Aussie Polling Wrong Too

So I actually watched some of the coverage of the Australian election results yesterday. It was quite interesting. I feel like I've went from complete ignorance to beginner level. I actually know some names of Australian politicians now. Though to be fair I was already vaguely aware of Tony Abbott - who lost his seat.

What interested me the most was that the polling predictions were wrong yet again. A victory for the Labor Party (who I think are like our Labour Party minus the "u") was expected, but the incumbent Liberal-National coalition won it. The bookmaker Sportsbet even paid out early on a Labor Party victory, costing them 1.3 million Australian dollars. So quite a big blunder.

(Costly ..great advertising though)

This expensive mistake comes as the latest in a long line of calls the pollsters have gotten wrong. Trump and Brexit being the two biggies of course.

My personal feeling is that the polls are wrong because of the way they're being conducted. However, I saw many people online yesterday putting it down to a different factor. They state that because of the politically correct atmosphere that's current in the world right now, voters are unwilling to share their true feelings with pollsters for fear of being judged right-wing, racist, uneducated, etc. Leading to an underestimation of the right-leaning vote. Which then emerges when people go to vote in private.

This seems like quite a logical evaluation, so it's swayed my thinking a little. I'm still more of the opinion that it's down to the polling itself though. (It could of course be a combination of the two).

I'm not sure how Australian polling is carried out, but most UK polling is now done online.

People freely sign up with online polling organisations, and then get paid 50p or a £1 (sometimes more) for every survey they complete. The invitations to complete the surveys are generally sent out to them by email. With maybe one or two invitations coming each week or so, depending on the circumstances.

In my view this immediately introduces a number of biases into the surveying.

Firstly, it's only polling people who are online. Now sure, most people are connected to the Internet these days. However, there are still a lot of people out there that either aren't connected, or that do have access but don't take advantage of it. We all have family members, particularly older family members, that don't like using technology. That barely know how to open a laptop even. All these people are omitted from these surveys.

Given that these people tend to be older, and are less connected to the wider world, they probably also tend to be more conservative in nature. It means that a whole demographic is missing. Sure, pollsters will say .."but we make sure to have people from all age groups represented in our polls!". However, it stands to reason that a 70 year man who is online, au fait with technology, and young at heart so to speak. Will tend to have a different outlook on life to a 70 year old man sat watching TV all day. A bit of a generalisation, but still..

Secondly, of all the people active online these surveys are only polling the people that have chosen to sign up with a polling company. Introducing yet another bias. Not everyone has the time or patience to sit around doing surveys. Many people won't have even considered the possibility.

Also the time spent doing it is offset by the monetary gain that comes from it. I know from doing these surveys myself. The sort of people that do them are usually people that are looking for ways to make a little bit of extra money. For example, when I used to do these surveys it was because I was a student, or in part time work (or wasting time making zero money blogging). A cheque in the post for £50 every now and then was something I was very pleased to get. I deemed it worth the hours it took to boringly fill out questionnaires on various things I had little interest in.

Of course most people with a proper job or a busy life aren't going to see it the same way. Again, it stands to reason that people willing to give up half an hour of their week to make 50p represent a particular demographic. Everyone not interested in making 50p this way - i.e. people who are relatively well off, people who don't need the money, people who just can't be arsed, etc - aren't likely to be represented in these polls.

Then thirdly, the people looking to make extra money this way will no doubt sign up with multiple polling organisations to maximise their money-making potential. I was signed up with three of the bigger polling companies myself. So consequently all these different polling organisations will likely be polling the same people again and again.

I really don't know how they can get around these biases without changing their methods.