from high school GPA and SAT scores. You've done a multiple regression analysis. What coefficient would you choose to examine the following
questions? (b, beta, partial correlation, semipartial correlation, multiple r, any coefficient squared, etc.)

What is the increase in R^(2) when you add SAT scores to a model already containing high school GPA?

Back 25

sr^(2)

Front 26

When drawing the Venn diagrams associated with MLR, we are always looking at variance in _____?

Back 26

dependent variable

Front 27

This is the proportion of variance in the dv (y) that represents how much R^(2) will increase when we add a predictor

Back 27

sr^(2)

Front 28

I am under the impression that the fundamental difference between sr^(2) and pr^(2) is...

Back 28

With sr^(2), you CONTROL for the variance associated with one variable in order to predict the proportion of TOTAL VARIANCE that is accountec for by another variable.

With pr^(2), you REMOVE variance associated with one variable in order to determine what proportion of the REMAINING VARIANCE in one variable can explain another.

Front 29

When determining the coeficient to use, what are some useful things to remember?

Back 29

sr^(2) and pr^(2) ask for proportions of variance either of the total (sr^2) or of what is remaining after removal (pr^(2)).

The standardized/unstandardized Bs are referring to how much dv increase/decrease for every one unit increas/decrease in iv

Multiple R^(2) seems to ask about optimal linear combination of variables.

Front 30

When determining the coeficient to use, what are some useful things to remember?

Back 30

sr^(2) and pr^(2) ask for proportions of variance either of the total (sr^2) or of what is remaining after removal (pr^(2)).

The standardized/unstandardized Bs are referring to how much dv increase/decrease for every one unit increas/decrease in iv

Multiple R^(2) seems to ask about optimal linear combination of variables.

Front 31

When determining the coeficient to use, what are some useful things to remember?

Back 31

sr^(2) and pr^(2) ask for proportions of variance either of the total (sr^2) or of what is remaining after removal (pr^(2)).

The standardized/unstandardized Bs are referring to how much dv increase/decrease for every one unit increas/decrease in iv

Multiple R^(2) seems to ask about optimal linear combination of variables.

Front 32

When determining the coeficient to use, what are some useful things to remember?

Back 32

sr^(2) and pr^(2) ask for proportions of variance either of the total (sr^2) or of what is remaining after removal (pr^(2)).

The standardized/unstandardized Bs are referring to how much dv increase/decrease for every one unit increas/decrease in iv

Multiple R^(2) seems to ask about optimal linear combination of variables.

Front 33

For every 1 SD increase in optimisim scores, recovery scores increase by .559 SD, holding severity constant.

The sentance above describes an interpretation of what coeficient?

Back 33

standardized b

Front 34

For every 1 point increase in optimisim scores, recovery scores increase by 1.31 points, holding severity constant.

The sentance above describes an interpretation of what coeficient?

Back 34

unstandardized b.

Front 35

If, in MLR, you are asked to compute the standardized/unstandardized regression weights using correlation formulas and interpret them, what general form will this take?

Back 35

*For MLR, you will be given two X values and one Y value.

You will need to create 4 interpretations:

Standardized:
For every 1 SD increase in x1, y scores increase by .559 SD, holding x2 constant

Unstandardized:
For every 1 point increase in X1, Y scores increase by 1.31 points, holding X2 constant.

Standardized:
For every 1 SD increase in x2, y scores increase by .559 SD, holding x1 constant.

Unstandardized:
For every 1 point increase in X2, Y scores increase by 1.31 points, holding X1 constant.

Front 36

Interpret the following:
Multiple R^(2) = .67.

Back 36

67% of the variance in recovery scores can be accounted for by our optimial linear combination of x1 and x2.

Front 37

What is the Venn diagram formula for multiple R^(2)?

Back 37

(a+b+c)/ (a+b+c+e)

Front 38

What is the Venn diagram formula for sr^(2)?

Back 38

(a)/(a+b+c+e)

Front 39

Represents the part of the variance that is uniquely predicted by one of the Xes on the Venn diagram.

Back 39

a. Also note that (a/(a+b+c+e))refers to the semipartial correlation squared.

Front 40

What is the Venn diagram formula for pr^(2)?

Back 40

with pr^2, you are taking out all of the variance associated with one of the Xes and looking at what remains.

I think that the Venn diagram formula is represented like this:
pr^(2) = a/(a+e)

Front 41

this coeficient is dependent upon the scales of the variables.

Back 41

unstandardized partial regression coeficient b

Front 42

this coeficient represents the average change in y for every unit increase in x.

Back 42

unstandardized partial regression coeficient b

Front 43

this tells us the average change in the z score for y for every 1 sd increase in x

Back 43

standardized partial regression coeficient b*

Front 44

This allows us to interpret the relative magnitudes of the b* weights in terms of relative importance of the predictors (thus, we can compare across predictors)

Back 44

standardized partial regression coeficient b*

Front 45

This coeficient answers the following question:
How much does this predictor add to the proportion of variance in y (which ranges from 0 to 1)?

Back 45

squared semipartial correlation coeficient sr^(2)

Front 46

this tells us the increase in the proportion of explained y variance when X1 is added to the regression analysis.

Back 46

squared semipartial correlation coeficient sr^(2)

Front 47

This tells us the proportion of variance in y,not explained by other predictors, that is explained by the predictor in question.

Back 47

squared partial correlation coeficient pr^(2)

Front 48

Represents the proportion of variance in y NOT associated with x2 that is associated with x1.

Back 48

pr^(2)

Front 49

Increase in multiple R squared when this predictor is added to the regression equation.

Back 49

sr^(2)

Front 50

X1 is a measure of stress, x2 is a measure of negative life events, y is a measure of depressive symptoms. Ry1=0.50 and ry2=0.70.

What is the proportion of variance in y that is accounted for by each variable alone?

Back 50

ry1^(2) = 0.25
ry2^(2) = 0.49

Front 51

X1 is a measure of stress, x2 is a measure of negative life events, y is a measure of depressive symptoms. Ry1=0.50 and ry2=0.70.

1. If X1 and X2 are uncorrelated with each other, what is the proportion of variance in y that would be accounted for if both variales were included as predictors in a multiple regression analysis?
2. Would this proprtion be the same if x1 and x2 were correlated with each other? Why or why not.

Back 51

1. If they are uncorrelated, .25+.49 = .74. 74% of the variance in y would be accounted for if both variables were included as predictors in mlr.

2. This proportion would not be the same if x1 and x2 were correlated with each other, be/c with correlated variables there would be overlapping variance and the overal proportion of variance accounted for would be smaller.

Front 52

X1 is a measure of stress, x2 is a measure of negative life events, y is a measure of depressive symptoms. Ry1=0.50 and ry2=0.70.

Imagine that you condicted a MLR analysis with x1 and x2 predicting y. You found that the unstandardixed b for x1 was 0.82. Interpret b in relation to the variables in the study.

Back 52

For every 1 pt increase in stress scores, scores on the measure of depressive symptoms will increase by 0.82 points, holding scores on the measure of negative life events constant.

Front 53

X1 is a measure of stress, x2 is a measure of negative life events, y is a measure of depressive symptoms. Ry1=0.50 and ry2=0.70.

Imagine that you condicted a MLR analysis with x1 and x2 predicting y. You found that the squared semi-partial correlation coeficient for x2 was 0.22. Interpret this value in relation to the variables.

Back 53

Controlling for stress scores, the proportion of the total variance in depressive symptoms that can be explained by scores on a measure of negative life events is 0.22.

Front 54

X1 is a measure of stress, x2 is a measure of negative life events, y is a measure of depressive symptoms. Ry1=0.50 and ry2=0.70.

Imagine that you condicted a MLR analysis with x1 and x2 predicting y. You found that the R^(2) value for the MLR was 0.43. Interpret R^(2) in relation to the variables in the study.

Back 54

This means that 43% of the variance in scores on the measure of depressive symptoms can be explained by the optimal linear combination of stress and negative life event scores.