The UNIVARIATE Procedure

Overview

The UNIVARIATE procedure provides data summarization tools, high-resolution graphics displays, and information on the distribution of numeric variables. For example, PROC UNIVARIATE

• calculates descriptive statistics based on moments

• calculates the median, mode, range, and quantiles

• calculates the robust estimates of location and scale

• calculates confidence limits

• tabulates extreme observations and extreme values

• generates frequency tables

• plots the data distribution

• performs tests for location and normality

• performs goodness-of-fit tests for fitted parametric and nonparametric distributions.

• creates histograms and optionally superimposes density curves for fitted continuous distributions (beta, exponential, gamma, lognormal, and Weibull) and for kernel density estimates

• creates quantile-quantile plots and probability plots for various theoretical distributions and optionally superimposes a reference line that corresponds to the specified or estimated location and scale parameters for the theoretical distribution

• creates one-way and two-way comparative histograms, comparative quantile-quantile plots, and comparative probability plots

• insets tables of statistics in the graphical displays (high-resolution graphs)

• creates output data sets with requested statistics, histogram intervals, and parameters of the fitted distributions.

The Default Univariate Analysis shows a default univariate analysis for student exam scores. The statements that produce the output follow:
```options pagesize=36;
proc univariate data=score;
run;```
By default, the tests for location examine the hypothesis that the mean is equal to zero. Optionally, you can request a test for the hypothesis that the mean is equal to a specified value .

A Univariate Analysis with Tests for Normality and Plots of the Data Distribution and An Output Data Set That Contains Univariate Statistics are the result of a more extensive univariate analysis. The analysis examines the data distribution of student exam scores and creates an output data set that saves percentiles that were not computed by default. The statements that produce the analysis also

• specify the null hypothesis for the tests for locations

• perform tests for normality

• plot the data distribution

• specify the analysis variables

• request confidence limits for parameters and quantiles

• list the five highest and lowest extreme values

• print an output data set that contains percentiles.

For an explanation of the program that produces both these reports, see Examining the Data Distribution and Saving Percentiles .