A 'Big Data' Approach to the Problem of Electoral Turnout


Why has New Zealand’s electoral turnout declined? What can be done about it? Previous research can partly answer the first question. But to answer the second, we need to know more about the people who move in and out of voting, a group inaccessible in previous research relying on surveys with low response rates among those not voting. Taking a ‘big data’ approach, using marked electoral rolls, and with a sample of 30,000 this research maps and seeks to explain stability and change in voting and not voting over two general elections and two local elections. It tracks movement between the general and Māori rolls, and by comparison of roll and census information generate simulated responses for the 7-8 per cent not on the rolls. The research tests four hypotheses: that turnout behaviour in the first election a person is eligible to vote will predict later behaviour; if the first opportunity to vote is a local rather than a national election, this will adversely affect later turnout behaviour; turnout behaviour becomes less fluid as people get older; and that Māori enrolment, turnout, and roll choice are affected by a mixture of socio-economic and cultural factors.

Principal Researcher: Jack Vowles

Statistical Consultant: Alistair Gray

Research Assistant: TBA

Supported by the Marsden Fund, 2017-2020