Sizing up the parties’ chances in November

Press, 27 May 1978, Page 14

Sizing up the parties’ chances in November

J. B. ATKINSON, lecturer in the Department of Political Science at the University of Otago, discusses the book “Election ’78,” written by two fellow political scientists, ALAN McROBIE and NIGEL ROBERTS.

The redistribution of electoral boundaries announced last year was the most extensive dislocation of New Zealand’s electoral geography in more than 30 years.

the major political parties. In 1975, the National Party won its seats more efficiently, electing one M.P. at the cost of 13,875 votes, while 19,826 voters were required for Labour to secure an equivalent return. The 1977 redistribution widens the National Party’s votes-to-seats advantage over Labour, each Labour seat costing 21,148 votes compared with 12,308 for National. Though hardly a countervailing advantage, the Labour Party’s organisation at the electorate level was less severely affected by redistribution than was that of National.

construct an electoral pendulum (familiar to 1975 election night televiewers) which ranks electorates in terms of the share of the two-party vote required for them to switch allegiance. The pendulum shows graphically the magnitude of Labour’s task in November. It suggests that within the new electorates, if there is a uniform two-party swing of 5.2 per cent throughout the country (National losing 5.2 per cent of the vote and Labour gaining the same percentage), the Labour Party will win every marginal National Party seat, plus New Plymouth, and secure a twoseat advantage in the House of Representatives. On the old boundaries, according to Mcßobie and Roberts, Labour needed a swing of only. 4.4 per cent to win. Because of redistribution it now needs to generate an additional 0.8 per cent swing to become the Government.

Some political implifications of these boundary changes are explored in a recently published study by two political scientists, Alan Mcßobie, of the Christchurch Teachers’ College, and Nigel Roberts, of the University of Canterbury. In “Election ’78,” Mcßobie and Roberts redistribute the results of the last two General Elections within the new electorates in order to provide a realistic basis of comparison for election watchers at the next two General Elections.

Mcßobie and Roberts convincingly argue that New Zealand’s redistribution is one of the fairest in the democratic world and that such biases as do occur are unintended and largely unavoidable. They are mainly the result of demographic changes beyond the scope of the terms of reference within which the Representation Commission operates. Enduring demographic tendencies create differential population throughout the country: the “drift to the north” both over the country as a whole and in each island; the depopulation of rural areas as a result of increasing urbanisation: and the depopulation of inner city areas as a result of increasing suburbanisation. Major boundary changes in response to these trends, especially around Auckland, necessarily create a “ripple” effect on boundaries throughout the North Island. Because of its emphasis on the numerical equality of electorate size, the Representation Commission is unable to pay much attention to communities of interest which may be disrupted by boundary changes. If the redistribution systejn remains an imperfect compromise between conflicting interests, however, Mcßobie and Roberts show clearly that it is better than it used to be. For more than 50 years after 1893, for example, electoral boundaries were determined more by the position of licensed hotels than by any other factor. in assessing the likely impact of the recent redistribution on the electoral fortunes of the two major parties, the authors introduce the statistical concept of “swing” which, they claim, is: “A compact measure of the extent to which the position of the two major parties has changed relative to one another.” The virtue of this concept, they argue, is that it enables the analyst to reduce a large number of variables to a single figure. It also enables the authors to

This unique study, based on extensive research (the authors had to locate more than 4000 polling booths throughout the country), deserves to be the bible of New Zealand's psephologists for the next two elections. In addition, its discussion of the background and implications of the 1977 electoral redistribution should be of more general interest. New Zealand’s electoral boundaries are changed at five-yearly intervals in response to the findings of a Representation Commission which, within extraordinarily narrow margins of tolerance, is charged with the task of creating electorates of equal size in order to ensure votes of equal weight. It is one of the curiosities of the electoral laws that these regular boundary changes, which so much disrupt the lives of local party organisers in the name of democratic equality, sometimes create electorates with markedly unequal voting populations. Although the total population of each electorate is established by the commission within 5 per cent either side of a common electoral quota, the size of the actual voting population in each electorate may vary quite considerably. After the 1972 redistribution, for example, Lyttelton’s electoral roll contained 21,144 names, nearly 16 per cent more than the average. Since electorates with large under-age, nonvoting populations are likely to require fewer votes to elect a single M.P., it might be said that the electoral laws discriminate against infertile constituences. The partisan implications of this bias are not immediately apparent, but as Mcßobie and Roberts point out, the latest boundary redistribution engenders some less abstruse, though equally unintended, biases which do have quite significant consequences for

The trouble with this simple statistic, as the authors are aware, is that it conceals what may be a very complex voting reality and is based upon assumptions which may be invalid at the time of the next election. Shifts in voting preference may be the result of straight conversion from one major party to another, but they may also be caused by differential voting turn-out, by the introduction of new voters into the electorate, by third party voting, or by personal rather than partisan allegiances.

Mcßobie and Roberts concede that “the very term ‘swing’ has linguistic connotations of movement from one fixed point to another, and change caused not by movement from one party to another but by movement from «one party into nonvoting is more like ‘drop’ than ‘swing’.” Thus an election result may turn on the ability of voters to enrol correctly. At the last election, for example, between 80,000 and 100,000 votes were, for various reasons, disallowed.

The authors argue that changes in seats primarily reflect nation-wide swings rather than the peculiarities of particular constituencies. Furthermore, they say, voters cast their votes for parties rather than persons. They may be correct most of the time, but the exceptions to these assumptions create more serious problems than the authors admit. For example, they concede that voters’ attitudes towards the background and personalities of the candidates in Dunedin North at the last election “were crucial factors in the National victory in the formerly safe Labour seat.”

They argue, however, that “the unique impact of personality in Dunedin North presents no problems for the calculations in this book,” because the boundaries of that electorate were untouched by the latest redistribution. But Dunedin North still appears in their analysis as a marginal National seat requiring a 3 per cent swing to return to Labour. In fact, of course, Dunedin North is likely to fall to Labour at the next election long before other marginal National seats which, according to their analysis, require smaller swings to change their allegiance. Dunedin North is really a strong Labour seat behind enemy lines.

The problem is that swing is not a universal scientific law. It is what philosophers call an “accidental universal,” relying for its predictive strength on the assumption of a static world—a future exactly resembling the past. The whole concept of electoral redistribution on the other hand is dynamic, based on the reality of population growth. Such growth is usually accompanied by more complex demographic changes. The new electorates are not simply old wine rebottled, but old wine mixed with new rebottled. Mcßobie and Roberts assume that the new electorates resemble the old in all politically relevant respects, but the swing of their pendulum can be thrown out

by a large number of factors: the rise of Social Credit as a third party; the effects of single-issue voting on intensely felt matters such as abortion which can override partisan preferences; increased Maori enrolments on the general roll; and so on. At least Mcßobie and Roberts have the courage of their convictions. Problematic as it may be, their analysis of the effects of electoral redistribution in New Zealand is likely to become a model for future exercises of this kind. Their study is also a valuable aid for the popular pastime of election-watching. It is worth buying the book just to see how close to the mark they get on election night.