Friday, 28 October 2011

Fifteen percent of people will say yes to anything: above that, it's significant

*

Having, like most people, been subjected to innumerable reports of surveys, it is clear that 15 percent of the population will say Yes to anything. 

"Are gibbons your favorite food?"

"Is the President doing a great job?"

"Have you got three legs?"

Yes, Yes, Yes - fifteen percent.

*

I conclude that fifteen percent of the population are simply crazed and incompetent (at least 15 percent - since it is the best case scenario that it is, in fact, the same 15 percent of people who make all the weird responses).

*

Fifteen percent sounds like a lot of people, nearly a sixth of the population, until you recognize that fifteen percent will say say (or no) to anything.

After which you can appreciate that any survey result showing <15% 'support' for something or another (or denial of it) should simply be ignored - indeed must be ignored.

*

Yet significantly more than fifteen percent is significant, by the same argument - and since that constitutes more than a sixth of the population then (by the logic of surveys) you have to take notice of it: this is way too big a proportion to ignore.

So fifteen percent is a sharp threshold: a clear demarcation of statistical significance.  

*

Yet of course it is much, much less than fifteen percent of the modern population who are correct about anything...

So the safest course is never do surveys...

Lest you be tempted to take notice of them.

*

8 comments:

  1. "a clear demarcation of statistical significance": come now, you know better than that.

    ReplyDelete
  2. Come now, dearieme, I don't expect Brits to fall into an obvious irony-trap like that one!

    ReplyDelete
  3. But you keep telling me that you are writing for Americans. I'm just trying to help the cousins here.

    ReplyDelete
  4. This is blimmin' brilliant!

    Brits probably comprise 15% of Amerika anyway...

    (Wonders what an irony-trap is).

    ReplyDelete
  5. 15% roughly corresponds to the percentage of the population suffering from severe psychopathy, doesn't it?

    I remember that figure because it also roughly corresponds to the percentage of Americans who profess to be atheists.

    ReplyDelete
  6. I suppose I shouldn't be able to see a suspicious "coincidence" in those numbers. No. That would never do...

    ReplyDelete
  7. @Crow - silly though it obviously is, this is a genuine empirical observation; and a description of how I react to surveys in the UK. Whether fifteen percent works for other places and times remains to be seen...

    ReplyDelete

Comments are moderated. "Anonymous" comments are deleted without being read.