Below are the study questions for the final exam. They cover correlation, regression, multiple regression, MANOVA, and discriminant analysis. Of these questions, 50 will be randomly selected for use as your final examination. The final examination is Tuesday, May 12, 11 am- 1pm
1. The basic, zero-order association between two interval or better level variables is measured by
a. discriminant analysis
b. ANOVA
c. Pearson’s r
d. MANOVA
2. The correlation coefficient, r, ranges between
a. 0 and 1
b. -1 and 0
c. -1SD and +1SD
d. -1 and +1
3. What would be the value of Pearson’s r, the correlation coefficient, when scores on X have a perfect negative relationship with scores on Y?
1. -1
2. 1
3. 0
4. Impossible to tell from the information given
4. When there is no relationship between X and Y, the value of Pearson’s r is
1. -1
2. 1
3. 0
4. .5
5. The angle of the straight line in the picture below
in relation to the vertical axis is called
the
1. ordinate
2. intercept
3. regression
4. slope
6. The pattern in the chart below
represents a
1. positive correlation
2. negative correlation
3. part correlation
4. partial correlation
7. What sort of statistical test of this hypothesis would you perform?
“People who have high scores on the computer self-efficacy test will have low scores on the aging anxiety test”
1. correlation coefficient 2. Wilks lambda 3. eta squared 4. eigenvalue
8. The significance of Pearson’s r statistic is tested with
1. Chi square 2. Z 3. t 4. DF
9. From the output below, what can we conclude about the regression equation for the regression of Y on X?
1. the value of F is significant, so the regression equation is a poor predictor
2. the regression equation is a significantly better predictor of Y than the mean of Y
3. the mean of Y is a better predictor than the regression equation
4. the residual SS is larger than the regression SS, so the predictive power is poor
10. Which parts of the diagram below are “removed” when a partial correlation between Y and X2 is calculated?
1. a, b, d 2. b, c, d 3. a, b, c 4. d, b
11. For the relationship between X and Y you have obtained a correlation coefficient of .67. When you computed the partial correlation coefficient between X and Y controlling for the effects of Z you obtained a partial r of .55. What can you say about the relationship of X and Y when you have this information?
1. The relationship between X and Z is stronger than the relationship between X and Y
2. The relationship between X and Y is less strong when the effects of Z are held constant
3. Z is likely to be the cause of both X and Y
4. the partial correlation is an artifact of the original correlation between X, Y, Z and some unknown fourth variable
12. In the example below, what is the equation for the regression of Y on X?
1. Y = 4 + 2x
2. Y = X
3. Y = 2 + X
4. ZX = ZY
13. The amount of change in Y that you can expect to occur per unit change in Xi , where X is the ith variable in the predictive equation, when statistical control has been achieved for all of the other variables in the equation, is represented by
1. ai
2. standard error
3. βi
4. value of t
14. Which regression weights are used in the predictive equation when the variables are expressed in standard scores?
1. the constant and the b’s
2. the Beta coefficients
3. the unstandardized coefficients
4. the error terms
15. Which regression weights are used in the predictive equation when the variables are expressed in raw scores?
1. the constant and the b’s
2. the beta coefficients
3. the standardized coefficients
4. the error terms
16. What is this a definition of? the case in which two or more of the predictors are too highly correlated, leading to unstable partial regression coefficients
1. partial correlation
2. multivariate condition
3. homoscedascticity
4. multicollinearity
17. What does tolerance measure?
1. partial correlation
2. multivariate condition
3. homoscedascticity
4. multicollinearity
18. What is a bad value of tolerance?
1. 1
2. zero
3. .5
4. 10
19. What is a “good” value of tolerance?
1. 1
2. zero
3. .5
4. 10
20. A procedure for testing the hypothesis that one or more independent variables, or factors, have an effect on a set of two or more dependent variables, is
1. MANOVA
2. factorial ANOVA
3. correlation
4. VIF
21. Suppose you had to test the hypothesis that ethnicity influences a set of work-related variables including salary, years at current rank, supervisor ratings, and relationships with co-workers. What procedure would you use?
1. test of multicollinearity
2. tolerance test
3. factorial ANOVA
4. MANOVA
22. Reducing the experimentwise error rate is a good reason for doing a
1. test of multicollinearity
2. tolerance test
3. factorial ANOVA
4. MANOVA
23. When it appears that your dependent variables are more meaningful taken together than considered separately, this is a good argument for performing a
1. test of multicollinearity
2. tolerance test
3. factorial ANOVA
4. MANOVA
24. In MANOVA, one looks at the results of the individual F tests of the several dependent variables only if
1. the Sheffe tests are positive 2. the overall F test is significant
3. the Wilks’ lambda value is above .90 4. the Wilks’ lambda value is below .5
25. In MANOVA, a value of Wilks’ lambda of .90 would mean
1. only about 10% of the variance is explained by the factor or between-groups effect
2. 90% of the variance is explained by the factor or between-groups effect
3. 81% of the variance is explained by the factor or between-groups effect
4. Wilks’ is not an indicator of variance explained but rather of confidence level
26. Suppose you have set your confidence level for your hypothesis testing to .01.
Which of the effects in the following table are significant?
1. main effect for region
2. main effect for interaction of region and hscat4
3. main effect for hscat4 and region
4. main effect for hscat4
Effect / Value / F / Sig.Intercept / Pillai's Trace / .989 / 1071.410(a) / .000
Wilks' Lambda / .011 / 1071.410(a) / .000
Hotelling's Trace / 91.835 / 1071.410(a) / .000
Roy's Largest Root / 91.835 / 1071.410(a) / .000
REGION / Pillai's Trace / 1.048 / 6.624 / .000
Wilks' Lambda / .237 / 7.645 / .000
Hotelling's Trace / 2.133 / 7.980 / .000
Roy's Largest Root / 1.568 / 19.335(b) / .000
HSCAT4 / Pillai's Trace / .423 / 2.024 / .043
Wilks' Lambda / .628 / 2.001 / .049
Hotelling's Trace / .516 / 1.931 / .056
Roy's Largest Root / .311 / 3.834(b) / .017
REGION * HSCAT4 / Pillai's Trace / .550 / 1.384 / .154
Wilks' Lambda / .528 / 1.400 / .148
Hotelling's Trace / .753 / 1.409 / .144
Roy's Largest Root / .518 / 3.195(b) / .013
27. One difference between the standardized and the unstandardized discriminant function coefficients is that
1. the unstandardized functions can be used to classify new cases based on raw scores
2. the standardized functions can be used to classify new cases based on raw scores
3. the unstandardized functions can be used to classify new cases based on Z scores
4. none of the above
28. One difference between the standardized and the unstandardized discriminant function coefficients is that
1. the standardized weights on the function can be interpreted and compared as the relative contribution of the variables to discriminating groups
2. the unstandardized weights on the function can be interpreted and compared as the relative contribution of the variables to discriminating groups
3. the unstandardized weights on the function can be interpreted and compared as the relative contribution of the groups to discriminating variables
4. the standardized weights on the function can be interpreted and compared as the relative contribution of the groups to discriminating variables
29. Considering the table below, what is the main purpose of Function 1?
1. to compare the means on the canonical variable of low wealth and moderate wealth
2. to separate the low wealth from the high wealth group on the canonical variable
3. to show how it explains more variance than function 2
4. none of the above
30. In discriminant analysis, the best way to test the success of the discriminant function in classifying new cases is by
1. re-classifying cases from the holdout sample with a discriminant function obtained on a derivation sample
2. re-classifying cases from the derivation sample with a discriminant function obtained on a hold-out sample
3. re-classifying cases from the derivation sample with a discriminant function obtained from the derivation sample
4. using the leave-one-out option in SPSS
31. If we were doing post-hoc pairwise comparison significance tests on three dependent variables in a multivariate analysis, what might be the reasons we would set our confidence level very low, such as .05/3 or .017?
1. reduce multicollinearity
2. reduce tolerance
3. reduce probability of experimentwise alpha error
4. all of the above
32. The stepwise procedure is an option in
1. MANOVA 2. multiple regression 3. discriminant analysis 4. 1 and 2
5. 1 and 3 6. 2 and 3 7. 1, 2, and 3
33. The statistic which measures the association between X1 and Y when the effects of a third variable, X2, are removed from the relationship between X1 and X2 and their relationship with Y is called
1. multiple correlation
2. partial correlation
3. part correlation
4. zero-order correlation
34. The statistic which measures the association between X1 and Y when the effects of a third variable, X2, are removed from the relationship between X1 and X2, but not from their relationship with Y, is called
1. multiple correlation
2. partial correlation
3. part correlation
4. zero-order correlation
35. The statistic which measures the degree to which one variable is correlated with two or more other variables is the
1. multiple correlation
2. partial correlation
3. part correlation
4. zero-order correlation
36. The coefficient of determination, R-squared, tells us how much of the variation in y, the dependent variable, can be explained by variation in
1. the mean of y
2. the mean of x
3. x2
4. x
37. The formula for r is the ratio of (a) how much score deviation the two distributions (X and Y) have in common to
1. the sum of the crossproducts
2. the maximum amount of score deviation they could have in common
3. the between group variance
4. the product of the error term and the within term
38. The equation Y = a + bX + e is called a
1. correlation equation
2. normal equation
3. regression equation
4. curvilinear equation
39. In the equation Y = a + bX + e , a is also called
1. the beta weight 2. the constant 3. the error term 4. the independent variable
40. In the equation Y = a + bX + e , e is also called
1. the beta weight 2. the constant
3. the error term 4. the independent variable
41. When I have two nominal level independent variables and a single interval or better level dependent variable, I should use what technique to analyze my data?
1. factorial ANOVA
2. multiple correlation
3. MANOVA
4. canonical correlation
42. When I have two nominal level independent variables and a single interval or better level dependent variable, how many main effects will I be testing for in the overall F test?
1. one
2. two
3. three
4. four
43. When I have two nominal level independent variables and a single interval or better level dependent variable, how many interaction effects will I be testing for in the overall F test?
1. one
2. two
3. three
4. four
44. In this design, identify the factors and how many conditions each has: 2 X 2 ANOVA
1. four levels of one IV, no interaction of factors
2. two levels of one IV, no interaction of factors
3. two levels of two IVs, interaction of factors
4. four levels of two IVs, interaction of factors
45. Another term for Beta weight is
1. partial regression intercept coefficient
2. standardized normal partial intercept
3. unstandardized partial regression coefficient
4. standardized partial regression coefficient
46. Which regression weights are you able to interpret in terms of the relative contribution of the predictors to explaining variation in the dependent variable?
1. the constant and the b’s 2. the Beta coefficients
3. the unstandardized coefficients 4. the error terms
47. Which regression weights do you use to predict raw scores on Y for new cases?
1. the constant and the b’s 2. the Beta coefficients
3. the unstandardized coefficients 4. the error terms
48. The sum of squared deviations of differences between the known values of Y and the predicted values of Y based on the regression equation is called the
1. residual sum of squares
2. mean square
3. regression sum of squares
4. between-group sum of squares
49. Which of the following is not an example of a specification error?
1. curvilinear relationships among the predictors