Social Engineering Blogs

Having been out of the country for the last fortnight, I have been fortunate to have missed the political back and forth in the lead up to the Australain Federal election.  However on the news websites and social media headlines that I saw while overseas they all referred to the polls saying that it was going to be a landslide against Australian Labor Party.  The polls suggested Labor would lose most of their seats in Queensland and it would be a bloodbath in some other marginal seats.

The Principle of Consensus says that when we are not sure of what we should do we look to the behaviour of others like us to guide our decisions.  So by rights Consensus says voters should have at the very least followed the polls and more probably the landslide should have been bigger as those swing voters followed the majority as reported in the polls.

But it didn’t happen.  Why? 

The polls were conducted by reputable polling companies and newspapers.

They had census representative samples.

So what happens when consensus goes wrong?  Or did it?

The answer, well perhaps a contributing factor anyway, was reported by Lucy Carroll from the Sydney Morning Herald on 2 September.  In Carroll’s article she said,

Some of the most influential political pollsters are ignoring millions of voters because they only call landlines….

 The number of people who use only mobiles grew by 20 per cent in the past year, to 3.3 million. Most users are city-dwelling renters, aged under 35.

While Nielsen and Newspoll have come in for the most criticism for not calling mobiles they say they ensure the spread of ages and gender by matching the ratios to the census.

But while mobile phone lists are expensive, online polls are not.

Carroll quoted statistician Nate Silver who reported in The New York Times following the 2012 presidential election that “a number of polling firms that conducted their surveys online had strong results” but those that only called “landlines… performed poorly and showed a more Republican-leaning electorate”.

Therefore did the polls get it wrong because the sample group was not representative of the general population; not based on age or gender but representative of lifestyle, mobility, and broader demographics?

Did the polls reach a certain type of constituent that is more inclined to vote a particular way?

Did people just switch off from the polls because of the increasing frequency of their use and variance between polls?

Here’s The Implication For You

When collecting your case studies, survey data, etc, to use with the Principle of Consensus make sure it is representative of the target of influence and communicate these similarities to them.

Don’t discount methods that allow for a more representative sample, such as online survey tools.

Finally don’t overdo the broadcasting of the results so people become immune to their effects.  Use Consensus when people are uncertain about what they should do and show them what people most like them are doing in that situation.

If you think political polls got it wrong, leave a comment or share this article to promote some discussion.