The SAS Workbook

CHAPTER 22 Basic Descriptive Statistics 149

Directions

Produce the following

A. One-way frequency tables for the variables GENDER, PRIM_DX, SEC_DX, VITAMINS,

and PREGNANT.

B. Vertical bar charts for the variables GENDER and PRIM_DX.

PROBLEM 4

Generating Histograms for Numeric Variables

Tools

PROC CHART

Data

SAS data set CLINICAL

Directions

Produce the following charts

A. Construct a histogram for the variable HR (heart rate). Let PROC CHART select the midpoints

and the number of bars for the histogram.

B. Rerun part A, but specify that the midpoints for the bars should run from 40 to 100 and be

spaced 10 apart.

C. Rerun part A, but specify 10 bars.

PROBLEM 5

Computing Means where There Is More than One Observation per Subject

Tools

PROC MEANS

CLASS and OUTPUT statements

Data

Raw data file CLIN_X.DTA

150 The SAS Workbook

Directions

If you look at the contents of the CLIN_X.DTA data file you will see that there are from one

to five observations for each ID. If you want to compute the mean HR, SBP, and DBP for your

patients, you can first compute a mean for each ID and then take the mean of the means, or you

can select a single observation for each patient (such as the first or last).

Using the CLIN_X.DTA data file, create a SAS data set and compute an overall mean for HR,

SBP, and DBP using both of these methods. For the latter method, use the last visit for each

patient. Compute all statistics to three decimal places.

151

CHAPTER 23

PROBLEMS

1 Computing a Chi-square and Fisher's Exact Test from Raw Data . . . . . . . . . . . . . . .151

2 Computing a Chi-square from Frequency Counts . . . . . . . . . . . . . . . . . . . . . . . . . . . .152

3 Computing Two-way Frequencies: Chi-square and Fisher's Exact Test . . . . . . . . . . .152

4 Computing Several Chi-square Statistics from Frequency Counts . . . . . . . . . . . . . . .153

5 Computing a Chi-square on a 2 X 3 Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154

6 Testing Differences in Proportions for Paired Samples: McNemar's Test

(Raw Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154

7 Testing Differences in Proportions for Paired Samples: McNemar's Test

(Frequency Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .155

INTRODUCTION

This chapter contains problems based on testing differences in proportions. Fisher’s Exact Test,

chi-square tests, and McNemar’s test are covered.

PROBLEM 1

Computing a Chi-square and Fisher’s Exact Test from Raw Data

Tools

PROC FREQ

Data

Raw data file CPR.DTA

Directions

Using the raw data in the CPR.DTA file, test if the proportion of patients surviving a cardiac

arrest is related to the three independent variables ventricular fibrillation (V_FIB), patient on a

respirator (RESP), and age greater than 70 (AGEGROUP). Compute both the chi-square statistic

and the Fisher’s exact test probability.

Tests of Proportions

152 The SAS Workbook

PROBLEM 2

Computing a Chi-square from Frequency Counts

Tools

PROC FREQ

WEIGHT statement

Data

Data were collected on GENDER (M and F) and voting preference (candidate A and candidate B).

The data appear as follows.

Gender Candidate Count

--------------------------------

M A 35

F A 45

M B 70

F B 50

Directions

Using the frequency count data in the Data section, compute chi-square.

PROBLEM 3

Computing Two-way Frequencies: Chi-square and Fisher’s Exact Test

Tools

PROC FREQ

Data

SAS data set CLINICAL

Directions

A. Using the SAS data set CLINICAL, create a two-way table of the variable VITAMINS by

the variable GENDER. The table should look like the following one.

GENDER

Female Male

-----------------

| | |

No | | |

VITAMINS -----------------

| | |

Yes | | |

-----------------