Last year’s Brexit vote sparked a public debate about long term election impacts and whether the people who will experience the longest term impact got the most important vote. This got us thinking about the idea of weighted voting. We wanted to investigate the effect that weighted voting might have had on the results of the 2014 NZ general election.
We used life expectancy as a factor for weighting vote importance, where young people’s votes count for more than older people’s votes. Next we considered political knowledge and general life experience, and decided to incorporate a weighting for this to provide some balance for the high weighting our “life expectancy” factor gave the youngest voters.
Finally, we thought about voter turnout, and giving the younger vote an equal say as the older vote. Younger voter’s turnout is around 65% whereas older age bands can have turnouts up to 88%. So we scaled our votes by age group to create a hypothetical situation where every age band had a 90% turnout.
We combined our three datasets; NZ Electoral Study, Stats NZ Period Life Tables, and Electoral Commission voter turnout. This provided a view of who voted for what party, their age, life expectancy, average political knowledge of their age band, and voter turnout of their age band. The NZ Electoral study included a series of questions testing general political knowledge, which provided us with the data for our political knowledge weighting. We used R and PowerBI for our data manipulation, creating the weightings, data analysis, and visualisation.
We created a website using ReactJS and the Azure platform which enables users to toggle the weightings of our factors to see the effect on the seats in parliament of each party. Users may select to use none, some, or all of the weighted factors, and choose the extent to which the weightings are used.
The biggest driver of change in parliament is the life expectancy weighting, which grants younger voters a higher say than older voters. We know that changing New Zealand’s voting system is practically unlikely, but our results also show that scaling up the young vote would change the seats in parliament. This leads to our call to action, which is that young people need to get out and vote.
We reused data from the New Zealand Election Study 2014, which is a dataset of responses from 2,835 people. The questionnaires were sent to people whose names were randomly selected from the electoral rolls, and people could respond by mailing their responses back in or filling out the survey online. The questionnaires were in the field 2-3 days after the general election. The questionnaire aims to investigate individuals’ perceptions of economic advancement and job security, and the effect on voting choices and turnout. We selected variables from the NZES including age, party voted for, and responses to general election/political knowledge questions. These variables were investigated to determine the effect of weighting people’s votes based on their political knowledge (by age band).
We reused life expectancy data from Stats NZ. The datasets are called Period Life Tables, for the period of 2012-2014. The “expected number of years of life remaining” variable was used to determine the life expectancy weightings that we created for every age from 18 to 70+. This enabled us to investigate the effect of weighting votes based on remaining years of life, where younger peoples’ votes are weighted higher than older people.
We reused Electoral Commission voter turnout data which followed the 2014 general election. It provides a breakdown of voting statistics by age and descent. We chose to investigate turnout based on those who voted out of those who were enrolled to vote. We created weightings for each age band based on turnout, which corrects for lower voter turnout in some age groups. This meant each age group was scaled appropriately to represent a 90% turnout.