HKU POP SITE ereleases regular survey findings today, and gives suggestions to the media on reporting such findingsBack

The Public Opinion Programme (POP) at The University of Hong Kong today releases, as scheduled, via its website the "HKU POP SITE" (http://hkupop.pori.hk) the latest popularity ratings of the "Top Ten Legislative Councillors" and that of Chief Executive Tung Chee-hwa. The date of its next release will be December 11, Tuesday, at 2 pm. The popularity rating of the most well-known political groups, people's satisfaction with the performance of Legislative Councillors, as well as Tung Chee-hwa's latest popularity rating will then be released. Since October 30, the "POP Site" has regularized its release of survey findings at a frequency of about once every two weeks, each time with a forecast of its next date of release. This practice will be reviewed monthly, with the objective of gradually increasing the frequency and content of such releases.

As a matter of fact, since September 1996, POP had been publishing its survey findings regularly via its monthly newsletter, the "POP Express". Additional and frequent releases of "extra" issues of POP Express also evolved gradually. However, due to a number of reasons, the editorial work of these monthly newsletters stopped in November 1999, after publishing the September 1999 issue. By the end of 2000, the irregular extra issues also stopped, their functions gradually taken over by the POP Site. In a way, the increase in POP Site's activity is a continuation of POP's effort in publishing the POP Express. Shall anyone have any question regarding the research design of the surveys published in the POP Site, our team members will be happy to answer them, but we will not comment on the findings. Such an arrangement would be reviewed when we have more resources.

Regarding media coverage of these surveys, POP has advocated for quite a long time that when covering opinion surveys, the media should follow the rules of minimum disclosure, whereby key information like target population, dates of fieldwork, sample size, whether the sample was drawn randomly, response rate, and sampling errors should also be reported. (Please refer to the POP Site under "Code of Practice" for more information.) However, although such information has been given in the POP Site for all its surveys, the media may have difficulty using it. POP would like to offer some suggestions.

Example One: In reporting POP's latest ratings of the "Top Ten Legislative Councillors", it could be mentioned that "the survey was conducted by POP at the beginning of this month, over a thousand people aged 18 or above were interviewed by telephone at random, with a response rate of over 60% … at 95% confidence level, the sampling errors of legislators' recognition rates were less than 3 percentage points, while those of support ratings were less than 1.4 marks."

Example Two: In reporting Chief Executive Tung Chee-hwa's latest rating, it could be mentioned that "the survey was conducted by POP at the beginning of this month, over a thousand people aged 18 or above were interviewed by telephone at random, with a response rate of over 60% … at 95% confidence level, the sampling error of the latest rating was less than 1.4 marks, so the change was not statistically significant."

Plenty of statistical concepts are involved in these suggestions. To put it simple, "at 95% confidence level" means that if we were to conduct the same survey 100 times, using the same question in each survey but with different random samples, we would expect 95 times getting a score within the error margins specified. The margin of error at 95% confidence level is twice the standard error calculated. POP will continue to introduce these statistical concepts via its POP Site.