Managers’ comfort with statistical information varies greatly. When shown a chart like the one on the right, they either know what questions to ask themselves, or they skip over it.

In this chart, the blue line shows the performance of a firm over the last year, and the other lines try to explain the data. Most corporate communications managers have ‘share of voice’ data that look like that blue line: it shows how their brand name is rising and falling in profile.

The more they understand statistics, the better they can understand what’s going on. A superficial look at the blue line makes it look like it’s been a bad year. If you simply look for the trend, then you get the black line – Excel’s linear trend line shows a downward trend, and it even forecasts for things to get worse. Few managers would produce a trend line on this sort of data and, indeed, few would even add the pink line, which shows the average value at each point in time. This also suggests a downward trend is still continuing. However, a weighted average – which puts more weight on more recent data, shows the more accurate interpretation that the trend is stable, and looks positive.

There’s a lot of seasonality in data like these as well: many businesses have recurrent annual and quarterly cycles that are reflected in both their stock price and in their share of voice.

This diagram shows one way of allowing managers to see beyond that seasonal data. You can use time series analysis to identify underlying trends. By subtracting that trend line from the current data you are collecting, you can see better what the situation really is.

These statistical calculations are pretty simple (especially with DecisionTools). But statistics are often avoided.

Personally, the most rewarding part of my job here at Lighthouse is helping AR managers to use statistical information to become more effective.

Statistics do not cure every problem, but I really want AR managers to be able to benefit from descriptive statistics. It’s been a major part of my working life. Before being an analyst at Ovum, I built Clementine and Business Objects systems. I wrote my masters’ thesis on data mining, and worked for City University researching that topic. I turned down an technical role at SPSS, the statistical software firm, to join Ovum. Perhaps because of this background, I still get surprised when I see people avoiding and even attacking the use of statistics.

There’s more avoidance that you’d think, especially in organisations that fight shy of metrics in case they uncover bad news. People sometimes get defensive and dismissive: they appeal to fear. My experience has taught me that without losing their fear of statistics, managers will live in fear of them. There are too many people who regard unfamiliar terms like ‘running average’, ‘weighting’ and ‘smoothing’ as mysterious snake-oil (and this is greedy reductionism, not just FUD). These fearful views are geographically concentrated, because there are some especially strong cross-cultural differences in numeracy education, as shown by the OECD’s PISA studies.

This trend is especially difficult in the United States. In the 2006 PISA study, it ranked 28th of 40 developed countries in math. The New York Times reports in this article this article that the gap is widening between countries with higher numeracy and those with lower numeracy. However, this is not widely recognised. Of the United States students surveyed by PISA, 72 percent said they got good grades in mathematics, more than in any other country. While the first step in solving a problem is recognising it, it seems that the American education system is living in denial. Ironically, it is in South Korea and Hong Kong, where students are more numerate, that students are most likely to describe themselves as not being good at math.

This background is reflected in the different preferences that analyst relations managers have for differing statistical tools that underpin the various AR measurement approaches. Managers are uncomfortable with what they don’t understand. However, this means that managers have the choice between either learning about statistics, or living in fear of them.

some commercial tools make the job of data mining and statistical analysis much simpler. check out the free webcast on Oracle Data Mining on Wed March 21, 2007 at http:/OracleBIWA.org

Thanks

Shyam Varan Nath

http://OracleBIWA.blogspot.com

A correspondant points out that the kind of calculations used in volumetrics are also useful with multi-dimensional seasonal data. http://www.cadtutor.net/acad/ktf/volume/volume.html