SPSS Dimensions Desktop Reporter 4.1 reviewed

Tim Macer puts the latest version of the Windows-based cross-tab reporting tool through its paces

There are some old programs that refuse to die. Even the original authors of Quanvert must be taken aback by its astonishing longevity. So must SPSS, its current owners, who had lined up mrTables, its new online cross-tab tool, to be a Quanvert killer. But the convenience of running an application at full tilt on your desktop or laptop continued to give Quanvert the upper hand for many users. Despite a gnarly old interface, Quanvert did many things that were hard, if not impossible to do in mrTables.

Now, SPSS has launched a second potential Quanvert-buster on the market. It is called Desktop Reporter, thus wearing its USP on its sleeve – and it is clear that this time SPSS are serious about providing the critical mass of functionality that should entice users away from old Q for ever.

Like mrTables, it is a Dimensions product, which means that it sits on top of SPSS’ tiered architecture of a standardised Data Model and another model relating to the creation of tables called the Tables Object Model, or TOM. The immediate benefits to end-users of this are not always easy to see, but it does make it a lot easer to integrate Dimensions software with other applications, or even use these products as a springboard to create your own analytical systems, information portals and so on.

Indeed, SPSS has tended to take an engineer’s rather than a designer’s approach to much of its Dimensions line. The user experience was often disjointed, leaving you with the feeling that it was not as easy as it should be, or could be in some of its rivals’ offerings. Desktop Reporter looks to me like a break with the recent past – it’s elegant, sophisticated, sassy and, best of all, highly intuitive.

The main screen has the now typical column on the left where selections take place, a large pane on the right, where the action happens, and buttons and controls on top to effect actions, change options and so on. Right-clicking always seems to bring up a sensible menu of options relevant to what you are doing, and these often duplicate buttons, menu options and keyboard shortcuts, so providing power users with many ways to skip through producing tables.

Some users like to handcraft all their analysis, question by question. A table is simply created by dragging and dropping items on to it. A separate tab in the main window lets you set up filters using parallel techniques.

But it is just as easy to throw all of your questions into the pot, and let Desktop Reporter produce one tab for each one, as a total or with a standard break, if you prefer.

Existing tables can be used as templates for new tables, to speed up the definition. Stats and sig. tests are also easily applied, and instead of overwhelming users with options (most of which 99% of users will never use), more obscure options are hidden away, but accessible generally from a “More…” button. Your default output can be a cross tab or a chart, or both; charts and tables are easily posted into Excel or PowerPoint, and both can retain links to the data for automated refreshing when more data arrives.

It even offers some multivariate analysis for the statistically challenged – letting you present a table of ‘difference attributes’ which lets you throw together any number of variables, and it will rank the combinations of answers to show those where the disparity is greatest at the top. You can alter the parameters to show affinity too.

There is much more that the program does very well – sensible defaults, right mouse button menus and tight navigation mean that whether you are exploring your data or finessing your output, a couple of mouse clicks is usually sufficient to get you where you want to be.

Quanvert users are likely to take to the program straight away. It appears to do much of what Quanvert did, with the benefit of a modern graphical user interface. Even hierarchical data are supported. Some Quanvert terminology, like Axes and Levels has come over too, which may be a puzzle to other users.

There are some constraints in the version I saw, however. The whole business of getting data in is a work-in-progress, as Jeff Thompson, our interviewee, points out. SPSS promise that version 4.5, due out this month, will remedy much of this. It probably will not remedy some of the other constraints that the Tables Object Model imposes on what tables look like, so that some of the manipulation features (often scripted in Quantum) are unachievable in this, and are likely to remain so.

An opportunity missed is to update significance testing. A faithful reproduction of the letter-code sig test tables from Quanvert will satisfy some users, but leave many more baffled – those who are used to looking at colour-coded tables based on exceptions. Unlike most of its rivals, this is something Desktop Reporter won’t let you do.

Despite its version 4 number, this is effectively the first version of Desktop Reporter (the number reflects the generation of the Dimensions suite) and, as a first version, it is most impressive. Will it turn into a real Quantum-buster we must wait and see what the loyal Quanvert users have to say.

Technology in action: Kantar’s early days report
Kantar Operations provides operational support for Kantar Group research companies, and is an early adopter of SPSS Desktop Reporter. Jeff Thompson, Director of Research Technology, based in Austin, Texas, describes his experiences.

“We are at the beginning of deployment, so it is not on everyone’s desktop yet – we have only used it with a few select teams. But we have now made it our preferred tool for delivering SPSS Dimensions-based output. We are already using it to work on Quanvert data. We did a really thorough gap analysis between this tool, Quanvert and some other in-house tools, and we found it really did do everything we wanted it to. There are a some limitations in this version, but 4.5, which is due anytime now, will probably address a number of these.

We have found that the ease of use of Desktop Reporter is so much greater than that of Quanvert. The user interface design is so much better. What we are seeing, overall, is that the number of things that people can do themselves is increased, so the number of requests to DP is diminished.

“For instance, it is so much easier for users to create nets, to create and edit filters and they can derive new variables very easily. These are often things that users had to come back to DP team for changes. It has some interesting new ways of looking at data too. It allows you to select a whole set of variables and run a report, and look for things that are statistically significant. But for us, the real advantage is the ability it provides for researchers to do things on their own and not have to come back to the operational teams to script new variables or tables.

“We are in very early stages and have only used it with a few select teams. They have been impressed, but these people are the ‘early adopter’ types who are often able to get the most out of new tools. We have yet to see if users as a whole will use it in the same manner.

In the past, in the Dimensions side of SPSS, there have been some weak user interfaces. The improvement in this over mrTables is enormous – it stands head and shoulders above anything the Dimensions team have produced to date. We have long felt they had good engines behind their software, but the interfaces were poorly, or perhaps quickly thought out. This interface seems to have been done really well, and I am really hopeful that we are seeing the start of a new age in their development – one where they put more thought into the user interface design of their tools.

The one awkward area is still there is not an ideal data store. You can easily hit Quanvert data but you cannot easily write out Quanvert; you can use XML data, but it does not perform well, or you can use SQL server, which performs quite well, but it just isn’t portable. The tools around the whole distribution of SQL server databases are lacking somewhat. We are expecting this to go away with 4.5.”

The verdict
SPSS Desktop Reporter 4.1
Windows-based cross-tabular reporting program for end-users which offers;

Ease of use – 4.5 out of 5
Cross-platform compatibility – 4 out of 5
Value for money – 2.5 out of 5

Cost
$1,500 (approx £750 ) annually for single user annual licence. Perpetual licence $2,450 (approx £1225 ) for a single user, plus 20% annually for support and updates. Volume discounts are offered.

Pros

• Powerful, but still simple to use

• Supports hierarchical data

• Outputs in Excel and PowerPoint can be auto-refreshed from live data links

• Will work directly on Quanvert databases

Cons

• Cannot colour-code or highlight variances or significant values

• Getting data in can be a challenge (in the current version)

• Some performance issues with large datasets in XML or SQL

Further info
www.spss.com/desktop_reporter

June | 2007