A national referendum was held in Iceland in 1993 on a restructuring of local government. A government committee had recommended mergers that would have affected local government areas in most parts of the country. These were put to the vote in the local areas. The majority in two-thirds of the local areas were against the committee’s proposal. But although the proposal was rejected, there was nothing to indicate what people actually wanted instead: it is not certain that they wanted the status quo to continue.

Shortly after this referendum, the chairman of one of the local councils held a survey of attitudes towards various alternatives for merging his local government area with others. Several alternatives were presented, but participants could only show a preference for one of them. The responses were spread widely across the alternatives, and were difficult to interpret. No alternative received overwhelming support. The chairman took the view that those who did not support the alternative that received the most votes had nothing in common except being opposed to that alternative; they were evidently not in agreement among themselves on what they actually wanted. This is typical of the situation in opinion polls: while people are seen as being opposed to a particular proposal, they do not have the chance to say what they would prefer, and where more than two alternatives are presented, the way in which the answers are distributed can make it difficult to interpret the results.

Shortly thereafter, six local councils in the same part of the country used sequential choice to examine people’s views on ideas concerning the restructuring of local government in the region. Voters in each local government area were faced with the question of whether to continue as a unit or to merge with one or more of the neighbouring units. In three areas, there were six possible merger alternatives; in two, there were seven and in one there were eight.

The method was explained in advance, and there was nothing to indicate that the voters had any trouble understanding it. The number of votes cast in each of the six areas was as follows (with the number of votes cast in the municipal elections shown after each in parentheses, for comparison): 53 (54), 88 (106), 83 (85), 135 (141), 169 (200) and 357 (372).

Many participants marked only one alternative. They were probably the ones who wanted the status quo to continue.

Inevitably, since many of the areas bordered each other, some of the alternatives were identical in many cases, though others were obviously not: in two areas sharing a common boundary running east-west, for example, each might have potential merger partners to the east and the west.

In this case it was important to know whether mutual interest in a merger existed on both sides of the boundaries involved. This became apparent through the use of sequential choice.

The results in one of the areas were as follows:

The chairman of the local council did not like this presentation of the results, in which one of the alternatives (E), which no one gave as their first choice, received 115 points simply for having been ranked at the bottom of the list by many voters. (E was, in fact, the alternative proposed by the government committee.) The picture would be slightly different for the last alternative, E, if, after the totting up the scores, the last alternative (E) were scaled down to 0 and corresponding adjustments made to the other scores. This revised scoring appears in the right-hand column below, with the results under the original scoring system in the left-hand column for comparison.

The alternative which no one placed above any other in their listings is here given a score of 0.

Democracy with sequential choice and fund voting 2003