Lesson 13 : Multivariate Analysis

Each question has only one best answer. Circle clearly the letter of the best answer. If you make a mistake, cross out the circle, and write the letter in capitals next to the question. If a question has both a capital letter and is circled, the letter will be considered to be the answer.

 

1.     Why would you want to use multivariate analysis?

a.     To describe the central tendency and distribution of multiple variables.

b.     To describe how an interval-level variable covaries with another interval-level variable.

c.      To test whether the relationship between an interval-level variable and a categorical variable is significant.

d.     To find out how multiple variables are related to develop a causal theory for these relationships.

e.     None of the above.

 

2.     What is the elaboration method?

a.     Complicated phenomena require elaborate explanations; hence the name elaboration method

b.     Elaboration involves teasing out the complexities of a bivariate relationship by controlling for the effects of an antecedent or an intervening variable.

c.      Putting multiple variables into a single equation that simultaneously takes into account the interactions among all the variables.

d.     Testing specific theories about how the independent variables in a multiple regression influence each other.

e.     None of the above.

 

3.     Tom finds that there is a moderately strong relationship between education and knowledge of email. He looks at the relationship across men and women and finds that, in the sample of women, the relationship is very strong while in the sample of men the relationship disappears. What kind of analysis is Tom doing?

a.     Partial correlation

b.     Multiple regression

c.      Elaboration

d.     Multidimensional scaling

e.     Bivariate analysis

 

4.     Holly is studying the relationship between computer use and levels of online purchasing. She is looking for a direct way to control for the effects of three other variables in this relationship. What kind of analysis should she consider?

a.     Partial correlation

b.     The elaboration method

c.      Zero-order correlation

d.     ANOVA

e.     None of the above.

 

5.     What does r 12·3 mean?

a.     The explained variance of variables 1,2, and 3

b.     The multiple R of a three-way intersect.

c.      The partial correlation of r  with variable 3, controlling for variables 1 and 2.

d.     The correlation between variable 1 and variable 2, controlling variable 3.

e.     None of the above.

 

6.     Which statistic represents the amount of variance in the dependent variable accounted for by two or more independent variables simultaneously?

a.     Multiple R.

b.     Multiple R 2

c.      Pearson’s r 2

d.     Partial correlations

e.     Elaboration method

 

7.     What is the general regression equation for two independent variables?

a.     y= a + b

b.     y= a + bx

c.       y = a  + b1x1

d.     y = a + b1x1 + b2x2

e.     y = a + b1x1 + b2x2 + b3x3

 

8.     Steve wants to know how age and income affect time online. He knows from the zero-order correlations that age accounts for 13% of the variance, and income accounts for 26% of the variance, in time online. What can Steve predict about the multiple 2  when he examines the effect of age and income together?

a.     2  = .39

b.     2  > .39

c.      2  < .39

d.     2   = .13

e.     He can’t tell until he runs the multiple regression with both variables.

 

9.     Tim wants to know what factors affect distance learning test scores. He collects data on 15 different independent variables. Tim uses a computer software package to identify a subset of these independent variables that explains the scores. The program first identifies the independent variable that correlates best with the dependent variable, then incrementally adds additional variables that increase the explain variance by at least 1%. What kind of analysis is Tim using?

a.     Factor analysis

b.     Path analysis

c.      Simple regression

d.     Multiple regression

e.     Stepwise multiple regression

 

10.            Emmanuel is trying to understand technology implementation in organizations. He thinks that technology use depends on the organization budget, its annual staff turnover, its IT training budget,  task security, privacy, and mobility requirements. Emmanuel wants to test a conceptual model for how the independent variables affect each other and technology use. What kind of analysis should Emmanuel use?

a.     Simple regression.

b.     Multiple regression

c.      Stepwise multiple regression

d.     Path analysis

e.     Multidimensional scaling

 

11.            When does multi-collinearity occur?

a.     Whenever a researcher uses more than three or four independent variables to predict a dependent variable.

b.     Whenever a researcher examines two or more dependent variables that are themselves related.

c.      Whenever the researcher is unsure of the direction of causality.

d.     When two or more independent variables are highly correlated such that it is difficult to determine their separate effects on the dependent variable.

e.     None of the above.

 

12.            Tina wants to measure what users experience when they view websites. During her exploratory data collection phase, Tina asks users who have just viewed a website to rate their experience on 30 different variables. Tina wants to reduce her 30 variables into a smaller number of related, underlying dimensions. What kind of analysis should she consider?

a.     Multiple regression

b.     Stepwise multiple regression

c.      Partial regression

d.     Factor analysis

e.     Multi-collinerarity.

 

13.            There are different kinds of factor analysis. What kind identifies factors that have as little correlation with each other as possible? 

a.     Orthogonal solution

b.     Diagonal solution

c.      Partial solution

d.     Bivariate solution

e.     None of the above.

 

14.            Factor analysis uses a set of original variables to generate a new set of super variables or factors. What are factor loadings?

a.     The degree to which factors are correlated with each other

b.     The correlations between the original variables and the new variables or factors

c.      The correlations of each respondent to each factor.

d.     The degree to which each factor explains the variance in the original data set

e.     None of the above.

 

15.            John has just finished asking informants to sort 10 kinds of web search methods into piles based on their similarities. After aggregating his data, John has a 10-by-10 similarity matrix. What kind of analysis should John use if he wants to visually depict how people think search methods are similar to each other?

a.     Factor analysis

b.     Path analysis

c.      Multidimensional scaling

d.     Discriminant function analysis

e.     Correspondence analysis.

 

16.            If you had a similarity matrix of 10 items, what is the maximum number of dimensions you would need to plot the items perfectly (i.e., without any stress) using multidimensional scaling?

a.     2

b.     3

c.      5

d.     9

e.     10

 

17.            Adam is looking at friendship networks among online users. Over one week, he records the number of times people interact with each other via email. Adam wants to see if there are groups who communicate together. What kind of analysis should Adam try?

a.     Factor analysis

b.     Discriminant function analysis

c.      Cluster analysis

d.     Path analysis

e.     Multiple regression.

 

18.            What are some methods for conducting a cluster analysis?

a.     Single-link

b.     Farthest-neighbor

c.      Nonmetric

d.     Metric

e.     a and b

 

19.            Bruce wants to identify which new computer owners decided to pay extra for warranties. Bruce sampled 100 new owners who purchased warranties and 100 new owners who did not. He collected data on each respondent’s age, income, marital status, family size, and online access. What kind of analysis should Bruce consider if he wants to see how well these data classified people into buyers or non-buyers of computer warranties.

a.     Factor analysis

b.     Discriminant function analysis

c.      Cluster analysis

d.     Path analysis

e.     Multiple regression.

 

20.            T F  Multiple regression assumes that the independent variables in a model are not correlated with each other.

a.     True

b.     False

 

21.            T F  Multiple regression, like simple regression, produces a PRE measure.

a.     True

b.     False

 

22.            T F  Path analysis lets you test a particular theory about the relationships among a set of variables.

a.     True

b.     False

 

23.            T F  Multidimensional scaling requires a matrix of similarity data (a square matrix).

a.     True

b.     False