The Secrets of Electoral Opinion polls


Opinion Polls arrows
Arrows representing various opinions

With the narrowing of opinion polls recently in the UK General election, I think people need to question what they actually show.

An opinion poll of voting preference is not simply asking 20 people (or 200 or even 2000) people what they think and then publishing the results.

They are a lot more involved than that and depending on the assumptions used the results can vary greatly.

Firstly a simple example.

To simplily matters and ignore opinion, consider an election on a factual basis where a person would vote whether they are old or young. Based on the criteris there are 20 participants that can vote – 10 old people and 10 young people

As a further simplification for this example assume that everyone will tell the truth.: What would the opinion poll say about the outcome of this imaginary election

What would the outcome would be?

The raw data would say 10 people are voting they are young and 10 that they old.

But that is not enough for electoral opinion polls. They know that not every one votes, that some people change their minds and that there are many other factors to consider.

As a first step, to normalise the data, they may look at previous election results to see how many people vote.

If previously on average 8 out of 10 old people voted and an average of 4 out of 10 young people voted, that would be used to manipulate the results.

Based on that data, it would be expected for 12 votes to be cast., 8 saying old and 4 saying young.

The initial raw data put the results at 50:50 but now taking the old elections into account Old wins by a landslide. 66% of the vote. Young loses heavily with only 33% of the vote.

That is basically how opinion polls work.

There are further manipulations that are done and are often useful to find best case and worst case scenarios. For instance it may be useful for a political party to request a worst case scenario to scare/motivate their voters into turning up and actually voting if they think theyr party’s victory is no longer assured.

Going back to reality, it is a bit more complicated and there are different voting patterns based on location, age and many other reasons including enthusiam and whether the find the candidate aspiring or some local issues that may cause an increase or decrease in support.

Generally older voters support conservatives and actually turn up to vote. Younger voters support Labour (or previously the Liberal Democrats), but they do not turn up.

There are many opinion polls out there. The ones that consider the Conservatives leading by massive margins do not expect the young to turn up in any greater numbers than the past few elections and expect the old to turn up. This is based on historical patterns and takes into account the elctions in 2015 where many polling companies were stung when they assumed that enthusiasm on Twitter and Facebook would result in them actually voting. Ofcourse the polls predicting a hung parliament had its on manipulating effects where some may not have voted due to expecting it to not matter.

Those that are showing the parties closer together are going against historical figures and assuming that this is a special occasion and more young people will turn up and vote that has been the case in recent history and are likely over valuing the Labour votes.

They may also expect more older voters to not vote Conservatives due to the “Dementia Tax” shambles.

Both of those assumptions could turn out to be massively wrong and on 9 June 2017 we may have  a huge majority for Theresa May and the Conservative party.

The only way to prevent this is to turn up and actually vote. Vote Labour and get friends and family to vote labour.

It matters.