Irregular Shed

Preamble du jour

Anybody who knows me even the tiniest fraction knows I like maths and physics. I make no attempt to hide it (and why would I?); I’m very much an ‘out’ geek. This is the t-shirt I wore today. And anybody who has dipped into my posts online over the past, ooh, nearly five years knows how much I hate it when maths is used badly in PR campaigns disguised as research, so as to get free advertising for a client when lazy journalists churn the PR pieces to fill column inches.

The most egregious of them all, and the one that started me off on my quest against faux formulae for advertising, is what has ended up being known as Blue Monday. My feelings about “the officially most depressing day of the year” are very well known. The only good thing about this supposed formula is that, thanks to mutual feather ruffling, I’ve made a couple of allies in science: very occasionally the darling of the tank-top Dr Ben Goldacre, but much more frequently the very lovely Dr Petra Boynton.

Petra recently pointed me at a mathsy PR mess that she was concerned by. In football terms, it was very much a Division 1 relegation zone affair, for Global Brands’ unpleasant booze, Vodka Kick. It had passed me by completely, which is good as far as I’m concerned because it meant it had also passed by the Premiership and Championship quality news outlets, and had only been picked up by Division 1 news websites - local papers and lazy radio DJs, that sort of thing. Of course, I’m not the sort of person who finds getting arseholed on artificially flavoured, watered-down ethanol appealing, so it may have been all over whatever news source the proletariat follow…

Anyway, Petra had serious concerns about the research that was being presented by the PR company, and told me she was working through that side of things for her blog. Would I be interested in taking a look at maths to see how figures worked out? Why yes, of course I would! Her comrade Tristan O’Dwyer was also looking at the same PR piece for his blog, so the three of us started comparing notes to pick through the whole debacle, using our own unique super-powers to analyse what was going on. Me, I’ve been on maths detail, and that’s what I’ll concentrate on here, but first some background…

The promised background

To flog flavoured vodka drinks to twenty-somethings, Global Brands set up a competitiony thing on Facebook-type things to find a young, “fun-loving” maths or science student to research nights out, to produce a “scientific” “thesis” on what makes a night out “perfect”.

Obviously, with this all being in the name of PR, munters need not apply. They didn’t actually say that (althoughh kudos would be due if they were that honest), but the way the cards were played made this obvious. The winner would be a moderately attractive woman; not “out of your league” material as far as men would be concerned, and the sort of woman who looks like she could be in your group of friends if you’re a woman. It wouldn’t be a man because if they were good looking they’d end up making potential male customers feel inadequate. An ordinary looking man would work if they were a bit funny, but VK is a sophisticated drink as well as a fun-loving one, so being entertainingly funny wouldn’t work.

What a surprise it was, then, that this young lady won; seen here posing with a formula that is all her own work, honest. (More on that later…)

As with Petra, I’m not going to name her (it’s not hard to find, though) in case she realises the travesty of it all.

Now, as is traditional with these PR formulae, the formula that is presented is poorly rendered. I’ve taken it upon myself to try and improve the appearance by using OpenOffice.org Formula to get something a bit more likely to gain a grade C GCSE. Here’s what has been shown off by the PR gang:

Here, your Perfect Night Out (PNO) is equal to your attractiveness+spontaneity factors, multiplied by the sum of your number of friends factor, venue factor, timing factor and fun time factor, all multiplied by an “end of the night factor”.

The PR company say this is all the student’s own work, but I find that suspicious. As she’s a biology undergraduate I’d expect her to be able to write a formula accurately, and that means no multiplication symbols and single character variable names. PNO is PxNxO, and N is used on the right hand side.

Looking at the promotion’s webpage, we can see that the whole formula is based on subjective feelings from leading questions, and any attempt to use actual units (like those of time, or the number of people or places) are thrown out the window in favour of a numerical scale from 1 to 10. At least, that’s how it appears - in their calculator they have eight variables, compared to the seven in the formula. Also, their calculator just adds your numbers together giving a total out of 80, instead of multiplying three parts of it which should, using the scales suggested, be out of 16000. (Of course, we can’t know that this is the case because there’s no example figures shown in the description of the formula.)

However, Tristan obtained the ‘thesis’ written by the student from the PR company, and in the interest of peer review (and seeing as it had to, and I quote, stack up to academic scrutiny by Britain’s best brains), we’ve got it hosted here for all to see. I recommend you have a look at it before I start picking holes and showing how easy it is to have infinite fun!

Reading through the paper was interesting for me, from a forensic point of view. It looks an awful lot like there are two distinct writing styles in certain places that would suggest, to me (and this is only an opinion) that the bulk was written by one person and then someone else chimed in to expand. I won’t dwell on this though, there’s no physical (or electronic) proof, but certain sections show a woeful lack of understanding of fairly simple maths.

Outlined in the paper are formulae used as the foundation of the formula above. A couple of them have really basic, obvious flaws that I would expect anybody tasked to draw them up to spot before even committing them to paper; they’re listed below. For them to be the foundations of any research on any topic would be pointless, and we can only be glad that it’s for a pointless exercise like a perfect night out, and not for something important like, I dunno, the amount of gas required to inflate a car’s air bag. Once again, I’ve used OpenOffice.org Formula to render them accurately.

The number of friends to go out with

Here, N is the magic number that represents how you feel being with your friends. W_K is a weighting factor, provided as a black box that allows wiggle room (“Formula doesn’t work for you? Clearly you’ve misinterpreted your weighting factors!”). L_L is the lower limit of the number of friends you like to go out with, U_L is the upper limit, and F is the number of friends you’re with actually out with on this occasion.

Looks impressive doesn’t it? The problem is, if you only like going out with, say, three other people - no more and no less - then you end up dividing by zero, resulting in infinity. Hey presto, infinitely good night out! This happens whenever you have two identical limits, so my favourite result is infinite perfection by being a complete loner.

The timing formula

Here, T is a numeric representation of timing - as a factor, not as a time, so no units. W_T is a magical weighting factor, D is a day idealness factor (D=1 on your favourite day to go out, 0.5 for any other day). S_Diff is the difference between your ideal start time for a night out and the actual time you start, and F_Diff is the difference between finish time.

This formula doesn’t stand up to scrutiny for several reasons. Firstly, there’s no discussion in the report regarding the units of time; in science normally this is done in seconds, but given that there’s 0.25 added (“to allow for inaccuracies”) I suspect this should be in hours - 15 minutes sounds like a more realistic amount of wiggle room than 15 seconds or 250 milliseconds. However, elsewhere in the formulae minutes are used - so there’s no consistency.

Secondly, units need to be included in scientific formulae. 7 seconds +7 is meaningless, either both factors have the same unit or no units at all. Those 0.25 wiggles require a unit identical to those of the differences. If you add them in, and let’s (to keep a pretence of science) assume that we’re doing it all in seconds, then the bottom of the formula is in seconds squared (aka s²) - an acceleration! Given that both the weighting and day factor are purely numerical, it results in the Timing factor being in s^-2, which is nonsense.

Conclusion

An acre of bullshit, but with the occasional nice flourish (more so than from - gah! - Arnall). All the same, there’s a disturbing lack of science coming from a biology student. I do sincerely hope she recovers from this and has a good career.

An aside - why do PR companies inflict this sort of rubbish on us? Well, as another commentator has pointed out, by getting things like this into news and blogs, instead of as advertising, you probably pay less (although you have the gamble as to whether it gets picked up by anyone), and you don’t have to play by the ASA’s rules, so you can claim absolutely anything…