Lies, damned lies, and statistics

I first heard about what the industry is calling "the world`s first ever Web census", aimed at profiling South African Internet users, at First Tuesday. My initial reaction was to wonder what everyone was talking about.

And yes, I responded twice - my actions may skew the results, but that`s what you get for using cookies.

Basheera Khan, Journalist, ITWeb

You see, despite the organisers` efforts to post the census on more than 50 local Web sites, I`m ashamed to say, I hadn`t got wind of it at all until someone dressed in a Spiderman costume handed me a notebook at First Tuesday, emblazoned with the words: I took part in the world`s first ever Web census and all I got was this lousy notepad.

Deeply original, and, I`m almost certain, slightly inaccurate. After all, considering the Internet has been around for some time, I`m convinced that someone, somewhere has done this before. I recall filling in all sorts of online surveys for various reasons, and passing on almost exactly the same information requested by the Web census survey.

I`d have facts to back up this belief if our Internet connection hadn`t been the victim of a spectacular technology failure over the weekend, but as the song goes, you can`t always get what you want.

The information asked this time around included location, gender, age, and the type of car one drives. In addition, job level, radio and TV stations of preference were queried, as well as whether one has a credit card, when last one purchased online, the number of hours one spends online each week, and how often one travels by air.

There were other more interesting questions too - such as the age of your children, whether you are responsible for the household shopping, what you`ve done online in the last month, and whether there is anything specific stopping you from starting your own business. And of course, permission to use the information supplied in ad or content serving.

So what was it all in aid of? And what confuses me more, why is it that some of the questions I was presented with this morning (naughty, naughty 5fm - keeping the survey up even though it was only meant to run until yesterday) don`t spark any trace of recollection in comparison to the questions I was asked on Friday, when filling in the questionnaire for the first time.

And yes, I responded twice - my actions may skew the results, but that`s what you get for using cookies, don`t you know.

At the end of it all, I`m still uncertain as to how seriously I should take results returned by the Web Census. A statement on the Web Census site (www.webcensus.co.za) says that it aims to record and thoroughly validate demographic, psychographic and other useful information on a total market as well as a site-by-site basis. Admirable enough, but how exactly does it intend to do this, with no real way of contacting the survey respondents?

I generally view statistics with an extremely jaundiced eye. I can`t help thinking that while they may be useful in certain applications, numbers on their own mean very little.

I wonder how the Web sites involved in the Web Census will use the findings to improve on their offerings. I wonder whether some form of the results will ever reach those who bothered to fill it in. Most of all, I wonder how events during the course of the upcoming year will affect the tone and angle of the questions asked next year - should it prove worthwhile this year. I await the results with considerable interest; watch this space.