Sunday, April 1, 2012

Polls, Can We Trust Them?

Are political polls trustworthy? For the 2011 general election not a single polling company came close. The two big polls TVNZ and Roy Morgan had National well above 50% set to govern alone in their polls just before the election. On the day however, National came up short. The polls also had New Zealand First at 2-3% when on the day they achieved 6%. Then there was the Horizon poll, which had National at below 40%. What do all of these polls have in common? They were all dead wrong.

Polls no longer accurately represent the opinions of the populace, but instead play a role in shaping them. the '08 and '11 elections had some of the lowest voter turnouts in New Zealand electoral history. Polls were everywhere and were often the talk of the town. I believe that such constant polling is detrimental to a democratic system. If people are constantly being told what the rest of the population thinks I can see two effects that are a cause for concern. The first effect of this is that a lot of people tend to align themselves with the status quo. If they see a poll with 60% of the respondents saying they think the economy is headed in the right direction, they will be influenced to believe the same. The second detrimental effect is voter apathy. Having polls predicting the results of the election will make many people think that the outcome is decided already, so may not bother to vote. Think of it from the perspective of a left leaning voter. If the polls tell them that National are going to be able to govern alone, what will their one vote do to stop it?

The sad thing is that if all of the people who didn't vote went to the polling booths on election day, the result would have been completely different.

So what about polls post-election showing National still having over 50% support, despite getting under 48% on election day and with all the bad press exposing their internal civil war and factionalism? I don't believe them for one second. If we truly believe in democracy we would stop these pathetic, inaccurate polls that do more harm than good.


3 comments:

  1. There's been a lot of commentary on polling in the modern age, and you're completely correct - not one of the polls predicted the actual outcome. Looking at the final week's polls (not a scientific sample sorry, just the four I followed and recorded) before election day:

    NZHerald Digipoll - National 50.9%, Labour 28%, Greens 11.8%, NZFirst 5.2%;

    One News (Colmar Brunton) - National 50%, Labour 28%, Green 10%, NZFirst 4.2%

    Three News (Reid research) - National 50.8%, Labour 26%, Green 13.4%, NZFirst 3.1%

    Roy Morgan Research - National 49.5%, Labour 23.5%, Green 14.5%, NZFirst 6.5%

    Actual election results - National 47.3%, Labour 27.5%, Green 11.1%, NZFirst 6.6%.

    Given that most of these polls have to work within a margin of error of around 3-4%, you could argue that in most cases they fell within the normal margin, so simply pointing out that they were wrong ignores the fact that you can't actually completely predict human behaviour.

    Obviously, those polled form a separate subgroup in society from those who go to the polls - being polled by these research companies often requires you to have a landline or internet connection, which can skew the results a little given that more people who are unable to afford such items generally vote for more left leaning parties, if they do vote.

    So, given these limitations, why even try?

    Seriously, what is the point of these opinion polls?

    I think we have to seriously question the role of media in our society. It is the media that make such a big thing out of these polls - in my opinion, it is so they can quantify their stories in the absence of any other 'solid' facts - essentially, it gives them something to talk about without having to go too far into details.

    The other thing we seriously need to consider is the state of our democracy. A functioning democracy is dependent on a well educated, motivated, and politically aware populace, one which is aware of the importance of politics and takes an active interest in it. Without such a populace, democracy becomes essentially an exercise in futility as the only people who become involved are the power-seekers and the axe-grinders. Where does the problem lie? Where else but the education system? Why else do you think the least politically active group in society is the one which has just recently left school? Civic responsibility and political awareness have been pigeon-holed into the social sciences area - the one area which is not compulsory for senior students. If we took democracy seriously, we'd have a compulsory subject like Civics or Government or Citizenship at Year 12 and/or 13.

    However, on saying that, the only people who actually make their minds up to vote based on polls are people who shouldn't be allowed to vote anyway. Democracy is only as strong as our weakest voters - and it's getting pretty fragile.

    ReplyDelete
  2. I'm in two minds over political polling, particularly during the week or two immediately prior to the election. On the one hand I agree that the results are invariably biased and certainly have the potential to influence voting decisions. But on the other hand there are a number of situations under the present MMP rules where having an indication of a party or candidates level of support is vital casting a meaningful vote.

    For example, if I was intending to vote for Mana or NZ First I'd need to be reasonably confident that those parties where going to get past the threshold before committing, if I didn't want to see my vote wasted.

    Polling data also gives voters an indication of what kinds of coalitions could occur post-election - potentially allowing for more informed decisions in the voting booth.

    ReplyDelete
  3. Wow, great comments!

    Karl, I agree with everything you said.

    DC, I agree and I think lowering the threshold would help combat that. A lot of Kiwis are still in the two-party mindset from the FPP days, and I think the 5% threshold has a substantial role in that.

    ReplyDelete