Chapter 15 Understanding Regression Analysis Basics

1) Which of the following is NOT true of prediction?

A) It is a statement of what is believed will happen in the future.

B) It may be based on prior observation.

C) We are seldom confronted with the need to make predictions.

D) It may be based on past experience.

E) Marketing managers are constantly faced with the need to make predictions.

Answer: C

2) One of the best tools for unraveling prediction complexity is:

A) probability sampling.

B) association analysis.

C) differences testing.

D) linear regression.

E) causal research.

Answer: D

3) What is the best way to make a prediction?

A) using simple statistical analysis

B) making a best guess based on past experience

C) employing residual analysis

D) building a predictive model

E) hypothesizing

Answer: D

4) A predictive model is an approach to prediction that:

A) relates the conditions expected to be in place influencing the factor that is being predicted.

B) observes a consistent pattern over time.

C) identifies a pattern and projects it into the future.

D) uses past experience to predict the future.

E) uses current experience to explain the past.

Answer: A

5) ________ helps the researcher to understand whether observed data is truly linear and

whether the data is a good fit to the model being used.

A) Analysis of prediction

B) Control data

C) Analysis of residuals

D) Analysis of variance

E) Values comparison

Answer: C

6) Which of the following residuals shows an exact prediction?

A) 0

B) +1.0

C) -25

D) +25

E) 100.0

Answer: A

7) Bivariate regression analysis is defined as a predictive analysis technique in which:

A) a pattern is identified over time and projected into the future.

B) a relationship that exists across time is observed to make a prediction.

C) one variable is used to predict the level of another by use of the straight-line formula.

D) one variable is used to predict the level of another by use of a scatter diagram.

E) a relationship that exists at one point in time is observed to make a prediction.

Answer: C

8) ________ is a simple technique for analyzing two variables to predict behavior or activity

in the marketplace.

A) Regression analysis

B) Variance analysis

C) Bivariate regression

D) Multiple regression

E) Stepwise regression

Answer: C

9) In the formula for a straight line, the intercept is known as:

A) the dependent variable.

B) the variable used to predict the dependent variable.

C) the change in y for any unit change in x.

D) the point where the line cuts the y axis when x = 0.

E) b.

Answer: D

10) In the formula for a straight line, the slope is defined as:

A) the change in y for any 1-unit change in x.

B) where the line cuts the y axis when x = 0.

C) the variable used to predict the dependent variable.

D) the dependent variable.

E) the predicted variable.

Answer: A

11) In bivariate regression analysis, the dependent variable is one that is:

A) used to predict the independent variable, and it is the x in the regression formula.

B) used to predict the independent variable, and it is the y in the regression formula.

C) predicted, and it is usually termed x in the regression formula.

D) predicted, and it is usually termed y in the regression formula.

E) predicted, and it is termed b in the regression formula.

Answer: D

12) In bivariate regression analysis, the independent variable is one that is:

A) used to predict the dependent variable, and it is the x in the regression formula.

B) used to predict the dependent variable, and it is the y in the regression formula.

C) predicted, and it is the x in the regression formula.

D) predicted, and it is the y in the regression formula.

E) used to predict the dependent variable, and it is the b in the regression formula.

Answer: A

13) What criterion is used to establish the best "fit" of a straight line through the points on a

scatter diagram?

A) the plum line criterion

B) the least squares criterion

C) the bearing line criterion

D) the b slope criterion

E) the right angle criterion

Answer: B

14) In evaluating your bivariate regression analysis findings, you first determine whether or

not a linear relationship between the independent and dependent variable exists in the

population. Which of the following best describes what you are doing in this step?

A) determining if the two variables have any covariation

B) determining if the two variables vary together

C) determining if the two variables belong in the same regression matrix

D) determining if the two variables are isotonic

E) determining if there is statistical significance

Answer: E

15) In evaluating your bivariate regression analysis findings, you first determine whether or

not a linear relationship between the independent and dependent variable exists in the

population and secondly you:

A) determine the significance of the intercept and the slope.

B) determine the significance of the covariation.

C) determine if the two variables vary together.

D) determine if the two variables belong in the same regression matrix.

E) determine if the two variables predict the intercept and the slope.

Answer: A

16) Which of the following SPSS commands allows you to run bivariate regression?

A) ANALYZE; BIVARIATE; REGRESSION

B) ANALYZE; REGRESSION; BIVARIATE

C) REGRESSION; BIVARIATE

D) ANALYZE; REGRESSION; LINEAR

E) REGRESSION; BIVARIATE; LINEAR

Answer: D

17) In bivariate regression analysis, the higher the Adjusted R Square value:

A) the lower the predictive power of the analysis.

B) the better the straight line's fit to the scatter points.

C) the worse the straight line's fit to the scatter points.

D) the closer to 0 it will be.

E) None of the above; there is no Adjusted R Square value in regression analysis.

Answer: B

18) The main purpose of ANOVA in bivariate regression is to:

A) tell us if there are significant differences between three or more means.

B) tell us if ANOVA is an issue.

C) tell us if the straight-line model fits the data we are analyzing.

D) provide a frequency table for further analysis.

E) None of the above; ANOVA is not used in regression.

Answer: C

19) In bivariate regression, if the F value is significant (say .05 or less), then:

A) we accept the null hypothesis that a straight-line model fits our data.

B) we reject the null hypothesis that a straight-line model does not fit our data.

C) we abandon our efforts to analyze the two variables.

D) we check for outliers.

E) we rerun the regression.

Answer: B

20) In bivariate regression, if the F value is not significant (say .051), then:

A) we accept the null hypothesis that a straight-line model fits our data.

B) we reject the null hypothesis that a straight-line model does not fit our data.

C) we abandon our efforts to analyze the two variables.

D) we check for outliers.

E) we rerun the regression.

Answer: C

21) Sometimes a researcher will find that the ANOVA F is not significant in regression

analysis or if the F is significant, the R square is lower than desired. It is appropriate in these

cases to:

A) examine the data using another stat package other than SPSS.

B) change the scaling assumptions from ratio or interval to ordinal and rerun the analysis.

C) run a confidence interval around the predicted values and then make the interval narrower.

D) run a confidence interval around the predicted values and then make the interval wider.

E) run a scatter diagram, search for outliers, and remove them and rerun the regression.

Answer: E

22) A form of regression analysis where more than one independent variable is used in the

regression equation is known as:

A) regression planes.

B) additivity.

C) multiple regression analysis.

D) independence assumption.

E) MANOVA.

Answer: C

23) If Maxwell House Coffee was considering a line of gourmet iced coffee, it would want to

know how coffee drinkers feel about gourmet iced coffee; that is, their attitudes toward

buying, preparing, and drinking it would be the dependent variables. Maxwell House might

consider developing:

A) a general conceptual model.

B) a general conceptual model that identifies the independent and dependent variables.

C) a specific conceptual model that specifies the variables that will produce residuals

analysis.

D) a specific conceptual model that will require additional modification to be used in

residuals analysis.

E) a conceptual model that identifies the residuals that are associated with the dependent, or

slope, variable.

Answer: B

24) A graph of the dependent variable in multiple regression analysis is referred to as:

A) confidence intervals.

B) multiple regression.

C) multiple scatter plots.

D) regression plane.

E) a multi-scatter plot.

Answer: D

25) A multiple regression equation is best described by which of the following forms?

A) The independent variable is predicted by the intercept plus a series of values of the slope

times each dependent variable.

B) The independent variable is predicted by the slope plus a series of values of the intercept

times each dependent variable.

C) The dependent variable is predicted by the intercept plus a series of values of the slope

times each independent variable.

D) The dependent variable to be predicted is equal to the intercept plus a series of values of

the slope times each independent variable.

E) y = a + bx

Answer: D

26) Which of the following in multiple regression is a handy measure of the strength of the

overall relationship?

A) Adjusted R

B) Multiple R

C) multicollinearity

D) VIF

E) Adjusted B

Answer: B

27) Which of the following stipulates that independent multiple regression variables must be

statistically independent and uncorrelated with one another?

A) independence assumption

B) multicollinearity

C) additivity requirement

D) regression plane

E) uncorrelation

Answer: A

28) In multiple regression, the presence of correlations among the independent variables is

termed:

A) independence assumption.

B) multicollinearity.

C) additivity.

D) regression plane.

E) multicorrelation.

Answer: B

29) Which statistic is used to determine whether or not multicollinearity is a concern in

multiple regression?

A) coefficient of determination

B) multicol Z

C) multicol R

D) VIF (variance inflation factor)

E) Q

Answer: D

30) When the statistic used to determine whether or not multicollinearity is a concern in

multiple regression is greater than ________, it is prudent to remove that variable and rerun

the regression.

A) .05

B) .10

C) .95

D) 1.00

E) 10

Answer: E

31) What is the proper SPSS command sequence to run multiple regression analysis?

A) ANALYZE; REGRESSION; MULTIPLE; GO

B) ANALYZE; REGRESSION; MULTIPLE

C) ANALYZE; REGRESSION; LINEAR

D) ANALYZE; REGRESSION; MLINEAR

E) ANALYZE; REGRESSION; MR

Answer: C

32) In multiple regression, you must test for the significance of the betas for each of the

independent variables. You would do this by looking for:

A) a significant t test for each independent variable.

B) a significant ANOVA for each independent variable.

C) a significant alpha level for each independent variable.

D) a significant nonlinear beta weight for each independent variable.

E) a significant R for each independent variable.

Answer: A

33) When you find "mixed" results in multiple regression (i.e., some betas are significant,

others are not), you:

A) eliminate, or "trim," the insignificant variables.

B) adjust the insignificant variables by applying a standardized weight.

C) accept the null hypothesis.

D) choose the result that fits your hypothesis.

E) none of the above

Answer: A

34) When we make a prediction using multiple regression, we can apply a 95 percent

confidence interval around the predicted dependent variable by multiplying:

A) 1.96 times the standard error of the predictor.

B) 1.96 times the standard error of the estimate.

C) 2.58 times the standard error of the predictor.

D) 2.58 times the standard error of the estimate.

E) 1.96 times .95.

Answer: B

35) While the scaling assumptions of multiple regression require that both the independent

and dependent variables be at least interval-scaled, we may use nominal independent

variables by using:

A) ratio-scaled variables.

B) standardized beta coefficients.

C) dummy variables.

D) temporary variables.

E) semi-ratio variables.

Answer: C

36) A standardized beta coefficient is defined as:

A) the result of adding the difference between each independent variable value and its mean

and the standard deviation of that independent variable.

B) the result of multiplying the difference between each independent variable value and its

mean by the standard deviation of that independent variable.

C) the result of dividing the standard deviation of an independent variable by the difference

between that independent variable value and its mean.

D) the result of dividing the difference between each independent variable value and its mean

by the standard deviation of that independent variable.

E) the result of subtracting the difference between each independent variable value and its

mean by the standard deviation of that independent variable.

Answer: D

37) Independent variables are normally measured in different units, so to determine the

relative importance of the beta weights between independent variables we would use:

A) a screening variable.

B) a trimmed model.

C) standardized beta coefficients.

D) betas measured in "like-units."

E) weighted beta coefficients.

Answer: C

38) Researchers applied multiple regression analysis to study mobile phone service in

Thailand, using overall satisfaction as the dependent variable. Standardized betas for

independent variables were calculated. Of those listed below, which is the most important

factor for a Thai mobile phone company trying to increase its competitiveness?

A) quality of service, standardized beta = .139

B) promotions by the company, standardized beta = .158

C) innovativeness by the company, standardized beta = .060

D) social status of the company brand, standardized beta = -.013

E) customer service quality, standardized beta = .155

Answer: B

39) Which form of regression is useful when the researcher has many independent variables

and wants to narrow the set down to a smaller number?

A) multiple component reduction

B) stepwise multiple regression

C) variance deflation regression

D) variance inflation regression

E) narrow regression

Answer: B

40) Which sequence of SPSS commands would you select in order to run stepwise multiple

regression?

A) ANALYZE; REGRESSION; LINEARSTEPS

B) ANALYZE; REGRESSION; LINEAR; METHOD; STEPWISE

C) STEPWISE; LINEAR REGRESSION; GO

D) STEPWISE; LINEAR REGRESSION

E) ANALYZE; REGRESSION; METHOD; STEP

Answer: B

41) Which of the following are warnings that the textbook authors give regarding regression

analysis?

A) It is complicated and requires large computer memory.

B) It does not give you cause-and-effect statements, and it is expensive to run.

C) It does not give you cause-and-effect statements, and the text's coverage of regression

analysis only scratches the surface of this topic.

D) It is expensive, and you should not apply regression to predict data outside the boundaries

of the data used to develop the regression model.

E) No warnings are given.

Answer: C

42) A prediction is a statement of what is believed will happen in the future made on the basis

of past experience or prior observation.

Answer: True

43) The two ways of making a prediction are extension analysis and baseline predictive

modeling.

Answer: False

44) A predictive model simply examines what has happened in the past and predicts the

future.

Answer: False

45) When we make predictions and compare the differences between our predictions and the

actual results, we are performing what is known as analysis of residuals.

Answer: True

46) When we want to use one variable to predict another and use the equation: y = a + bx, we

use the technique known as multiple regression.

Answer: False

47) In the following straight-line formula, y = a + bx, the variable being predicted is the beta

weight, b.

Answer: False

48) In the formula for bivariate regression analysis, the change in y for each one-unit change

in x is known as the slope.

Answer: True

49) In the formula for bivariate regression analysis, the point where the line cuts the y axis

when x = 0 is known as b, the beta.

Answer: False

50) In regression, the variable being predicted, y, is known as the dependent variable.

Answer: True

51) In regression, the variable being predicted, b, is known as the dependent variable.

Answer: False

52) In regression, the variable used to predict the dependent variable is known as x, the

independent variable.

Answer: True

53) In regression, the line that runs through the points on a scatter diagram is positioned to

minimize the vertical distances away from the line of the various points because of the least

squares criterion.

Answer: True

54) A regression line using the least squares criterion will result in high residuals.

Answer: False

55) Immediately in bivariate analysis, the researcher must find out whether or not a linear

relationship exists in the population.

Answer: True

56) The R Square value is very important because it tells us how well our regression line fits

the scatter of data points. It may range from 0 to +1.00 because it is the square of the

correlation coefficient, which may range from -1.00 to +1.00.

Answer: True

57) If the ANOVA F test is not significant in bivariate regression analysis, we must trim the

model by eliminating the insignificant dependent variable(s).

Answer: False

58) In bivariate regression analysis, t tests are used to test the significance of the slope and

the intercept of the multiple dependent variables.

Answer: False

59) If the tests of the significance of the slope and the intercept are significant, this means

that the straight-line relationship depicted by the slope and the intercept actually exists in the

population and, therefore, the regression equation may be used as a prediction device.

Answer: True

60) In bivariate regression, we can calculate an upper and lower range within which we could

expect the values of the independent values to fall if they were calculated.

Answer: False

61) We can sometimes improve a regression analysis finding by removing outliers and

rerunning the regression analysis.

Answer: True

62) An outlier refers to Multiple Rs that are above expected norms such as above 95 or 100.

Answer: False

63) In multiple regression analysis, we are trying to predict an independent variable using

more than two dependent variables.

Answer: False

64) Multiple regression requires specification of a general conceptual model that identifies

independent and dependent variables and shows their expected relationships.

Answer: True

65) A regression plane is the shape of the independent variable in multiple regression

analysis.

Answer: False

66) The multiple R, also called the coefficient of determination, in multiple regression ranges

from 0 to +1.00 and represents the amount of the dependent variable "explained" by the

combined independent variables.

Answer: True

67) Multicollinearity refers to correlations among the dependent variables and makes

predictions much more accurate because predicting one variable also allows you to predict

the correlated variable(s).

Answer: False

68) The VIF is useful for identifying multicollinearity.

Answer: True

69) VIF is an acronym for "Very InFrequent."

Answer: False

70) The SPSS command for running multiple regression is: ANALYZE; REGRESSION;

LINEAR.

Answer: True

71) In multiple regression analysis, t tests are used to test for the statistical significance of

betas. If a beta is insignificant, it means that its respective independent variable plays no

meaningful role in predicting the dependent variable, and the independent variable should be

"trimmed" from the model.

Answer: True

72) When you have independent variables that are not significant in multiple regression

analysis, it is appropriate to take them out and rerun the regression. The new model is

referred to as a "trimmed" model.

Answer: True

73) In multiple regression we make a prediction, but we cannot put confidence intervals

around our prediction as we can in bivariate regression.

Answer: False

74) In dummy coding, the 0-versus-1 code is traditional, but any two adjacent numbers could

be used, such as 1 versus 2.

Answer: False

75) We must use standardized beta weights to compare the size of beta weights in multiple

regression because the independent variables they represent are often measured with different

units.

Answer: True

76) Multiple regression may be used as a screening device in the sense that it may be used to

reduce large numbers of potential independent variables in order to spot those that are most

salient for the dependent variable.

Answer: True

77) There is a type of multiple regression, called stepwise multiple regression, that does the

trimming operation automatically.

Answer: True

78) If we wanted to use a type of regression that first enters the variable that explains the

most variance, then the variable that explains the second highest level of variance and so on,

we would use ordinal regression.

Answer: False

79) Stepwise multiple regression is useful if a researcher has many dependent variables but

needs additional dependent variables in order to obtain a good predictive model.

Answer: False

80) Once we establish, through multiple regression analysis, that certain independent

variables are statistically significant in predicting a dependent variable, we may assume this

relationship to be one of cause and effect.

Answer: False

81) When a researcher uses gender as a dummy variable in a study for a client, it is important

not to distort the findings by highlighting this fact in the final report and presentation.

Answer: False

82) When regression is used as a screening device, the items to report are (1) dependent

variable, (2) statistically significant independent variables, (3) signs of beta coefficients, and

(4) standardized beta coefficients for the significant variables.

Answer: True

83) A weather forecaster studies relationships among phenomena such as wind direction,

barometric pressure, humidity, jet stream flow, and temperature. Based upon his or her

knowledge of the relationships between these variables and the weather, the forecaster

predicts there is an 80 percent chance of rain tomorrow. Which method of prediction have

you used?

A) guessing

B) statistics

C) extrapolation

D) predictive model

E) forecasting

Answer: D

84) If the intercept is found to be 2 and the slope is found to be 5 in a regression result

formula, then:

A) y = 2 + 3x.

B) y = 2 + 5x.

C) x = 2 + 5y.

D) x = 2 + 3y.

E) y = a + 2(5).

Answer: B

85) In a straight-line formula, the intercept is 4, the slope is 2, and the independent variable is

6. The predicted variable's level is:

A) 12.

B) 14.

C) 16.

D) 18.

E) 26.

Answer: C

86) You run bivariate regression analysis and you find that the ANOVA results indicate that

you have Sig. value for your F of .05. Now, looking under your Coefficients output, you have

an intercept value and a slope value. You should use these values only when:

A) they are both significant, i.e., have Sig. values above 1.00.

B) they are both significant, i.e., have Sig. values below 1.00.

C) they are both significant, i.e., have Sig. values equal to or below .05.

D) at least one is significant, i.e., has a Sig. value above .80.

E) None of the above; because the F value in the ANOVA was not significant, you should not

have even looked at the Coefficients output.

Answer: C

87) Assume you are developing a general conceptual model for the Advanced Automotive

Concepts dataset. Which of the following variables would most likely be your dependent

variable?

A) probability of purchasing a particular auto model

B) customer income

C) customer gender

D) customer education

E) customer occupation

Answer: A

88) The National Football League office discovered data covering attendance at professional

football games in the late 1940s and early 1950s. The game with the highest attendance was

between the St. Louis Cardinals and the New York Giants. The office also found considerable

information that someone had collected on each game day, such as the level of GDP, the

DOW, numbers of persons employed, number of new businesses formed during the week

preceding the game, and the population. A student intern took the information and built a

regression model to predict game attendance for the upcoming season. The model:

A) should accurately predict game attendance.

B) should not predict game attendance accurately because the variable levels of today (i.e.,

population, DOW, etc.) are out of range of those used to build the regression model.

C) should predict game attendance accurately because the variable levels of today (i.e.,

population, DOW, etc.) are out of range of those used to build the regression model.

D) should predict game attendance accurately because the variable levels (i.e., population,

DOW, etc.) are within range of those used to build the regression model.

E) does not have enough information.

Answer: B

Test Bank for Marketing Research

Alvin C. Burns, Ronald F. Bush

9780133074673, 9780134895406, 9780134167404