The Joy of Stats
You might not realise it, but statistics rule your life. Think about that for a minute. Because even if you claim you know nothing about maths, maths knows a lot about you.
Consider what you did even when you woke up this morning. After you brushed your teeth, you might have put on some face cream that you bought because the TV advert told you 87% of people who used it reported an improvement in the appearance of fine lines and wrinkles. But did you also read the small print at the bottom which said it was 87% of 93 people, tested for a week?
Maybe you've taken a multi-vitamin pill. Why? Because you read somewhere that even if you have a balanced diet, a higher dose than the recommended daily requirement might boost your immune system. You've seen those people coughing and sneezing on the train to work, why take the chance? But where did that story you read come from, what statistics showed that it might work? And come to think of it, what statistics lead to a "recommended daily requirement" in the first place?
As you drink your pro-biotic yoghurt that's "guaranteed to rid your gut of 92% of harmful gut bacteria", you turn on the TV to catch the weather forecast. You catch the sports news first, your favourite football team won last night despite incredibly low possession in the first half and this puts them 5th in the league table, but they may move down if another team gains points for an away win tonight. You're fairly confident your team will stay in place, though, as statistically the other bunch have only won 2, drawn 3 and lost 7 of their away games this season. Out of a total 24 games played, the odds don't seem that great.
Finally. The weather. As if you didn't know already, it's been declared the coldest May for 40 years and rainfall is well above the average. You check the UV risk and pollen count for the day (ha ha, as if) and wonder what on earth happened to this global warming business (because you've never read the statistical analysis which shows this cool, chaotic weather pattern is all within the prediction model).
I could go on. On the train you could read a survey which says that most of the respondents are having more sex than you and in more interesting ways, you could check your horoscope and ponder the probability of That Thing actually happening to you and every other person born between 22nd of September and 22nd October in the world (this is why horoscopes are so vague), and you could read a local piece about parents choosing schools for their children based on their exam rankings. Statistics bombard us all the time, you get the point.
You could just throw your hands up, claim you'll never understand it and trust the people who throw the stats at you indiscriminately. Or you could make an effort to understand and interpret the basic information yourself. It seems like most people go for the former option. An IPSOS MORI poll carried out out on behalf of The Royal Statistical Society and Kings College London shows that people neither feel very confident with statistics themselves, nor confident about the way others use them to tell us news. In fact, 46% of people trust their own "anecdotal" experience, or that of those close to them, more than statistics.
And this doesn't just go for your every day life. Healthcare workers face an onslaught of medical statistics, and they have to make decisions based on them every day. And what's more, those Decision Makers are as human as you or I. Statistics are everywhere in health care systems, from the percentage of nurses in a hospital who work part time, through the ethnic groups of the babies born there, to the incidence rate of MRSA in an administrative area, to the systolic and diastolic readings from a 24 hour ambulatory blood pressure check on a single outpatient.
In Evidence Based Medicine, we can see statistics as the science of making sense of the uncertainty in our decision processes. There will always be uncertainty. There is such huge variety between populations and individuals that we can't possibly make exact predictions. Not only are we talking about variations in age, sex and heredity, we're talking about all those things we looked at while checking our health privilege, difference in methods and diagnostic tools, differences in whether someone remembered their medication every day or was a bit forgetful, differences in the training and medical knowledge of the people supervising the treatment. So many variations, so many ways results may be uncertain. But we can at least use statistics to make them as close to being certain as we can.
And it's not just about the number crunching at the end. We can use statistical methods from the very beginning of our investigations, even in choosing our sample in the first place to make sure they're as representative of the people we'd like to treat as possible and not subject to bias. From there we can decide on the best kind of study to use, whether it's a questionnaire or a whole heap of instrumental measurements from a machine that cost millions. We can train our research team in standard terms and measurements to make sure, say, that the data collected on newborn babies or pregnant women with anaemia or children under 10 with bad asthma or men who need heart surgery.... anything you can think of.... is standardised and usable to process results at the end.
As well as being interesting in themselves, statistics help us to think logically and scientifically about what it is we want to find out and how we'll need to do it. They help us to properly assess the evidence we make decisions on, to be aware of the possible risks associated with those decisions, and also to identify the stuff that, while instinct may lead us down that path, actually just doesn't make sense. We're only human after all, and anecdote is tempting. The also help us to make sense of variations in data, which is a big deal in epidemiology where the world is pretty much our laboratory.
But a knowledge of statistics doesn't only help us to devise our own studies, it also helps us to make sense of the work done by other people, to read journals and find evidence. Even very plausible sounding studies can pull a whole load of statistical stunts that make them sound a lot more certain of their result than they actually are. Even if you're no mathematical whizz kid you need to be able to check that the right kind of study was done on a decent number of representative people, that good results have not been given the spotlight and ones that don't fit hidden in the shadows, and that the results are not only unlikely to be pure chance, but also have an effect worth the money you'll be paying for them.
Statistics and Evidence Based Medicine go together like Laurel and Hardy, like Holmes and Watson, like Kevin Bacon and incredibly bad mobile phone adverts. It's a match made in heaven if we want to make any kind of sense of the evidence before us and weigh up how useful it is. In a world crammed with data and statistical conjuring tricks it's pretty important to get our facts straight. After all, in medicine the lives of thousands of people might depend on it.