Data Use: A review of four powerful data analysis programs for Windows 95

Editor’s note: Steven Struhl is vice president, senior methodologist, at Total Research in Chicago.

Statistical analysis programs for the PC have become increasingly powerful over the last several years - as many of you doubtless already know. The most widely used programs, SPSS and SAS, have incorporated many major enhancements and new features. SAS has even conceded that its suite of programs should run under Windows, adding an interface that more or less looks like other Windows programs - and an extra 1,000-page manual to describe new procedures and general "enhancements through version 6.11."

SPSS, Inc., though, has kept an even more torrid pace in product development. Its main program has had three major releases in the last 18 months (and we will discuss the newest, Version 8.0, in an upcoming issue). This program has made major strides toward more formatted, finished-looking output in the form of "objects" which other Windows programs can use and edit. It has refined its menu system, and added new procedures. And still, at the same time, SPSS has added a large number of new products to its line of offerings.

This review discusses four programs with widely divergent capabilities in analyzing, displaying, translating, and exploring data. They share one common thread: SPSS now manufactures and/or distributes all of them. (These programs work as stand-alone products, not requiring you to have SPSS to use them.) Each of these programs will add substantially to your data-handling abilities, even if you already own SPSS and all its options.

DBMS/COPY

This program takes a basic idea and executes it with tremendous power and skill. DBMS/COPY bills itself as "the tool for software connectivity," and this is as precise a description as you will find for a software product. DBMS/COPY allows you to talk to (and use the data from) nearly any analytical or database program that you have ever seen - and many that you have not.

The need to connect becomes more important as we make more use of data stored on mainframe-style computers. In particular, now that we can dig into huge customer and prospect databases, we find ourselves in territory where Windows cannot make it easy to cut and paste the data we want into our favorite programs.

DBMS/COPY fills this gap nicely. It imports and exports to and from about 75 basic file formats and includes several versions, or flavors, for many of them. I find two specific features of DBMS/COPY of most interest. First, this program lets you use SAS data files in their "native" formats even if you do not have SAS. Second, it gives you an easy, visual method to read in multi-line ASCII data files. (That is, files that the "text import wizards" in Excel, for instance, cannot handle.)

The range of DBMS/COPY is almost confounding. Need to use Rbase data? Rats? Forecast Pro? Abstat? NCSS? DBMS/COPY will handle all of them.

Have a client or colleague who insists on getting data in Y-Stat format? PRODAS? Again, no problem. (Perhaps you are luckier than your author and have not met people like these - but then, things always can change suddenly for the worse.)

In addition to translating data, DBMS/COPY lets you view the data, sort it, filter it, transform it with numerous functions, and (for programs that support this) change the data type as needed.

The program works swiftly and efficiently. It launches quickly and plainly, with no fancy opening screens or "splash panels." File translations run nearly as fast as doing a simple file copy. The commands have an amusingly quirky, but direct, quality. The "interactive" mode, for instance, has the label notifying you that "it doesn’t get any easier than this." Also, I believe DBMS/COPY is the only data translation program having an author who appeared in a production of Shakespeare’s Much Ado about Nothing. A careful examination of the DBMS/COPY Web page (www.conceptual.com) will reveal this and several other astounding facts.

There are just a few things DBMS/COPY does not do that I would put on a wish list. The current version does not yet read long value labels from SAS, but you can expect this soon. (SAS does not cooperate with meddling outsiders who want to know about their file formats, so the team at DBMS/COPY has had to crack the code themselves.) Those of you who are not familiar with, or are rusty about, the notion of value labels will find a description in the chart on page 15. Mavens can skip this explanation, but miss a couple of awful jokes.

Also, DBMS/COPY cannot handle ASCII files where the records vary in the number of "cards" or lines of data. This trick is fairly difficult, though, and fortunately, the need only rarely arises - but still it would be a useful feature to have.

If you need real "connectivity," or file translation power, you will find DBMS/COPY one of the great software bargains of our time, at $295. You can get more information about this program at the SPSS Web site (www.spss.com), or directly from the manufacturer, Conceptual Software.

DeltaGraph 4.0

This program provides charting and graphing power of the highest order. It has the best mix of simplicity of use, chart customization, and depth of features of any charting/graphics package I have encountered. DeltaGraph has a long history on the Macintosh, and now in its Version 4.0 series (the current release is 4.04) has become fully compatible with Windows 95.

Even though programs like Excel, Harvard Graphics, and Lotus Freelance - and SPSS itself - have strongly improved charting abilities, DeltaGraph has stayed several steps ahead. It offers so many different chart types - grouped into "galleries" - that just browsing through its offerings may give you new ideas about ways to display data.

Figure 1 shows a small sample from the chart gallery. In addition to providing many preset graph types, the program allows you to customize nearly anything on a chart, and then save the result in the gallery. This is definitely the program to have if you need everything "just so," down to the size and placement of the tick marks at the border of the chart.

Among the many nice features of this program, one that your author probably likes best is its ability to make scatterplot diagrams with labels next to the points. This feature works quite handily with the various types of perceptual maps (whether actually from discriminant analysis, correspondence analysis, or whatever). You just put in the coordinates and the labels, and the program does nearly all the terrible plotting work you once had to do by hand. You will have to nudge some of the labels if the chart is crowded, but the program makes this type of on-screen editing quite simple.

Other useful charts rarely seen elsewhere include "x-y bars" and "bubble charts." In x-y bars the widths represent one series of numbers, and the heights another. For instance, you can make the widths of the bars represent the sizes of groups being analyzed, and the heights represent market shares among those groups. Unlike a simple bar chart, this can give you a quick visual impression of how much, for example, total sales volume goes into each group. (In the example here, the area of each bar -- height times width -- would show the proportion of volume accounted for by the group.) Bubble charts are useful because they can show both an "x-y" position for a point and represent, for instance, its importance by its size. This can add very nicely to several types of maps.

DeltaGraph has many analytical extra features, some of which have become more or less expected of a charting package. For instance, it can calculate and plot regression lines and fit various types of curves (power, exponential, logarithmic, etc.), and calculate new data with built-in formulas. More advanced features include the ability to add "error bars" to exact specifications (for instance, at 1.5 standard deviations around plotted points in either or both directions, if you wish), and an editor specifically for equations.

You can make a sort of a slide show with DeltaGraph alone, but I prefer to use it as a supplement to programs like PowerPoint or Excel, when they do not have enough charting power. Charts from DeltaGraph paste very nicely into these applications as "enhanced metafiles" which print at the best resolution your printer can offer. The charts also can be "ungrouped," and edited one element at a time, in PowerPoint and several other programs. DeltaGraph is particularly useful as an adjunct to these programs in part because, unlike them, DeltaGraph does not think it knows better than you when it comes to labeling. On bar charts in particular, DeltaGraph will include all the labels you request, and not skip some to satisfy its own sense of aesthetics.

Of course, Delta Graph can make all sorts of astonishing, and sometimes mind-boggling, charts with 3-D and 3-D effects. Unfortunately, while these seem incredibly interesting in the making, many audiences do not find them much fun, or highly comprehensible. It may take a little experience with a program this powerful to realize what an invitation it offers to overdo your charts.

There are only a few features on your author’s wish list for DeltaGraph. Salient among these is the inclusion of a "recently used file" list on the file menu. Nearly all Windows programs now have this feature, and it certainly can be very handy in opening and editing recent work.

Also, it would add to the program if the user could control the placement of labels on bar charts more closely. Now you have some general options like "inside," "at end" or "outside." The ability to specify labels’ distances from the ends of bars would help. At the least, the program would work better if it made sure labels fell beyond the ends of 3-D bars, when you ask for them to go "outside."

Some of DeltaGraph’s charting power has found its way into the newest release of SPSS, but even so you likely will find this a remarkably versatile and useful piece of software. It packs a tremendous amount of charting power, regardless of price -- and at $295 looks like another exceptional value for the money.

SPSS Diamond

Diamond is the most unusual of the programs discussed here. Its data visualization and exploration tools not only are unique, but they also can prompt you to think about patterns in data in new ways. It comes from impressive original work on data visualization done at IBM’s impressive-sounding T.J. Watson Research Center. It makes heavy use of color to define patterns in data, and uses some display methods that likely are not familiar even to advanced data analysts. At the same time, though, Diamond still seems to need somewhat more work to reach its full potential than the other programs reviewed here.

Let’s start with what Diamond can do, and then move on to the assorted wishes for, and scattered complaints about, its operations. Just the list of Diamond’s capabilities looks highly intriguing. These features include:

• Parametric snake plots
• Quadwise plots
• Parallel coordinates plot
• Triplewise plots (3-D scatterplots)
• Fractal foam plots
• Numerous univariate statistics
• Numerous bivariate statistics
• Best fit lines with "goodness-of-fit" thresholds for display

Let’s first briefly describe some of these display methods. Perhaps most intriguing, in name if nothing else, are fractal foam plots. This type of plot shows a special pictorial view of a bivariate correlation matrix. You first choose a variable to become the focus of the chart. This variable then is plotted as a large bubble at the center of the map, with other variables in the analysis mapped around its perimeter. The diameter of each bubble around the focus variable is proportional to the strength of their correlation. The shape and orientation of each bubble (i.e., its roughness, tilt, and flatness) also can be used to represent univariate statistics (skewness, kurtosis, and standard deviation) for each variable.

This plot allows you to view large numbers of correlations quickly, and to see how variables cluster together. You can use these plots to check assumptions, and to reduce data before doing statistical tests or procedures. Fractal foam plots can reveal several possible problems in a data set, such as multicollinearity, violations of assumptions of normality, or correlation structures in the data set that do not work well with a particular type of model. Figure 2 shows you a sample of a fractal foam plot. Labels for the variables clustered around the focus appear only in a space below the diagram as you move the mouse pointer across them - as will be discussed shortly.

Parametric snake plots extend standard scatterplots. In them, the points in the scatterplot are connected in the order of another variable. For instance, you can connect the scatter points in order of a time variable. This allows you to see how relationships between two variables develop with respect to the third.

Quadwise plots give a view of any two scatterplots with linked lines drawn between corresponding points. They therefore show the relationships between two pairs of variables. This can provide extra analytical power when used with Diamond’s flexibility in assigning colors to data based on the numerical range in which the data falls, and in making subsets of data based on these colors.

Parallel coordinates plots look at the relationships across many variables at once. Each variable is plotted on a vertical axis with lines connecting points, case by case. Again, these plots can work by color-coded group. You can sort and rearrange the axes interactively.

Triplewise plots (3-D scatterplots) give a 3-D view of a scatterplot of three variables in a transparent cube. You can animate the display, making it spin and tilt in any direction.

Other capabilities include numerous summary displays, most notably in the program’s Directory window. This window (which needs a more descriptive name) holds a matrix of scatter plots in which each variable is plotted against every other variable. Histograms and cumulative histograms for each variable also appear. You have the option of adding best-fit lines, which you can draw according to an adjustable goodness-of-fit threshold. (For instance, the lines can be set to appear only for variable pairs with an r-value of greater than 0.25.) You can sort the plots, and scroll through them. The large device in the lower corner of Figure 3 controls the scrolling.

Note how the color-coding works on the slopes of the best-fit lines, as well as on the variables displayed themselves. This feature can quickly show the relationships among large sets of variables. This directory window by itself is practically worth the price of admission for the entire program.

Now we move on to your author’s wish (and complaint) list for Diamond, one that’s longer than comparable sections for the other products reviewed. While it should appeal to fans of grousing and random criticism, I approach it with mild trepidation. I hope that SPSS will continue developing this program into the remarkable analytical tool it can become, even though a fair amount of work remains.

Perhaps most vexing are Diamond’s limitations in formatting and modifying output. Most of the displays are strictly fixed in size, and do not fully use the screen at 1,024 by 768 resolution (which is fairly standard for Windows). You cannot zoom in on most of the displays. For instance, the "Directory" window shown in Figure 3 cannot be modified, except for changing the color-coding, or scrolling or sorting the pre-sized rows and columns. In the "fractal foam" display, your only way to find what the bubbles represent is running the mouse across them. You cannot even put labels onto the largest bubbles in the display screen. In addition, you cannot zoom in on this display - a feature it seems to beg to have. It eludes me how this display could get into a report, or a presentation, in an intelligible form - without forcing the user to scratch out labels with a pencil and paper, then apply them with another program.

Unfortunately, the "gee-whiz" aspect of the displays does not carry over well from the screen to paper. Saving the displays in the only format available (standard Windows metafiles, apparently) did not produce high-resolution output, but rather was limited to the pixels (dots) in the rather small display windows. The saved images did not come out in color, in spite of their huge file sizes (the directory window file ran to some 1.3MB). The pictures from Diamond in this review in fact were "finessed," using a dedicated screen-capture application.

For such a visually oriented program, Diamond relies heavily on use of the manual to explain what it does and how. Unusual for SPSS, the manual was fairly diffuse. Pointers for practical applications for the program are scarce. Worse, the authors drag out the dreaded "Fisher’s Iris" example, which must be nearing its 70th birthday, and seems to show as little as possible about what you can do with any modern analytical procedure.

In sum, Diamond is an intriguing application, but needs more work in customizing its data presentation and in its on-screen controls. It likely would be of most interest to you if your goals center on looking at data and understanding it yourself - rather than on presenting your insights to others. This is unfortunate, because as we look at messier data sources, such as customer databases, we always need new and more powerful methods to find and display hidden patterns. With a little more work, Diamond could provide some of the needed tools.

SPSS prices Diamond like its options, meaning that it could cost less if you already have other options, or if you buy others at the same time. If you use SPSS, you should call your SPSS representative to determine the price. As a stand-alone product, Diamond costs $395.

SYSTAT Version 7

SYSTAT and SPSS, for years major competitors, now have joined forces. These two general purpose statistical packages now share a common manufacturer, and have been brought more closely in line in file handling and overall appearance. This gives rise to two key, and highly related, questions about SYSTAT. First, if you already own SPSS, do you need SYSTAT? Second, if you don’t own SPSS, could you use SYSTAT as your only statistical package?

We’ll start with the first question. What does SYSTAT add to the analytical power you already may have in SPSS? After all, if you have the SPSS base product, and a few of the optional modules, you may feel that you have covered all the analytical areas you ever will need to. You may believe that SYSTAT offers largely the same things as those you have. Perhaps SPSS, Inc. plays into this perception somewhat by labeling SYSTAT a "scientific" product, keeping it separate from the "SPSS family." (That is, they seem to suggest something along the lines of, "It’s really like SPSS, but just for people calling themselves scientists.")

While these two programs overlap in many of the basics, SYSTAT provides a wealth of analytical methods that add to those in SPSS. These programs in fact complement each other, in spite of the inevitable duplication involved in each offering a full set of statistical procedures.

The "new feature" list for SYSTAT Version 7.0 will reveal most, but not all, of the important differences between SYSTAT and SPSS. Several of these "new features" will be familiar to some readers as former extra-cost options in earlier versions of SYSTAT. Whatever their provenance, these include some major additions. Here’s the list first, then a brief discussion of some of the procedures likely to be of more interest.

SYSTAT Version 7.0 major new features:

• Bootstrapping
• Probit
• Classification and regression trees
• Test item analysis
• Conjoint analysis
• Set and canonical correlations
• Correspondence analysis
• Signal detection analysis
• Logistic regression
• Survival analysis
• Partially ordered sets
• Two stage least squares
• Perceptual mapping

Bootstrapping is not actually a "module" in SYSTAT. Rather it is a procedure available to use in nearly all the other modules. It allows you to estimate errors in coefficients and other measures that you otherwise could not. Or, as the SYSTAT manual states it, it produces "estimates of parameters in samples taken from unknown probability distributions." (And of course, there are no questions about this, class, are there?)

Most notably, bootstrapping can provide the standard errors of coefficients from multinomial logistic regression. This makes bootstrapping a very useful option for analyzing discrete choice modeling problems. (For the more statistically inclined readers, bootstrapping works better than the [perhaps] more familiar Wald tests for the coefficients of nonlinear models.) You also can use bootstrapping to estimate the standard errors of medians, the standard errors of Spearman correlations, and the standard errors of regression coefficients where predictors are highly intercorrelated. Bootstrapping as implemented by SYSTAT actually includes three related estimating procedures, more correctly called jackknife, simple replacement, and (finally) bootstrap. These procedures determine errors empirically, with calculations based on drawing many subsamples or subsets from the data set. Since you run bootstrapping procedures hundreds, or thousands, of times to get the required estimates, plan to leave a little extra time for it. You probably will want to wait until just before lunch - or better, a few minutes before quitting time - to turn the computer loose on this type of problem, then check back later. (Two or three years from now, though - if computers continue increasing speed as they have - you probably will get this all done while you have a cup of coffee.)

Logistic regression is perhaps the biggest bonus of all "new" SYSTAT modules. This is an update of a formerly extra-cost module (called "Logit"). With this module, you can easily and efficiently do nearly all the analysis needed for discrete choice modeling. No other widely available program now offers a dedicated module designed to do this form of analysis. (I would like to note here that you can force or trick SAS into doing the required analysis, and the SAS Institute Web site provides some remarkable papers by Warren Kuhfeld showing exactly what you need to do. Also, when I stated in an earlier review that SPSS could not do the required multinomial logit analysis, a Very Alert Reader [Keith Crum of IntelliQuest] quickly pointed out that you can in fact cajole the SPSS Cox-Regression procedure to give you an answer in a similar way. For most readers, though, this definitely will remain Something Not to Try at Home, especially because SPSS does not offer program-specific guidelines like the ones from SAS.) Finally, Sawtooth Software also offers the CBC program, which is an all-in-one type of solution, handling everything from setting up the problem to gathering the data and doing the analysis. CBC, though, has a serious limitation in that you cannot analyze any product with more than six attributes. Also, you need to accept CBC’s underlying analytical model, in which brand is treated as an attribute. Some users may not care for this lack of flexibility.

In any event, you can analyze discrete choice modeling problems directly with the SYSTAT logistic regression module. You can handle multinomial logit and related modeling procedures of very large size - so you do not need to worry about running up against limitations in the number of attributes, levels or choices you test. SPSS has trapped and resolved a few bugs that sometimes emerged in earlier versions of this module, and it now appears to run perfectly well.

I have only a few scattered wishes for this module. Principally among these, it does not directly support testing for "Independence of Irrelevant Alternatives" (or IIA). Multinomial logit (and related methods) require that this assumption not be violated, to produce reliable estimates. Doing this type of testing without any automation in the program is both time-consuming and difficult. (Here is a place where I would very much like an Alert Reader to write in with a simple solution.) Fortunately, IIA rarely gets violated in real-world applications, though. The absence of this feature, then, while regrettable, will not critically injure most analyses.

Another wish concerns the manual itself. Uncharacteristically for SYSTAT, the manual seems a little diffuse. Even very experienced readers may find the descriptions somewhat unclear, in terms of what you need to do to structure the data for each of the logit variations. The manual really needs very explicit instructions, in brief (even bulleted) form, on how to make sure the data is set up properly. You can figure this out with some experimentation now, but the manual could go further in explaining this key area. Also, you will find no discussion that helps in setting up and using the procedure as it is most often applied in marketing and market research.

Doing market-share simulations remains time-consuming and cumbersome. You are really better off building your own simulator that makes use of the output, or finding somebody who knows how to do this.

In spite of these reservations, this module can truly add crucial analytical power to the procedure you can find in SPSS and other widely available PC-based packages.

Classification and Regression Trees also is an update of another former extra-cost SYSTAT module: the familiar (perhaps) CART program. This program has many excellent features for constructing and testing classification trees, using both CHAID and CART (or as SYSTAT now calls it, "C&RT"). Unfortunately, this sophistication in developing models apparently brings a limitation: the program only can make two-way splits in the tree diagram. (More formally, it can do only "bifurcation.") At the risk of disputing a point with Leland Wilkinson (the driving force behind SYSTAT, and doubtless one of the 50 smartest people in the Western hemisphere), building classification trees based only on two-way splits cannot be exactly equivalent to building trees in which many-way splits are possible. Perhaps most important, correct statistical testing for many-way splits must use different methods from the testing used for two-way splits. (Yes, I have a citation. If you care and want to know, please contact me.) The upshot of all this is that, in my experience, classification trees based solely on two-way splitting at times may obscure, or even miss, some useful patterns in the data. (Now that I’ve said this, you may fire at will.)

Please note, though, that the type of tree-building used in this module has become almost mandatory in certain types of medical research, where the goals are somewhat different from those related to marketing. Finally, classification tree analysis provides so many strong benefits, in both understanding and analysis of information, that anybody who works with data really needs to have some program that does this competently. For many, the SYSTAT module will more than meet all their needs.

Conjoint Analysis is perhaps not quite what you would expect, and certainly is different from the conjoint module in SPSS. This is actually a general-purpose modeling program that will fit additive models to data that you cannot measure with more specialized conjoint models. You can, for instance, fit trade-off models to data that does not come from experimental designs. This program can address the question of whether this type of model could fit, once you have data that was not collected with a standard conjoint procedure. As such, it could work as a useful supplement to the standard conjoint methods more familiar to some readers.

The various mapping procedures in Perceptual Mapping and Correspondence Analysis include a wide variety of procedures for representing and plotting data multidimensionally. If you already have the SPSS Categories module, you still should find some new approaches and features in SYSTAT. In addition, the discussions of mapping in the SYSTAT manual go into greater depth then those in the fairly slender SPSS Categories booklet.

Most of the other new procedures either seem highly similar to those in SPSS, or just seem to get fairly sporadic use in market research and allied fields. (I’ll apologize in advance if I just insulted your personal pet procedure.) Some of these methods look like they could get more application. For instance, signal detection, used mainly in psychological experiments, might be a useful analytical tool where you have an overall rating, an actual response, and specific rating items. If any of you have explored this possibility already, it might make an interesting article for this very column in another issue.

Now, we will move, at last, to the second question. Could you use SYSTAT as your only analytical program? Certainly, SYSTAT offers a wide enough range of features to cover many needs. In addition to the "new" features, SYSTAT has all of these analytical methods:

• Analysis of variance
• Linear regression
• Correlations, distances, and similarities
• Loglinear models
• Descriptive statistics
• Multidimensional scaling
• Design of experiments
• Nonlinear regression
• Discriminant analysis
• Nonparametric tests
• Factor analysis
• Path analysis (structural equation modeling)
• Frequencies and cross tabulation
• T-tests for means
• General linear models
• Time series

Note in particular that the SYSTAT path analysis (structural equation) module included in the program has plenty of features and power. It even has a name of its own: RAMONA. (RAMONA is another one of those long acronyms, standing for "reticular action model or near approximation." Anyhow, it does structural equations. Now you know.) RAMONA does not have the graphical Windows-type interface, and consequent ease of generating output, that you will find in the SPSS add-on module, AMOS. (AMOS, another happy acronym, stands for analysis of moment structures). Analytically, though, you should find RAMONA thoroughly top-notch.

The differences between SYSTAT’s RAMONA and SPSS’ AMOS exemplify the differences between the two main programs. SYSTAT remains somewhat more command-oriented than SPSS, meaning you may type a little more, and probably look in the manual more if you are not familiar with parts of the program. SYSTAT’s command language is logical, concise, and easy to use, but if you have not seen it yet, or have not used it for a while, you will need to take a few minutes to learn (or re-learn) its workings.

SYSTAT’s analytical output is still text, not the nicer-looking "table objects" you get with most procedures in SPSS. This also means you cannot paste the output from SYSTAT directly into a spreadsheet program - where it neatly becomes cells you can manipulate or format - as you can with SPSS output. You usually also get somewhat more control over the contents of the output with SPSS, although most of the larger SYSTAT modules have gone well beyond the old days, when you could ask only for "short" or "long" (and sometimes "medium.") SYSTAT’s charting and graphing capabilities have always been excellent, though, and remain so in this release. Now many charting features are available from handy menus and check-boxes, as well.

SYSTAT has one advantage over SPSS: price. Before we discuss this, you should recall that statistical analysis programs, selling as they do to a small, select audience (i.e., you), cost quite a bit more than programs like Excel and Word Pro. That said, you get everything in SYSTAT for just under $1,000, which is quite a bit less than assembling the comparable modules in SPSS.

I have only one major request for SYSTAT. I wish the program would incorporate "progress meters" into its display screen, similar to those that appear in SPSS. These are little displays letting you know how far the program has gone in the analysis at hand. Especially if you are running computationally intensive procedures, like logistic regression, you want to know that the PC is busily working on the problem - and not frozen.

Overall, then, if you need the latest developments in Windows-related features, including output, displays, and menu-based controls, you still will want SPSS as your main statistical analysis program. If you use the program only occasionally, and want to get up to speed quickly, then SPSS again is the logical choice. If you are a more experienced user, and want the absolute maximum in analytical bang for the buck, then SYSTAT would be the better product for your needs. As suggested earlier, though, if you want a truly remarkable range of analytical tools, and your budget can tolerate it, go for both.

Scarcely go wrong

If any of these programs meets your analytical needs, you scarcely can go wrong by choosing it. As a reviewer of software, I can tell you that it is very gratifying to conclude a review in this way. Not too long ago, computer programs in general did far fewer things than you hoped they would, and many others that you wished they would not.

In summary, then: DBMS/COPY now has become about as speedy, comprehensive, and powerful a file-translation program as you can find. DeltaGraph provides tremendous charting and graphing capabilities, and remains my program of choice for times when programs like Excel or PowerPoint just don’t have enough power to get the job done. SPSS Diamond offers unique data visualization and exploration capabilities. It needs some work to become more useful for presenting insights (as well as finding them), and I hope SPSS will continue developing this very promising program. SYSTAT provides a tremendous range and depth of analytical procedures, and remarkable charting and graphing abilities. Even users of SPSS will find many useful features in this program adding to those they now have. All of these programs could make highly useful additions to your analytical armamentarium.

You can get more information on SPSS products at their Web site: www.spss.com. You will find more in-depth information on DBMS/COPY at www.conceptual.com.