27: Nonparametric

Question 1

Which of the following conditions does not warrant the use of nonparametric statistics?
a. Samples are very small.
b. Data are at the interval level.
c. Data do not show homogeneity of variance across groups.
d. Data do not come from normal distributions.

Answer 1

B

Rationale: Parametric statistics can be used with interval level data. warrant use of nonparametric tests, even if data are ratio/interval.

Question 2

Compared to their parametric counterparts, nonparametric statistics:
a. Tend to have slightly lower power.
b. Do not use critical values to determine significance.
c. Need larger samples.
d. Can only used with ordinal data.

Answer 2

A

Rationale: Nonparametric statistics tend to be less powerful than their parametric counterparts, although the difference is often negligible.

Question 3

A researcher designs a study to measure differences in anxiety (measured 1-10) in patients who do
and do no receive an anxiety medication.
a. The Kruskal Wallis ANOVA
b. The Wilcoxon signed-ranks test
c. The Mann-Whitney U test
d. The sign test

Answer 3

C

Rationale: The Mann-Whitney U test compares two independent groups

Question 4

CA study was designed to look at the difference in pain across 4 groups, each receiving a different dose of the medication. The results of a Kruskal-Wallis test showed H = 8.4 (p = .02). Based on the results of a multiple comparison, shown above, which statement is not a correct conclusion?
A. Using .14 mg is better than any other dosage.
b. Using .10 is not more effective than using .14 mg.
c. Using .10 mg will be more effective than using .6 mg.
d. Using .2 mg or .6 mg is not as effective as either .10 mg or .14 mg.
.

Answer 4

A

Rationale: The .2 mg and .6 mg ranks are not different. The .10 mg and .14 mg ranks are not different. However, .10 mg and .14 mg are better than two smallest doses. We cannot say that .14 is better than all others, as it is not better than .10 mg.

Question 5

AA study was done to test the difference in heart rate with four different exercises. A sample of 10 patients were tested, each using the four exercises in random order. The appropriate test for this study would be:
a. Kruskal-Wallis ANOVA
b. Mann-Whitney U test
c. Friedman ANOVA
d. Wilcoxon signed-ranks test

Answer 5

C

Rationale: The Friedman ANOVA is a one-way analysis for repeated measures.

Question 6

Significance for the sign test is determined using:
a. The t-distribution
b. The binomial distribution
c. The F distribution.
d. The 2 distribution

Answer 6

B

Rationale: The binomial test is used to determine significance for the sign test based on whether there are more positive or negative differences.