Exam 16: Nonparametric Tests

At the Seymour Clinic, the total number of patients seen by three doctors over three days is as follows: $\quad$$\quad$$\quad$$\quad$$\quad$$\quad$$\quad$$\quad$$\quad$$\text { Physician }$ Day Dr. Able Dr. Baker Dr. Chow Mon 19 25 27 Tue 20 22 27 Wed 22 24 32 Thu 19 20 22 Fri 20 34 27 Which nonparametric test would you use to compare the median number of patients seen by the doctors on each day?

The Wilcoxon signed-rank test has good power over a range of nonnormal populations.

The Kruskal-Wallis test is equivalent to comparing medians in c groups.

The Kruskal-Wallis test requires c independent samples (where usually c > 2).

The Friedman test resembles the Kruskal-Wallis test except that there are two grouping factors (rows, columns) instead of one grouping factor (columns).

The one-sample runs test uses binary data (only two possible values).

At the Food Barn, children can order from the adult menu (A) or the child's menu (C). Below is the pattern of menu orders for 24 children last Wednesday evening. A, A, A, A, C, A, A, C, C, C, A, A, A, A, A, A, C, C, C, A, C, C, C, C Which statement is most accurate?

Nonparametric tests generally are more powerful than parametric tests when normality cannot be assumed.

Which nonparametric test is used to detect nonrandomness in sequential observations?

Spearman's rank correlation always lies within the range -1.00 to +1.00.

The one-sample runs test compares medians of two or more groups.

Which is a nonparametric test for runs (autocorrelation) in a sequence of binary data?

The Wilcoxon signed-rank test is analogous to a one-sample parametric test of a mean.

We sometimes call the Wilcoxon rank sum test the Mann/Whitney test, but the two are the same.

The Wilcoxon rank sum test (Mann-Whitney test) can be used as a test for equality of two population medians.

If there are three or more populations with ordinal data, what is the appropriate test to determine whether the distributions are equal?

Hoping to reduce the waiting time, a doctor's office tried a new way of scheduling its appointments. Populations of waiting times are believed to be similar except in central tendency. However, waiting times may not be normally distributed, so a nonparametric test was chosen to compare the waiting times. The Wilcoxon/Mann-Whitney test results are shown below: $\text { Mann/Whitney Test }$ n sum of ranks 21 600.5 Old Way 25 480.5 New Way 46 1081 Total $493.50$ expected value $45.31$ standard deviation $2.35 \mathrm { z }$ , corrected for ties $.0094 \mathrm { p }$ -value (one-tailed, upper) Which is the best conclusion?

Spearman's rank correlation is named for a British brewer who was testing beer samples.

The one-sample runs test is similar to a test for autocorrelation.

The Kruskal-Wallis test is almost as powerful as one-factor ANOVA even when normality can be assumed.