Back

Note: This article represents the view of the author and not the University of Hong Kong.

Since the Legislative Council Direct Elections in 1991, whenever there are large-scale district elections, the Public Opinion Programme (POP) at the University of Hong Kong would conduct exit polls to study voter opinion and behaviour. This year's District Council Election makes no exception.

According to Chapter 14 of the Guidelines on Election-related Activities in respect of the District Councils Elections issued by the Electoral Affairs Commission (EAC), any organizations intending to conduct exit polls must provide to the Registration and Electoral Office, 7 days prior to the election day, personal particulars of all persons-in-charge and interviewers, as well as the polling station(s) selected. Whether the names of such organizations and persons-in-charge belongs to the public domain is not clear to me.

Three HKU research teams with different features

The election studies led or co-ordinated by the author include three teams of exit poll interviewers, with a total of 160-odd people, conducting interviews outside different constituencies territory-wide. Their operation is briefly summarized as follows:

(1)

Research Team A is a group of professional interviewers employed by POP, with a total of 52 people to be deployed to 28 polling stations. They will sample voters during prescribed time segments according to a prescribed method.

(2)

Research Team B, formed by 76 HKU students, is a self-initiated volunteer research team which will practise exit poll operation and election observation in 45 polling stations. Leaflets will also be handed out to passers-by outside the No Canvassing Zone in order to invite people to express their views at a certain website.

(3)

Research Team C, formed by 23 students from the Department of Politics and Public Administration of HKU, will perform course-related election studies and conduct exit polls in 8 polling stations.

The three HKU research teams are all politically neutral, and their questionnaires have all been designed independently. They are not connected to any political party, and the survey data collected will not be disclosed to any political figures and parties. Data obtained by Teams B and C will not be announced on the election day.

Research Team A is directly under the supervision of POP. Sponsored by the media, its findings will be shared with, and broadcast by, the sponsor, roughly according to the following schedule:

Early stage: The exit poll will start as soon as voting begins. After the sample size reaches an appropriate level, the media will begin to announce figures relating to voters' motivation, their decision process, political demands, etc.

Middle stage: Around noon, the media will begin to release figures on the profile of supporters of individual candidates, such as their education attainment, age distribution, reasons for voting, political demands, etc.

Final stage: After polling ends, the media will broadcast POP's prediction of winners. However, considering the small size of district council constituencies, and the small sample size in each constituency, the number of calls will be limited.

Functions of exit poll

Media all over the world sponsor exit polls, mainly because they can give early predictions of election results, as well as allow detailed analysis of party strengths and their support profile. However, as Hong Kong decentralizes the counting of ballot papers to individual polling stations, instead of transporting ballot boxes to central counting locations, thereby shortening the counting time, the predictive function of exit poll will become less important.

Back in 1991, when POP conducted its first exit poll, we have decided that to focus more on exit polls' function in studying voting behaviour. Therefore, the exit poll questionnaires used by POP over the years are relatively long and complicated. Between sample size and data quality, we have chosen the latter. For this year's exit polls, both long and short questionnaire will be used. It is hoped that, apart from making predictions, survey data collected from exit polls would also contribute to the long-term study of voter behaviour. This position is vastly different from many people and organizations which also conduct exit polls.

In the past 12 years, the author has witnessed many exit polls manipulated by political bodies. These bodies, ranging from left to right, simply use exit polls for election engineering. When their exit poll shows that they are lagging behind in certain areas or among certain types of voters, they would simply re-deploy their manpower strategically to boost their support. In recent years, beginning in the afternoon, some candidates would chant slogans like "Exit poll shows that we are losing by a small margin, come out and vote for us!"

For the record, POP has never published any predication of winners before the close of poll, nor have we ever communicated our findings to any candidate formally or informally. Even for our media sponsors, they will only receive our preliminary predictions in the late evening, and with the understanding that they will not release such predictions before the polling stations close. The self-discipline demonstrated by the local media in this aspect is very commendable.

In any free society, the free-flow of information is part of everybody's right. In the guidelines issued by EAC, the media and research organizations are urged to refrain from announcing any predictions on the voting results until after the close of poll, otherwise an open censure may be made by EAC. This is acceptable. Without the binding of draconian laws, the consensus reached by EAC, the media and research organizations over the past 12 years is indeed a hard-earned achievement.

Exit pollsters must never cheat voters

However, if individual candidates or political parties conduct exit polls in private, and use them for electoral engineering purpose, the aforementioned consensus would then become unnecessary. The effect of such election engineering on polling results may be much more significant than the media's broadcast of predications throughout the election day.

The author does not oppose to political parties' conducting exit polls to facilitate their electoral engineering, everyone has the right to conduct research after all. The author only wants to condemn those organizations which lie to respondents about their true motives, and camouflaged their activities under the name of school projects. If a political party is to conduct exit polls, it should do so frankly and openly. When asked by respondents about their background and objectives, there must not be any concealment.

Whether they are conducted by political parties or academic bodies, exit polls should always apply scientific methods of sampling, rather than snatching everybody coming out of the polling station within a short period of time. If every single voter, soon after stepping out of the polling station, is "taken hostage" by a dozen interviewers, all of whom simply want the voter to vote again, wouldn't it be too disturbing and meaningless?

If we are really doing academic studies, or genuinely gauging voters' demands, why can't we pool our resources together to collect and share the data? We can even open the data for public consumption after the elections. Wouldn't it be more meaningful that way?