New Program Puts It All on the Table

Have you noticed that the kinds of tables of data that most programs are able to print don’t look nearly as good as the ones you commonly see in books and journals and even in newspapers?

TPL Tables, $895, from QQQ Software Inc. of Addington, Va. (703) 528-1288, will produce publication-quality tables from virtually any data source. What’s more, it can structure those tables in much more elaborate ways than can any spreadsheet, database or statistical software program.

Tables are the most concise and least biased way to present a large amount of data. The stock market tables are a good example, as are the financial statements in corporate annual reports.

Some tables are fairly simple to construct, and spreadsheet or database software will do the job. Others are complex and difficult to produce. A good example are the summary tables you often see in articles reporting the results of public opinion polls.

Those tables that show you the percent of men and women who say they would vote for various political candidates are called “cross tabulations.” They summarize hundreds or thousands of separate data elements into a few succinct categories.

But even the few powerful database and statistical programs that can produce cross tabulations give you little or no control over the printed result. They don’t let you mix type sizes along with boldface and italics for easy readability.

The better spreadsheet programs, such as Excel and Lotus 1-2-3, do give you some control over the typography of tables they can create, but they don’t do cross tabs, nor do they handle very large databases.

TPL Tables, which runs on IBM PC and compatible computers, does it all. It does cross tabulations of virtually any size database, performing most any mathematical transformation or summarization you could need. It also organizes your data into any order desired, which is both important and difficult or impossible to do with other software.

TPL Tables also lets you specify the printing format for any table you design and use a printer or typesetter equipped with the PostScript page description language so that you get true typeset-quality tables.

This is decidedly not Windows or graphics-oriented software, despite the elegant tables it produces. It is pure mainframe in look and feel and heritage. That’s not surprising, since TPL Tables was inspired by mainframe software created by the federal Bureau of Labor Statistics to produce the massive numbers of tables that it publishes. That software, dating from the 1970s, is called TPL for Table Producing Language. Stephen Weiss and Pamela Weeks, who created TPL Tables for the PC, played major roles in the development of the mainframe version.

The advantages of the mainframe programming approach on the PC are processing speed and simplicity, once you learn the basic English language commands that will construct a table. The disadvantage, especially for typeset-quality PostScript-formatted tables, is that you can’t see what they will really look like until they’ve been printed.

TPL Tables is not a database or spreadsheet program, so you can’t use it to create or store your data. Instead, it will read most any data file you can get into your PC, no matter what program created it, and produce easily understood tables from that data. The catch is that you have to understand the file structure and be able to describe it in a set of text instructions called a code book that tells TPL Tables what each element of the database is and how to interpret it.

I found it much harder to write a successful code book describing a data set than to write the subsequent program to display the tables from the data.

There is an interactive program for creating a code book in which you respond to a series of questions. But many questions are cryptic; there is no on-line help, and you often can’t back up to correct an error several steps back.

Table request programs have to be written with a text editor, but they are quite simple and logical and there are numerous examples in the thick, loose-leaf manual to guide you.

TPL Tables’ ability to organize your data just the way you want to see it is impressive. For example, a simple four-line set of commands would build a table of sales organized by sales people, grouped by sales region down the side, with revenue grouped across the top by each product within each product category. You could instantly spot who can and can’t sell different products among your product line.

Basic statistics such as percents, maximums, minimums, quantiles (equal groupings of data), means and medians can be computed, as can numerous other mathematical formulas. Various categories of data can easily be weighted, which is especially important in preparing survey results.

I think TPL Tables would quickly become indispensable to those who need to prepare tables.

Computer File welcomes readers’ comments but regrets that the author cannot respond individually to letters. Write to Richard O’Reilly, Computer File, Los Angeles Times, Times Mirror Square, Los Angeles, Calif. 90053.

RICHARD O’REILLY is director of computer analysis for The Times.