As a teacher of VCE maths methods, one of the concepts that I have to explain to students is that of sampling. In particular, how can we ensure a non-biased sample? This might seem arcane but a real-world situation has recently arisen where the implications have affected the political debate.
Soon after the UK EU referendum vote, Sky Data were projecting that only 36% of 18-24 year-olds voted. This was essentially an educated guess that was based, from what I can ascertain, on a variety of measures including polls of voting intentions.
Nonetheless, the genie was out of the bottle and the impression formed that young voters were responsible for the UK making a bad choice. If they really felt strongly about Brexit then they should have voted.
Later, the results of a study by the LSE were published showing that ‘up to’ 70% of 18-24 year-olds did vote after all. This was based upon a poll conducted in the days after the vote where young people were contacted and asked, amongst other things, whether the result had made them cry.
Now the ball was hit to the other side of the court. It was no longer the fault of couch-loving youth voters. A poll with a better methodology than Sky Data proved this. The bad referendum result could not be pinned on them. Instead, those selfish old people were clearly to blame.
But here’s the rub.
Imagine that you are a 20-year-old. You wake up after the vote to the news that the world is about to end. Stocks have plummeted, the pound has fallen, old people and racists have sold your future down the river. They have ruined your future prospects. Then the phone rings.
“Did you vote?”
“How did you vote?”
“Did the result make you cry?”
We should already be alert to the notion that British voters may not always tell the truth to opinion pollsters. The failure of polling at the last UK general election was quite profound. It seems to me that a sample like this is likely to be biased. Even if the pollsters attempted to ‘correct for’ the fact that people tend to over report actually voting, how can we be sure that they corrected appropriately for this very unique set of circumstances?
One way to avoid this problem in future would be to collect data on the ages of voters at source, on actual ballot papers. But there is a good reason why we don’t do this.
In a democracy, everyone’s vote is equal. It doesn’t matter, in principal, how long you will have to live with the result. We don’t discount votes on the basis of life expectancy just like we don’t weight votes on level of educational attainment. The only data that matters is the number voting for a position or candidate.
I am not sure that I like the choice Britain made on the 23rd June but it was a democratic one. Slicing and dicing it doesn’t change that.