Classic Computer Magazine Archive COMPUTE! ISSUE 43 / DECEMBER 1983 / PAGE 196

TI Statistics

Roger B. Crampton

In many professions there is a need to analyze something statistically. Engineers, medical researchers, psychologists, and social scientists often must generalize from data samples and make predictions concerning the probability of events. Not many years ago this data analysis was a tedious and expensive task, using calculators and many clerical assistants to perform manual computations.

In addition, because the mathematics of statistics appear so formidable, professionals often hesitate to try to explain the implications of their data.

Texas Instruments has helped remove some of this anx­iety with its Statistics Command Module, a series of programs that perform dozens of the most commonly needed statistical techniques.

The module leads the researcher through the procedures of statistical analysis in a friendly and efficient way. The only hardware requirements for running complicated statistics programs are the TI-99/4 or 4A console, a monitor, and the module. While not essential, a printer and a cassette or disk drive will eliminate having to reenter the data set and file structure if you want a second look at your findings.

Learn The Basics First

Before plugging in the module, it is important that you thoroughly read the 48-page instruction manual at least twice. The time spent will be rewarded with a clear understanding of the module's capabilities and a basic understanding of statistics itself.

When the module is inserted into the console, a title screen is displayed, followed in a few seconds by the first of several menus (see Figure 1).

Figure 1: Program Options

PRESS

1 TO CREATE A NEW FILE
2  LOAD AN EXISTING FILE
3  USE SIGNIFICANCE LEVEL CALCULATOR
4 QUIT

Typing 1 allows you to set up your file structure. You name each variable, determine its type (alphanumeric, integer, decimal, or scientific notation), and enter the maximum number of digits of each variable. The number of variables allowed depends on the width of each entry and the number of observations. Conversely, the number of observations that you will be able to enter depends on the number and specifications of the variables you have selected. It is important to carefully define the parameters of the problem so that you will be able to use all of your observations without getting a MEMORY FULL message.

Another reason for care when you specify the initial file structure is that there are no provisions for editing file specifications once they have been set up.

When the file structure has been established, the next menu will be displayed (see Figure 2).

Figure 2: Basic File Structure

MAIN INDEX

PRESS

1 TO SEE FILE DEFINITION
2  ENTER OBSERVATIONS
3  CHANGE OBSERVATIONS
4  ANALYZE DATA FILE
5  SAVE DATA FILE
6  QUIT

At any time, you can return to the main index, select option 1, and review the specifications of the file. But remember, you don't have a chance to change any­thing, unless you're willing to reenter the entire file definition.

Entering Data

When you are certain that your file is arranged exactly as you want it, it's time to select option 2 and begin entering data. The module will prompt you with the names of the variables as each is typed in.

Data entry is slow. A fast typist must slow down to about half speed because the module will not accept entries at usual typing speed. The first variable value will be accepted, but the initial digit of the second or succeeding variables often gets lost. An entry of 84 becomes 4, an entry of 1.3794 will become .3794.

After all your data has been entered, you can verify its accuracy by selecting option 3 from the menu and single-stepping through your data set, making any changes that are necessary. There is no provision for LISTing your data to a printer to check each observation for accuracy. This would be desirable, especially to see that decimal data is properly entered.

Analyzing The Data

At last the preliminaries are completed, and you're ready to get down to the real purpose of the program: looking at your data from a statistical point of view. By pressing option 4 of the main index you are given a new menu (see Figure 3).

Figure 3: Analysis Options

ANALYZE DATA FILE

PRESS

1 FOR DESCRIPTIVE
     STATISTICS
2   CORRELATION
3   LINEAR REGRESSION
     ANALYSIS
4   INFERENTIAL STATISTICS
5 TO EXIT THIS SECTION

Each of the four options is thoroughly described in the user's manual. Few researchers will need all of the procedures available. In fact, it may be best to learn to use one technique at a time. The enormous amount of information from the analysis of even a simple data set can be overwhelming.

Although the Statistics Command Module can provide volumes of information about a data set, it does have limitations. Evaluation of a great deal of information can be hampered by memory problems if the module is used without memory expansion. In addition, no provision exists in the program to screen out data entry errors by specifying acceptable ranges for each variable.

For nonprogrammers who need a means of analyzing fairly simple data sets, however, the module can be a useful tool. And for anyone seeking a relatively painless introduction to statistics, it is superb.

Statistics
Texas Instruments
Box 53
Lubbock, TX 79408
Dallas, TX
(800)858-4565