Problems and Purpose

The referendum asked Australians to vote Yes or No on a constitutional amendment to make Australia a republic, cut its remaining constitutional ties to the British Crown, and replace the Queen and governor-general with a president chosen by Parliament. The president was intended to have roughly the same, largely ceremonial role as the governor-general. Accompanying legislation spelled out the details, including that the president would be nominated by the prime minister, seconded by the leader of the opposition, and approved by two-thirds of the bicameral Parliament sitting as one body and could be removed by the prime minister, subject to after-the-fact ratification within 30 days, by a majority of the House of Representatives. The president would inherit the uncodified “reserve powers” by which the Crown and governor-general can dismiss a prime minister, grant or refuse parliament’s dissolution, and issue election writs. 

Background History and Context

The referendum stemmed in large measure from agitation, over the preceding decade, by the Australian Republican Movement (ARM), formed to work toward “a Head of State who is an Australian citizen, who is appointed by Australians and who represents the independent and sovereign nation of Australia.” The ARM urged a minimal presidency on the lines of the eventual referendum proposal, fearing that a directly elected and possibly stronger presidency would be too radical and too complicated a change. The opposing case was argued most prominently by Australians for Constitutional Monarchy (ACM), formed soon after to “defend the Australian Constitution, the role of the Crown in it … and resist its replacement with a republic.” The ACM defended the status quo on the grounds that Australia was already a “crowned republic”; that the governor-general, an Australian, not the Queen, was actually head-of-state; and that it was foolish and possibly dangerous to tinker with a constitutional system that had served well for a hundred years.

The governing Liberal-National coalition and the opposition Labor party took visibly different positions. Labor Party leaders tended to favor the proposal, while the governing Liberal party was more evenly split, as to a lesser extent was its coalition partner National Party. The then and current Liberal prime minister, John Howard, personally opposed any change. Nevertheless, his government convened a people’s constitutional convention to consider the issue in February, 1998. After two weeks of fractious, nationally televised debate, above all over the issue of whether the president in a republic should be parliamentarily appointed, as urged by the ARM, or directly elected, the “Con-Con” endorsed a republican model involving a parliamentarily appointed president. The result, after Howard was forced to retreat from an initial, less palatable version, was the referendum proposal described above.

The government appropriated $20 million for educating voters about the issue. Five million dollars went to the Electoral Commission to send every voter a neutral information document about the referendum. The remaining $15 million was divided equally between 3 officially constituted “Yes” and “No” committees, each consisting of ten prominent advocates. The No committee included two direct election republicans. The three-month campaign saw newspaper and television advertisements, talk-show discussions, TV debates, town meetings, and appeals by past prime ministers and governors-general, eminent jurists, and sports stars.

The Yes campaign described the proposed change as “small and safe,” sneered at the prospect of eventual rule by “King Charles III and Queen Camilla,” and cited the support of distinguished Australians. The No campaign railed against “Chardonnay-swilling elites” fomenting a republican plot, claimed that the referendum would create a “politicians’ republic,” and argued that “real democrats,” favoring a directly elected president, should vote No. Some extreme anti-republicans raised the specter of secession by states refusing to adopt republican governments, of Australia’s ouster from the Commonwealth, and even of the republic’s being like Weimar Germany waiting for Hitler.

In the end, the referendum was rejected by a margin of 55-45. It failed in all six states and the Northern Territory. Only the Capital Territory, containing Canberra, registered a majority in favor. A very sizable portion of the No vote clearly came from direct-election republicans. Polls showed that fewer than 10 percent of those who voted No liked having the Queen as Australia’s head of state and that up to half of them, depending on how the question was worded, would have voted Yes if the proposal had been for a directly-elected president. 

Participant Recruitment and Selection

In the event, the participant sample was highly representative. The overall response rate (the number of participants divided by the number of designated interviewees) is comparable with those of the best commercial polls, honestly calculated (Brady and Orren 1993; King and xxxx, Krosnick?). Comparing the participants with the nonparticipants shows that the sociodemographic differences between them are almost entirely minor and of a piece with known biases in ordinary polls. The participants are somewhat better educated, older, more affluent, and more interested in and knowledgeable about the subject area than the nonparticipants. When it comes to beliefs and attitudes, moreover, the statistically significant differences are fewer and still smaller. The participants were somewhat more inclined (57% versus 51%) to favor the referendum proposal, although the difference is not statistically significant.

Methods and Tools Used

In the present case, we had the commercial survey house Newspoll interview 1220 respondents, of whom 347 attended the deliberative weekend, October 22-24. The interviews were by telephone, and the sampling by RDD with quotas for sex. The deliberations took place in Old Parliament House in Canberra. Other elements of the quasi-experimental design rested on the close collaboration between the ADP and the Australian Constitutional Referendum Study (ACRS) of the Australian Election Studies. All the ADP’s original interviewees, both participants and nonparticipants, were mailed ACRS questionnaires to get their views just after the referendum, at the same time as the ACRS’s own independent random sample. Since the ACRS questionnaire included most of the ADP questionnaire’s items, the responses from the ADP participants constitute a third wave a measurement on the participants and a second wave on the nonparticipants. The latter constitute a before-after quasi-control group, while the ACRS respondents constitute a rather better “post-test only” quasi-control group. Ensuing reports will make more use of these additional data; here we advert to them only occasionally in passing. 

Influence, Outcomes, and Effects

The bottom line was a huge, 16 point increase in the Yes vote, from 57 to 73 percent. Among the ADP participants, the referendum would have carried easily. 

The public’s preferences were actually divided among three broad alternatives: a republic with a parliamentarily chosen president (the referendum proposal), a republic with a directly elected president, and the status quo. Before deliberation, direct election was by far the most popular option, the first choice of almost 51.5 percent of our participants. Those favoring direct election opposed the referendum proposal as too little of a good thing. By the end of the deliberative weekend, however, only 20.5 percent ranked direct election as their first choice, apparently because many participants came to believe that it would make the office and the selection process political. Thus many of those initially favoring direct election moved toward the referendum proposal as what they now saw as the best republican alternative. First-choice support for the status quo also declined, from 27.5 to 15.7%, and first-choice support for the referendum model increased from 21 to 64 percent. 

The participants learned a lot. Between the initial interview and the end of the deliberative weekend, their “domain-specific” knowledge of the referendum proposal and the status quo increased dramatically. At the beginning, our participants, like the citizenry as a whole, knew little about these matters. On average, only 39% got the five domain-specific knowledge items right, not so much more than the 27% who would have done so by blind guessing.

By the end of the deliberative weekend, however, they knew a great deal more. The mean percentage getting these same five domain-specific items right rocketed to 70%. Sixty-one percent knew that “the governor general can decide whether or not to dismiss the government,” 73 percent that under the referendum proposal, the prime minister “could remove the president at any time but must later obtain approval from the house of representatives,” 85 percent that “the Queen appoints the governor general only on the advice of the prime minister,” and fully 92 percent that the role of the president under the referendum proposal would be like that of the current governor general (rather than the prime minister, the American president, or the British prime minister).

Only the question about which party was more in favor of a republic showed any decrease, and that for understandable reasons. The panelists appearing before the ADP included Yes and No advocates within both Liberal and Labor parties, obscuring the difference in central tendency between them. The percentage answering this party location item correctly decreased from 47 to 40%. That one item aside, the percentages getting these items right increased by an average of 40.5%. The percentage knowing about the prime minister’s ability to remove the president under the proposed republic increased by 57%.

“General” knowledge of politics more broadly also increased, if much more modestly. On average the percentage knowing that Jennie George is the president of the ACTU/leader of the Australian workers’ union, that Aden Ridgeway is an aboriginal senator in parliament, that the Liberal party “is closer [than the Labor party] to business,” and that the Labor party “is more concerned [than the Liberal party] about social and welfare issues” increased by 6%. From the standpoint of the referendum debate, these items represented incidental, background knowledge. Neither the briefing materials nor the discussions directly concerned them. Yet the participants learned something about them anyway.

With only one exception, both referendum and general political knowledge increased further during the several weeks between the deliberative weekend and the ACRS survey. The percentage knowing that under the referendum proposal, the prime minister “could remove the president at any time but must later obtain approval from the house of representatives” increased further from 73 to 81 percent. Most strikingly, the percentage knowing that the Labor party was more in favor of a republic than the Liberal party, which had slipped from 47 to 40 percent, shot up to 65 percent. Doubtless this increase stemmed partly from Howard’s publicly opposing the referendum proposal during the last two weeks of the campaign, i.e. after the deliberative weekend, but we suspect that it also occurred because the ADP participants were by that point very actively looking for answers to the factual questions they thought they might have missed, as their further gains on the other knowledge items suggests. 

Analysis and Lessons Learned

One reasonable concern is that much change in preferences could result simply from conformity mechanisms. The mean opinion of each small group might be expected to shift toward the same side of the mid-point as it originally lies on. Predominantly pro-referendum small groups might be expected to be come more pro-referendum, predominantly antireferendum ones more anti-referendum. At the same time, the variance within each small group might be expected to shrink The second pattern does obtain: a large majority of the small groups do show diminishing variance of opinion. That is partly, however, because so many people in all groups moved to the Yes position, and so few to the No position. That same across-the-board shift also limited the number of groups changing in the expected direction. On vote intention, only 45.8% of the groups moved toward the same side as they were already on. On the trichotomous preference item, only 8.7% of the groups saw an increase in the frequency of what had been their model first choice category, and only 26.3% saw an increase in the frequency of what had been their first and second choice categories combined.

In the real world, the Australian constitutional referendum went down to defeat. Fifty-five percent of the voters voted No. Only 45% voted Yes. From the extenuationist point of view, this must have been something very like the full-information outcome—how the voters would have voted if they all had known as much as the experts about the referendum proposal and the status quo. Indeed, it must have been even closer than usual to the full information outcome, given the great wealth of information readily available from the Yes and No Campaigns and the Electoral Commission, among other sources.

From the perspective of the Deliberative Poll, however, the voters got the referendum quite wrong. The AD participants gained a very great deal of information. They thought about the issue much harder than average. They discussed it much more, and with a much wider variety of their fellow citizens. And they voted resoundingly for the proposed republic.

What was the difference? It seems to have lain in a mix of changed values, changed perceptions, and the clearer reasoning from values and perceptions to votes that comes with increased knowledge. The participants both came increasingly to value having a president from outside politics and increasingly to see the direct election model as likely to produce presidents from within politics. They came both decreasingly to think that Australia should maintain its traditional ties with Britain and increasingly to think that the referendum proposal would boost Australia’s independence. Etc. We shall soon begin more extended statistical modeling of just what went into the changes of preference, and to what degree.

Of course this is just one referendum. There may be others in which fuller information would make no appreciable difference. Based on the results of statistical simulations for candidate and party elections and for policy preferences, however, we doubt they can be common. In any event, the Australian constitutional referendum is a very clear case of heuristics and simple cues’ not being enough. One may take the precise 73-27% margin with some salt. Perhaps a different set of experts, of performances by the experts we had, or a different sample of voters would have produced a different outcome. Perhaps, but our experience with a series of eight regional Deliberative Polls about electric utility issues, which varied considerably in this sort of operational detail, suggests otherwise. Despite countless differences in the composition of expert panels, what questions the panelists were asked, what they said in response, the briefing materials, and of course the participant samples, the results came out remarkably similar across the series (Luskin, Fishkin, and Plane 1999). Thus while the margin might not have been 73-27, it does seem abundantly clear that the proposal to make Australia a republic would have passed, if the whole electorate had learned and thought more about the issue. 

See Also

https://cdd.stanford.edu/2005/deliberation-and-referendum-voting/

References

Taken directly from https://cdd.stanford.edu/2005/deliberation-and-referendum-voting/

External Links

https://cdd.stanford.edu/2005/deliberation-and-referendum-voting/