B2B International
B2B International

January 23, 2009

In his first Thursday Night Insight, David Ward – B2B International’s Head of Data Processing and resident number cruncher – warns us of the dangers of taking statistics at face value.

Numbers and statistics are quoted at us all the time but how often do we stop to think about what they really mean before we’re drawn in by them? Outside a farm shop a mile or so from where I live, I saw a sign which was advertising fresh milk for sale. This was no ordinary milk. In fact it was very special 96% fat free milk. For a split second I thought about it, was impressed, and then realised it was effectively ‘full fat’ milk. The way in which the information was presented made all the difference. The impact of the sign was as much about what it didn’t say as what it did.

The same can be said of the stores on the high street that are trying their hardest to get us to part with our money at the moment, and sales signs are in every store (or at least it seems that way). I’ve seen many signs offering up to 50% off, but what does this really mean?

Perhaps this is a little cynical but an alternative sales sign could read…

“Up to 50% off items in store, well to be honest most items have 15% off. It’s the 4 items that nobody wants and that we’ve been trying to sell for 6 months that have 50% off”.

Judging by a story in my local newspaper, it seems that some of the shoppers hoping to pick up a bargain at the closing Woolworths store left with thoughts not too dissimilar from my alternative sign. No one can deny that 50% off is a very tempting offer on the face of it, but my alternative sign may be more representative of the reality of what awaits the shopper.

There is a serious point here and it is that numbers can be made to say or infer almost anything you like. In market research, different messages can be shown depending on how the numbers are presented and indeed which numbers are presented. Even something as seemingly simple as indicating a typical respondent or trend isn’t as straight forward as is often assumed. Is it better to depict the “typical” respondent with the mean or median? How often are we told what the average person thinks without being told whether that figure has been taken from the mean or median?

Working in data processing, I spend a large proportion of my day working with numbers, whether it is column locations, frequency counts or checking that the figures in the tables are presented well. Not only do I put the effort into getting the raw data into table format, I spend equally as much effort in making sure those figures are correct. I want to be confident that this effort is not wasted and that the figures are being used in the most appropriate way.

Just like the sales signs I saw in the Woolworths shop windows, reading something into the figures that isn’t really there is all too easy.The ability to sift through the sea of information that a quantitative project can generate, be able to pick out what is relevant and interesting, and be able to apply the correct conclusions based on those numbers, is a real skill. It is an important skill in market research, and one that should not be underestimated.