Exploration for beginners

Tim Macer reviews two products that allow the non-statistician to explore data in greater depth.

Espri 3.3
For most people involved in research, there are cross-tabs and there are statistics packages and you normally progress from one to the other. Cross-tab packages struggle to produce stats, and stats packages struggle with cross-tabs. Espri, the data exploration and analysis tool from New Zealand specialists Information Tools, takes the best of both approaches to offer a ‘third way’ in data analysis.
Espri was ahead of its time when Interface last reviewed it in 1996, by putting clever data visualisation and reduction tools into the hands of researchers. Now, in Espri 3.3, the program has had a welcome makeover - more charts, new options, some rationalisation - but it is not dramatically different. Inexplicably, seven years on, it is still in a class of its own.
Gone is the separate selection of techniques and variables. Now, you start each round of investigation by selecting questions to go into a data cube, then viewing this cube for confirmation. Colour-coded significance testing based on Chi2 makes values above or below the expected total stand out. Just a few mouse clicks away are unfussy options to control the significance level and other options. Where others would label the pop-up window ‘presentation options’ or some other bland technical term, in Espri, the window is entitled ‘Show me…’ This approachable style is repeated throughout to reinforce the program’s intuitive qualities.
However, it is all too easy to see Espri as just another 3D interactive cross-tab tool. Here the similarity with other programs ends as the data cube is simply a handy springboard for the more interesting stuff. Pick any column or any group of columns and apply the ‘what’s different’ tool, and the sig, tests will be recalculated to show how your subgroup differs from the rest. It is perfect for time-series data, but is equally applicable when segmenting your data. This is just one of many statistically-based discovery techniques.
Where other segmentation tools may leave you guessing as to the exact make-up of your clusters or selected segments and whether they have any statistical validity. Espri will rank all the other criteria in your data cube against your target segment using correspondence analysis. Very easily, you can produce a very long list of attributes ranked on a scale of most like to most unlike the group. At the top are those characteristics that strongly define the segment; at the bottom are all the telling affinities the group lacks.

Espri 3.3 & Harmoni 1.2

At a glance:
Intuitive survey data exploration suite that makes statistical analysis and interpretation of survey data accessible and meaningful to the non-statistician. Produces a wide variety of cross-tabs, charts, statistical measures from survey data, and in Harmoni, combines this with raw or aggregated data from other sources too.

  • Easy, intuitive and non-technical interface
  • Advanced statistical techniques for the lay user
  • Powerful “what’s different” and other trend spotting features
  • Saved tables and charts are updated automatically with new data
  • Limited report automation features
  • Uses a proprietary database

One of a kind. For sig tested tables, consider Pulsar, U-Tabs, StatXP; for data visualisation, consider SPSS or BrandMap

Harmoni 1.2:
Espri’s younger sister
Harmoni, with an overlapping set of features in common with Espri, extends the capabilities to combining and comparing data from different data sources. You can add either raw or aggregated data by populating the cells of an Excel spreadsheet with the values you wish to import. No specialist help is required.
Both programs offer extensive charting: the usual culprits plus great correspondence maps for that a qual-type feel to quant analysis. There is a handy zoom tool to help you see meaningful detail even if you ambitiously plonked every variable into your map.
Results dissemination is also well provided for, with good integration with MS Office products, and a new proprietary Cube Viewer, available under separate licence. Any cube, charts or reports associated with it can simply be cut and pasted into Cube Viewer then emailed to anyone you choose. Recipients can slice and dice, but do not have recourse to the raw data.
If it falls short anywhere, it is in its range of tools to aid repetitive tasks and report automation. While data get refreshed automatically - which is good

- it lacks any macro or iterative functions. Neither, in its current guise, is it web enabled, and it uses a proprietary database which you normally commission Information Tools to build for each project.
Catherine Malcolm, an Espri user in her current role as MR Manager for Norske Shell in Oslo comments: “I use it for everything, as a profiling tool, for interrogating the data we have, and even to track activities. Furthermore, whenever someone asks you a question, it’s easy get to the answer. The program has a very solid feel - as if you can’t go wrong with it. By having all the different options visible, it makes it easy to try one thing then if that doesn’t work, try something else.” Malcolm described the significance tests as ‘a godsend’ along with the ‘what’s different’ feature, which she also uses a lot. She notes “The good thing is, you don’t really have to know anything about statistics to use Espri. You just need to think.”

Information Tools:

Tim Macer is MD of Meaning Ltd. He writes as an independent specialist and advisor on software for market research. His website is at www.meaning.uk.com

Published in Research, the magazine of the Market Research Society, April 2003, Issue 443.
© Copyright Tim Macer/Market Research Society 2002. All rights reserved. Reproduced with permission.
top of page | Back to list of reviews