10: Non-Parametric Analyses

Question 1

Shapiro-Wilk

Answer 1

Tests the null hypothesis that sampled data is drawn from a normally distributed population.

Question 2

Spearman’s rho

Answer 2

Most common alternative to Pearson’s r.

Question 3

Kendall’s Tau

Answer 3

Alternative for Pearson’s r. Better for fewer than 20 cases, or tied ranks.

Question 4

Mann-Whitney U

Answer 4

Alternative to the independent t-test.

Question 5

Wilcoxon

Answer 5

Alternative to paired t-test.

Question 6

How is effect size calculated for t-test alternatives?

Answer 6

z / sqrt(N)

Question 7

Kruskal-Wallis

Answer 7

Alternative to independent one-way ANOVA.

Question 8

How is effect size calculated for Kruskal-Wallis?

Answer 8

x^2 / (N - 1)

Question 9

Friedman’s

Answer 9

Alternative to repeated measures one-way ANOVA.

Question 10

One-variable Chi square

Answer 10

Goodness-of-fit test. Deals with categorical (not percentage) data. Difference between expected and obtained frequency.

Question 11

Chi-square test for independence

Answer 11

Measures association between two categorical variables. Calculates the difference between expected and observed frequency.

Question 12

2 x 2 Chi-square

Answer 12

Expected frequency should be >5.

Question 13

What is Fisher’s exact probability test for?

Answer 13

When the expected frequency assumption of 2x2 Chi-square is broken. Less sensitive to small expected frequencies.