HKU POP SITE wraps up findings of CE Election rolling pollBack

Press Release on March 26, 2007

Special Announcement

The Public Opinion Programme (POP) at the University of Hong Kong today releases via the 「HKU POP SITE」 (http://hkupop.pori.hk) figures and analyses of the CE Election rolling polls for the last time, including daily tracking figures. Because the last rolling survey was conducted on March 23, the figures released at the POP Site today are already the final figures of rolling poll series. From today onwards, POP will shift its attention to the appraisal and wrapping-up of the entire election process. Findings will be released in due course.

Background

In 1996, the Public Opinion Programme (POP) at the University of Hong Kong conducted many rounds of surveys on the Chief Executive election, most of which were conducted after Tung Chee-hwa formally announced his intention to stand for the election. During the 2005 CE by-election, because a number of well-known political figures had declared their intention to stand, POP therefore also started our CE election polling routine. Between the end of January and early February this year, both Alan Leong and Donald Tsang embarked on their CE election campaigns, POP therefore switched on our CE election series accordingly. In our three rounds of feature survey, respondents were asked to rate the suitability of Leong and Tsang, as well as to vote hypothetically. The results have already been released long ago.

POP's CE Election rolling poll began on February 26 using similar questions, whereby respondents were asked to rate the suitability of the two candidates and to vote hypothetically. The first batch of focus analyses was released on March 9, in the form of 4-day rolling analyses. According to schedule, beginning from March 15, POP increased the daily samples of the polls from 250+ to 350+. On March 21, POP wrapped up the findings again and released them in the form of 4-day and 2-day analyses. Between March 22 and 24, POP further included daily tracking figures and charts on top of 4-day and 2-day rolling analyses in its daily releases via the POP Site. The wrap-up figures and charts released in the POP Site today also include 4-day rolling, 2-day rolling, and daily tracking analyses. This press release reprints some of those charts for focus analyses. Readers who would like to examine the figures in detail, or to study the effects of various demographic variables can check them out at the POP Site.

It should be noted that as the period of rolling gets shorter, the sample size gets smaller, and the sampling errors get bigger. Take for example the latest rolling figures released at the POP Site, roughly speaking, the sampling error of all percentages for 4-day rolling samples should not exceed +/-2.6 percentage points, at 95% confidence level using the full sample. For 2-day rolling samples, it should not exceed +/-3.8 percentage points. For daily tracking samples, it should not exceed +/-5.4 percentage points. Sampling errors of rating figures are calculated according to the distribution of the scores collected.

Charts

(1) Across the rolling polls from February 26 to March 23, daily changes of Donald Tsang's support rate with respective to respondents' different political inclinations are as follows:

(2) Across the rolling polls from February 26 to March 23, daily changes of Donald Tsang's support rate with respective to respondents' different educational attainment are as follows:

(3) Across the rolling polls from February 26 to March 23, daily changes of Alan Leong's support rate with respective to respondents' different political inclinations are as follows:

(4) Across the rolling polls from February 26 to March 23, daily changes of Alan Leong's support rate with respective to respondents' different education attainment are as follows:

Commentary

In POP's analyses of rolling poll findings, we have focused on the extent of support for each candidate with respect to the education level and political inclination of the respondents, because we find them worth-noting after analyzing all our CE election surveys. Moreover, because daily tracking figures can reflect the effect of daily events more accurately within the constraints of sampling errors, they can be mapped with election diaries to study people's changing support of the candidates.

On the four charts carried in this release, Robert Ting-Yiu Chung, Director of Public Opinion Programme, observed, 「According to our daily tracking surveys, people's support of the two candidates in general has remained rather stable over the 26 days of rolling survey. People's support for Donald Tsang in general fluctuated between 70% and 80%, some times over. People's support for Alan Leong in general has remained between 10% and 20%, mostly around 15%. Throughout the survey period, there is a structural difference between people's own political alignment and their support for the two candidates. The election forum of March 1 and the election debate of March 15 have definite effects on some demographic groups, but such effects are shallow and short-lived.」

Concluding Remarks

Other than questions on the research design of the surveys mentioned, POP will not provide further comments on the findings. We welcome questions for follow-up purpose, please email them to us at <[email protected]>. Please note that everything carried in the POP Site does not represent the stand of the University of Hong Kong. Dr Robert Ting-Yiu Chung, Director of POP, is responsible for everything posted herewith, except for column articles which represent the stand of their authors.