Advertisement

We need your help now

Support from readers like you keeps The Journal open.

You are visiting us because we have something you value. Independent, unbiased news that tells the truth. Advertising revenue goes some way to support our mission, but this year it has not been enough.

If you've seen value in our reporting, please contribute what you can, so we can continue to produce accurate and meaningful journalism. For everyone who needs it.

Sam Boal
VOICES

What if we could call the election result before it even happens?

David Higgins has spent the past year building a statistical model which he hopes will accurately predict the outcome of the general election next week.

DATA IS ALL around us like never before. The growth of the internet and the advancement of new technology has led to the creation of more information than any of us could ever imagine.

One study suggests that 90% of the world’s data was created in just the last two years. We only have to look at Facebook’s recent jobs announcement to see that “big data” is now a big industry which plays a big role in our country.

‘Data analytics’ has become the broad term to describe the work in this field. Put simply, it attempts to draw value or insight from all of this information. We may wish to crunch Google’s database for Ireland’s most searched terms, or Amazon’s for the most purchased items.

Do you ever wonder how Facebook knows which friends to suggest to you, or why web ads seem to know exactly what you’re looking for? In all these examples, data analytics is at work behind the scenes.

As a student of statistics and computers I had the privilege of spending four years at the core of this field. Week after week we learnt the tools to take seemingly meaningless sets of numbers, and unlock the most insightful of knowledge.

In 2012 we studied the work of Nate Silver, a US statistician analysing the US Presidential election that year. On his website, FiveThirtyEight, he correctly forecast 49 of the 50 states at the 2008 election, and all 50 states in 2012.

Coding the Irish system

Nate_Silver_2009 Nate Silver Wikimedia Commons Wikimedia Commons

Inspired by the work of Nate Silver, I began to explore how this approach could work for Ireland. I’ve had an interest in politics since the age of 16, so I understood the workings of our PR-STV voting system, but translating it into coding language was another task entirely.

In the US, the election is simple, two candidates, two parties and the person with the most votes wins. In Ireland, there are 20 registered political parties, with some running multiple candidates in each constituency, plus independents. You then have a count system with numerous rules and dynamics. Voters not only have a vote for their favourite candidates, but preferences run right down the ballot.

In late 2014 I began putting the project together. Past election results are key to any forecast model. With count data from the last general election and local elections, I was able to build a profile for each party in each constituency. We know intuitively that some parties perform better in certain parts of the country.

Newer and more radical parties tend to draw their votes from urban and poorer areas with the established parties more evenly distributed across the country. We’ve known these trends for decades, but quantifying it more accurately allows us to unlock so much more.

Predicting the marriage referendum

In May 2015, Irish Elections Stats launched as a blog, detailing the work so far and a vision for what the project may become. In the same month, Ireland went to the polls in the marriage equality referendum. Numerous polls emerged in the closing week of the campaign and I tried my hand at forecasting the result.

Polls up to that point suggested 70% support for the amendment, although few took notice of the many undecided voters at the time. Taking an average of polls, and incorporating a strong swing of undecideds against the referendum, I determined a final Yes vote of 63%. On 22 May, Ireland voted 62% in favour.

screenshot.1455897384.94082 www.thejournal.ie www.thejournal.ie

So far so good, but general elections are much more complex. The summer months gave me a chance to expand the code and simulate the Irish system in more detail. By autumn, the model was able to produce probability figures for each candidate.

Fast forward to today, and I have launched Irish Election Stats as a full website with data for each party and each constituency. Users can explore how these forecasts evolve over time. As new polls emerge, the forecasts change. All general election counts are simulated 1,000 times to determine the most likely seats.

One of the most notable stats is a forecast for the number of female TDs. The introduction of gender quotas has put this issue into sharp focus. The projected rise in seats (c. 34 at the time of writing) is a noteworthy improvement over the 25 female TDs elected in 2011.

No clear government

The coalition options also make for fascinating reading. No clear government appears likely. Fine Gael and Labour have moved upwards since last year, however a handful of independents and smaller parties may be required to see them over the line. On current polling, the prospect of an alternative coalition, or Sinn Féin overtaking Fianna Fáil cannot be ruled out entirely.

The statistician George Box famously wrote that “all models are wrong, but some are useful”. Only after the final count will we know the true accuracy of the model. Even taking into account historic trends, we know that elections in Ireland can set new trends very quickly.

In particular, more work is needed to better capture the strong personal vote held by many candidates nationwide. There are also various local factors at play which may fail to be captured, including any constituency based events taking place since the 2014 local elections.

Whatever the outcome, I will have spent the past few months working with the most interesting dataset you could find. At the end of the day, elections direct the path of our country for the next five years. An understanding of the stats is fundamental for understanding our democracy.

I hope over the coming week you will follow the site and take a deeper interest in the data of this election. It is likely to be one of the most fragmented votes in our history.

David Higgins is a political analyst and founder of Irish Election Stats. You can follow him here @higginsdavidw

All our Election 2016 coverage is right here > 

Your Voice
Readers Comments
57
    Submit a report
    Please help us understand how this comment violates our community guidelines.
    Thank you for the feedback
    Your feedback has been sent to our team for review.