The numbers game

In a month of steady advertising, Colin Fallon warns that you shouldn’t believe all the stats you see – not even these ones

Last month I predicted the possibility of a spike in advertising as we head towards the end of the year. Sadly, it hasn’t happened – but that’s no reason to start crying into the pages of Gun Trade News just yet. We’re around six pages down on last month’s total, but since that’s less than 2 per cent in relative terms, it’s safe to say the industry has advertised consistently month on month. A rise before the end of the year is – without trying to sound too much like a broken record – still fairly likely.

That’s not entirely speculation, either – looking deeper into this month’s figures, even though July and August issues still abound, there are already signs that winter is on the way. Foremost is the huge growth the clothing sector has seen this month – at 24 pages, or 7 per cent of the total, it’s nearly quadrupled in size in the space of a month. We’ve seen the return of seasonal advertisers such as Seeland, and big distributors choosing to give their clothing brands time in the limelight (such as Garlands, which advertised the Rocky brand over several pages, and Highland Outdoors with Ridgeline). We aren’t seeing many clothing advertisers trouble the top 10 yet – compared with last year when for a short while they were fi ghting for the top spot – but that could be a sign there’s even more growth to come in this sector.

No particular sector has seen a big drop to counter the growth in clothing, either, with most other categories remaining steady from last month or posting very small drops. The biggest drop was in the classifieds, which have gone down from 15 to 12 percent of the total (or around 11 pages). Since looking for meaningful patterns in classifieds sections is too small a needle in too large a haystack for even this column to contemplate just yet, we’ll make the executive decision to put that down to the vagaries of statistical distributions for now, and expect the classifieds to rebound in coming months.

What of the wider, flatter advertising market, which I devoted signifi cant column space to discussing last month? At first glance it seems to have just been a flash in a pan, with one of the ‘top 10’ charts actually having 10 companies in it and the other 11 (in contrast to last month’s 13 and 14). But I’d argue the pattern is still there, just a little harder to spot. That’s especially true in the retailer charts, where the market clearly flattens out at the two-page mark – just like last month. In the supplier chart things look signifi cantly linear – but in compiling the data I found that things get a lot more equal just below the top 10, with a host of companies just missing out with 4.5 and 4 pages. These are businesses that could easily make the charts any other month, like RUAG, Garlands and Mossy Oak. So I’d say that if the market isn’t exactly a ‘flatter’ one, it is at least a very competitive one. Which many would say is good news.

Lies, damned lies

Statistics is the science of uncertainty. Regular readers of this column will be familiar with the monotonous regularity with which I warn that these figures on their own aren’t statistically significant – you need many months’ data together before concrete claims can really be made.

But if you think that’s bad, be warned: there are many ways – ways often seen in the mainstream press and even scientific journals – in which statistics can be not just uncertain but downright wrong. And if enough people believe these wrong statistics, that can start to influence public opinion and even government policy.

As a hypothetical example, let’s say you see the headline ‘Gun Crime Doubles’ in a newspaper. Putting aside the fact that firearms offences are continuously dropping in this country, it seems like the sort of headline you might see, and the sort that people should also sit up and take notice of. Gun crime doubling? That’s a huge rise, right? See also: ‘Use of new drug doubles’, ‘Miracle cure 100 per cent more effective than before’, and so on.

These declarations sound impressive, but they’re entirely relative – there’s no clue as to the absolute numbers involved. If that miracle cure was only effective once in a million times, and now is effective twice in a million, it’s still not very effective. It’s a similar story if gun crime of drug use was very low to start with – it’s not really a meaningful increase, but it can be twisted to make a good headline. If you see ‘100 per cent increase’ in a headline, be wary.

Here’s another phrase you see an awful lot: “Fastest growing”. If you look hard enough, you can see it used in articles declaring anything from Islam being the fastest-growing religion to ultimate Frisbee being the fastest-growing sport. But this, too, suffers from a base-rate error, since the rate of growth is measured entirely by percentages. If I start my own shooters’ association, then my mate joins, it’s just grown by 100 per cent, making it technically the fastest growing organisation out there – but that’s hardly worth reporting.

That second fallacy only becomes useful fact if you can predict the future. If my shooting association continues to increase by 100 per cent every day, it actually will be the biggest organisation in a matter of weeks. But will it actually grow at that rate? You wouldn’t bet on it, I’m pretty sure.

That’s where the association to this very column comes in – Industry Data is forever uncovering trends only to stop shy of forecasting that they will continue. That’s because I, just like everyone else, can’t predict the future. To do so, even with all the data in the world, would be a fallacy. Advertising has dropped very slightly this month – would I bet that this drop will continue? No, I wouldn’t – I’d be more inclined to bet the opposite..

Posted in Features
Follow Us!