Wednesday, 22 July 2015

The plural of anecdote is not data

If scientists had mantras two would stand out:
  • The plural of anecdote is not data
  • Correlation is not causality
I'm sorry they're both negative - I'm sure it's not spiritually sound or something, but it does make them very valuable reminders of two key errors that crop up again and again in everyday life, and they are errors that even scientists can be prone to outside their field.

I point this out because I've been semi-swamped on Facebook and Twitter by people, often scientists or with a science background, sending me stories about the way a particular doctor had worked at the weekend, so the government is entirely wrong. (For non-UK readers, there is a spat between the government, who want hospitals to operate the same at weekends as on weekdays, and the medical profession who say things don't need to change.) Spot the error from above?

I'll come back to the weekends business in a moment, but let me illustrate why this is a terrible way of countering an argument. 

Let's say I was running a campaign to get rid of all out of work benefits. (Let's be clear: I don't want to do this, I'm pointing out the flaw in the doctors' campaign approach.) I could make an impassioned video saying that I have never claimed out of work benefits, so they clearly aren't needed. That's ludicrous, right? And equally it's ludicrous to use a video of someone saying 'I'm working at the weekend' to counter the suggestion that hospitals should operate the same way at weekends as they do on weekdays.

No one is suggesting that hospitals don't operate at weekend - but I don't think anyone would disagree that at the moment the weekend operation is pared down. And there are statistical implications from that.

The government has gone about this in an unnecessarily aggressive and stupid way, granted. But the medical profession don't make things better by using an argument with no scientific validity in response.

No comments:

Post a Comment