Wednesday, August 19, 2009

People Are Liars

I've been looking at the British Social Attitudes survey* because I'm extremely boring. One of the questions is "Which party did you vote for in the 1992 General Election?" and it's interesting how the numbers changed over the course of a decade as the Tories became ever more unpopular:

1994: Conservative- 32.28, Labour- 29.26, Liberal Democrat- 12.5
1996: Conservative- 30.24, Labour- 30.90, Liberal Democrat- 9.21
1999: Conservative- 21.34, Labour- 39.30, Liberal Democrat- 9.18

In reality the Conservative vote in the 1992 election was over 40%, yet by the end of the decade at around half of thosr voters had blotted the memory of voting Tory from their minds.

A similar process is probably underway with Labour voters now.

Despite the headline I doubt that this many people are liars, it's just that they tend to subconsciously rewrite their own histories to fit in better with whatever they believe now, unless they were unfortunate enough to have been on the record about their earlier opinions at the time. This is why the vast majority of people who discuss the Iraq war now insist that they were always opposed to it, yet the opinion polls at the time showed public opinion was split more or less equally.

* It isn't as useful as it's American equivalent the General Social Survey because you can't cross reference as many variables.


James Higham said...

The public have short memories.

Laban said...

Maybe half the Tory voters have died ? 17 years is a long time.

"People Are Diers"

Labour won the 1945 election in a landslide, yet only a tiny minority of voters admit to voting for them now.

Laban said...

And no one admits to voting for Lloyd George.

Ross said...

It only goes up to 1999, just 7 years after the election!

Matthew said...

My thought was the same as Laban's, but as 1992 was high turnout it's not going to work with even extreme assumptions. Let's say in 7 years 3.5m died, and let's say they all voted Tory in 1992 (Those of you who are paid up believers in ZanuPFNulab might think this possible). That would reduce the votes (14.1m, 11.5m, 6.0m) of the three main parties to (10.6m, 11.5m, 6.0m), and their shares would become Con 35%, Lab 38%, Lib 20%.

In fact your figures show Lib Dem deaths to be similar to Tory deaths - they got 17.8% in 1992, but had lost nearly half that by 1999. Maybe 10m Tories emigrated (perhaps Laban can fill in the details?).

There's also the collapse of 3 party support - in 1999 we're down to 70% when in fact the polled 92%. Which parties gained in the survey?

Matthew said...

Btw this thing is known to pollsters, isn't it? I think Anthony Wells has written a bit about how they correct for this, and have started correcting for 'shy labour'

Ross said...

"your figures show Lib Dem deaths to be similar to Tory deaths - they got 17.8% in 1992, but had lost nearly half that by 1999. "

I assumed that becaus ea certain proportion of Lib Dem voters are tactical supporters, a lot of them might be Labour supporters who happened to vote Lib Dem but forget about it soon after.

Ross said...
This comment has been removed by the author.