A Research Note on the 1990 New Zealand General Election

Abstract

In 1968 one of the authors developed with C. Gillion a method of analysing the outcome of the 1966 General Election based on a linear regression model, and essentially the same technique was subsequently applied to the 1975 and 1981 elections.' The method is applicable when there are no boundary changes between successive elections, so the 1990 election may be analysed with its aid.The model assumes that changes in the size of electorates, as measured by the number of potential voters, and the geographical distribution of electors between successive elections (in the absence of boundary changes) are relatively minor influences on the outcome, as are the qualities of individual candidates and other local factors. Instead it is hypothesized that the dominant influence is nation-wide voting swings between Labour, National and Other, the last comprising votes for minor parties, informal votes, special votes disallowed and non-votes so far as they can be assessed.It should be noted that &dquo;swing&dquo; in this paper means the proportion of votes secured by a party in 1987 which it retained in 1990 or which went to another party (or to wastage). The proportions, assumed to be the same throughout the country, were estimated by conventional regression methods across all (93) general electoral districts. They represent &dquo;average&dquo; voting trends, party to party, against which outcomes in individual electorates may be assessed.The number of Electors on the Master Roll (EMR) is not a good measure of the number of people qualified to vote but it is the only one available. It is in fact an under-estimate, possibly a serious one, for at least three reasons, namely that EMR excludes those who: -qualified after the roll closed, notably late enrolments.were on the dormant roll (having voted at the previous election), but who did not re-enrol, or were not on the dormant roll and did not re-enrol.However, for the country as a whole EMR increased by some 4% between 1987 and 1990, in line with the increase in population of voting age.1