Deck 7: Non-Parametric Models

Question 1

In an enterprise competition, 10 entrepreneurs are asked to rate two business plans for potential success on a scale of 1(low value) to 5 (high value). What procedure would be used to test the hypothesis that the ratings are not significantly different?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Answer 1

Wilcoxon signed-rank test

Question 2

The results from Q12 are below. What conclusion should be drawn?
Ranks

Test Statistics (a)

A) There is a significant difference in ratings, and Plan B is higher than Plan C which is higher than Plan A.
B) There is a significant difference in ratings, with Plan A being higher than Plan C which is higher than Plan B.
C) There is no significant difference in ratings,
D) Not enough information.

Answer 2

There is a significant difference in ratings, and Plan B is higher than Plan C which is higher than Plan A.

Question 3

The non-parametric equivalent of the one-way between-measures ANOVA is the?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Answer 3

Kruskal-Wallis test

Question 4

A survey was conducted with customers of two gas suppliers - Atomic Gas and English Gas. The objective was to assess whether one group of customers spent more (in £s) on their annual gas bill than the other. Initial tests suggested the absence of normality in the dependent variable. A non-parametric test was deemed appropriate, but which do you think that the researchers should use?

A) Paired t-test
B) Mann-Whitney test
C) Wilcoxon signed-rank test
D) Independent t-test

Question 5

The same entrepreneurs in the above study are asked to rate a third business plan (C). What test would be used to see if there is a significant difference in potential success ratings amongst the three business plans, once again assuming that a non-parametric test is necessary?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Question 6

An analyst at your company has told you that the non-parametric equivalent of the paired t-test is the 'non-paired u-test'. You don't believe him, but what is the correct answer?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Question 7

A colleague has asked you what the Jonckheere-Terpstra test assesses. You tell them that it?

A) Does essentially the same thing as the Kruskal-Wallis test
B) Does essentially the same thing as the Wilcoxon signed-rank test
C) Tests for an ordered pattern to the medians of the groups you're comparing
D) Does both a and c

Question 8

The data in Q10 produce the following results. What conclusion should be drawn?
Test Statistics (b)

A) There is a significant difference in rating, with Business Plan A being rated higher than Business Plan B.
B) There is no significant difference in ratings
C) There is a significant difference in rating, with Business Plan A being rated lower than Business Plan B.
D) Not enough information is available to make a decision.

Question 9

You would like to test whether male or female managers incubate teams with higher employee satisfaction. You therefore run the Mann-Whitney test on the data. What do you expect will be approximately equal in the two groups if you accept the null hypothesis?

A) Means
B) Ranks
C) Variance
D) Mean ranks

Question 10

The non-parametric equivalent of the two-sample independent t-test is the?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Question 11

What do you think is the primary reason why some business analysts prefer to use parametric tests even when the normality assumption has been violated?

A) Non-parametric tests are harder to carry out.
B) Non-parametric tests are less powerful.
C) Parametric tests are less robust.
D) Parametric tests are more likely to show causal effects.

Question 12

A group of 8 stockbrokers were assessed for the worth of their investments over the course of a year. The following year the CEO reduced the commission rate payable to stockbrokers making good investments. Once again, the value of the broker's annual investment was totalled. The data were not found to be normally distributed. Which non-parametric test should the analyst therefore use on this project?

A) Paired t-test
B) Mann-Whitney test
C) Wilcoxon signed-rank test
D) Wilcoxon rank-sum test

Question 13

The non-parametric equivalent of the one-way repeated-measures ANOVA is the?

A) Mann-Whitney test
B) Wilcoxon signed-rank test
C) Friedman test
D) Kruskal-Wallis test

Question 14

Your manager has asked you to find out the alternative name for 'non-parametric test'. Which of the following will you feed back?

A) Non-normal test
B) Data-free test
C) Non-continuous test
D) Distribution-free test