Deck 14: Nonparametric Statistics

Question 1

What is the purpose of non-parametric statistics?

Answer 1

Non-parametric statistics allow us to test for statistically significant effects even when the dependent variable(s) may not be normally distributed.

Question 2

What kind of parametric statistic is a Mann-Whitney U test analogous to?

Answer 2

It is similar to an independent-samples t test, but is more appropriate for use with a dependent variable that may not be normally distributed. This is the non-parametric test to use when comparing two independent samples on a single dependent variable.

Question 3

What kind of parametric statistic is a Wilcoxon Signed-Rank test analogous to?

Answer 3

It is similar to an dependent-samples t test, but is more appropriate for use with a dependent variable that may not be normally distributed. This is the non-parametric test to use when comparing a single sample of two variables (e.g., a pre-test/post-test situation) or when comparing matched samples on a single dependent variable.

Question 4

What kind of parametric statistic is a Kruskal-Wallis test analogous to?

Question 5

What is the purpose of conducting a chi-square test of independence?

Question 6

What kind of variables are appropriate for use in a chi-square test of independence?

Question 7

What is a nonparametric statistic and when should it be used?

Question 8

Suppose that I have been hired by a large company to investigate an employee's accusation that there are more men than women in the higher-paying, upper management level positions within the company. I select a random sample of 100 male and 100 female employees and find out how many of each gender are in entry level, middle management, or upper management positions within the company. I get the data that a presented in the table below:
a. Calculate the expected frequencies for each cell in the table.
b. Using the formula in Table 14.5, find the values needed to calculate the
c. Calculate the χ² value and decide whether it is statistically significant.