Q: When interpreting findings, should we focus more on the positive or negative figures?
Q: After the A45 Concern Group turns into a political party, can we accumulate the popularity ratings of the new party with the Concern Group?
Q: Some obsolete or non-existent political groups have their names appeared in our Stage One naming survey, would this affect the final result?
Q: Why does sample size for different questions vary in the same survey?
Q: Would different sample size for different questions affect the sampling error?
Q: What is the meaning of "weighting" mentioned in the press release?
Q: How can opinion polls avoid asking supporters or opponents of certain political groups?
Q: How can people know the contact information of each survey?
Q: In measuring the popularity of political parties, would it be better to measure hypothetical choice rather than popularity ratings?
Q: Can we deduce respondents' choice of voting based on popularity rating questions?
Q: How can systematic biases be eliminated in sample surveys?
Q: Why not interview experts and professionals instead of members of the general public?
Q: Does HKUPOP provide comments on surveys conducted by other organizations?
Q: Does POP initiate surveys upon the suggestion of members of the general public?
Q: Political parties are not legally defined in Hong Kong, how does HKUPOP define an organization as a political party?
Q: If a political party or group is disbanded to set up a new political party or group, should we treat the two groups as the same?
Q: Certain survey findings are at record high or low, but they are not significantly different from those recorded last time. How should they be reported?
Q: Will HKUPOP reprimand newspapers for inaccurate reporting?
Q: When the proportion of positive answers to a tracking question drops, does it mean that the proportion of negative answers will rise?
Q: Is there an "ideal" support rate?
Q: Is there a "successful" support rate?
Q: What would be the negative counterparts for "ideal" and "successful" support rates?
Q: Are there other benchmarks of popularity other than "ideal" and "successful", "depressing" and "disastrous"?
Q: In describing the support rates of government officials, there are benchmarks like "ideal", "successful", "depressing", "disastrous" and "inconspicuous". Are there similar benchmarks for describing a government's popularity?
Q: In HKUPOP's latest survey, judging from government officials' support rates alone, how can we describe the popularity of the CE and the principal officials using various benchmarks?
Q: There are a number of column articles recently, on the functions of polls and public opinion. Does HKUPOP have any comment?
Q: Would HKUPOP consider including think tanks and political forums into the survey?
Q: When interpreting findings, should we focus more on the positive or negative figures?
答： 今天的新聞公報提供了一個很好的例子。如果我們集中討論各項指標的正面數值，則情況似乎變化不大，但如果我們聚焦在負面的數字，則部分數字其實已經回落至接近歷史新低。明顯地，部分意見已經由負面轉為中性。民研計劃在描述數據時，通常都是先從主流意見著眼，然後再處理小眾意見，不論是先正後負，還是先負後正。如果主流意見已經達到六、七成的水平，其他意見的變化便屬次要。 (初版：2006年2月16日新聞公報之附加資料)
A: Today’s release provides a very good example. If we look at the positive figures, things seem to have remained unchanged, but if we look at the negative figures, some of them have actually dropped to almost record low. Obviously, there was a shift of opinion from the negative end to the middle position. In describing the findings, our general practice is to describe the majority view first, be it positive and negative, and then address the other findings. If the majority view comprises 60% to 70% of the population, shifts in the other end would be of minor importance. (First release: Supplementary section of our press release of 16 February 2006)
Q: After the A45 Concern Group turns into a political party, can we accumulate the popularity ratings of the new party with the Concern Group?
答：我們不會這樣做。坊間報導「基本法四十五條關注組」即將變身成為「公民黨」，屬實的話，除非「關注組」繼續以某種方式存在，否則，我們會把「關注組」在「十大政治團體評分調查」中剔除，而「公民黨」將會以全新政團名義納入調查中。 (初版：2006年2月21日新聞公報之附加資料)
A: We won’t do that. If the news report about A45 Concern Group turning into Civic Party is true, unless the Concern Group still continues to exist in a certain mode, we will take away the Concern Group from our ratings of the top 10 political groups series, and Civic Party will be included in our survey as a brand new political group. (First release: Supplementary section of our press release of 21 February 2006)
Q: Some obsolete or non-existent political groups have their names appeared in our Stage One naming survey, would this affect the final result?
答：不會，因為我們會在第二階段評分調查前剔除有關團體。由於香港沒有政黨法，政治團體的定義相當含糊，因此我們在提名階段盡量寬鬆，之後才核實入選團體是否屬於社會人士公認的政治團體。 (初版：2006年2月21日新聞公報之附加資料)
A: No, because we have already taken away those political groups before the second stage rating survey. Since political groups are not yet legal entities in Hong Kong, such definitions are rather vague, we will try to be as lenient as possible in the naming stage, and then those groups which fall outside the popular definition will be eliminated in the next stage. (First release: Supplementary section of our press release of 21 February 2006)
Q: Why does sample size for different questions vary in the same survey?
答：這是一個在調查中經常出現的情況。原因有三：（一）調查中部分題目在邏輯上只涉及若干次樣本。例如：回答「贊成」或「反對」某項建議者被追問「為什麼贊成」或「為什麼反對」，二者不能共存。（二）部分題目的組合在設計上只涉及若干隨機樣本，以減輕個別訪問對象的負擔。例如：問卷全文包括甲、乙、丙三個冗長部分，調查因而在設計上只容許每次訪問中包含甲乙、乙丙或甲丙三種提問方式，又或者因為甲部屬於核心問題而只容許甲乙和甲丙兩種組合出現，各部分的樣本數目便會有所變化。（三）部分被訪者拒絕回答部分題目，題目的有效基數因而降低。 (初版：2006年3月21日新聞公報之附加資料)
A: This is quite common, ands there are three reasons. (1) Some questions in a survey logically only apply to certain respondents. For example, respondents supporting or opposing certain propositions are further prompted with questions “why support” or “why oppose”. They cannot be asked at the same time. (2) Some questions are deliberately skipped by design, in order to reduce the load on individual respondents. For example, a questionnaire may comprise sections A, B and C, all with many questions. By design, each interview questionnaire may only comprise AB, BC and AC, or perhaps only AB and AC because A comprises core questions. The sub-sample size for different parts of the survey would then vary. (3) A number of respondents refused to answer a certain question, and the effective sample size is reduced. (First release: Supplementary section of our press release of 21 March 2006)
Q: Would different sample size for different questions affect the sampling error?
答：絕對會。因此，在閱讀調查報告時應該注意個別題目的樣本數目是否與總樣本有異。(初版：2006年3月21日新聞公報之附加資料)
A: Absolutely. Readers should therefore pay attention to the sub-sample size of different questions, if they differ from the overall sample. (First release: Supplementary section of our press release of 21 March 2006)
Q: What is the meaning of "weighting" mentioned in the press release?
答：由於種種原因，抽樣調查得出的樣本可能與調查對象出現偏差。例如，如果一項調查成功訪問了1,029名巿民，當中43.2％屬於男性，但政府的統計資料卻顯示香港成年人口中男性應佔47.0％。為了增加樣本的代表性，調查機構便把樣本中每個男性個案的比重，按比例調高約0.088倍，女性個案的比重則按同樣比例調低，以配合調查母體的男女比例，這便是「加權調整」。民研計劃的慣例，是把原始調查數據按照最新人口統計數字，根據年齡及性別的分佈情況加權調整，並把所有原始及加權後的樣本資料與調查數字同時公佈。 (初版：2006年3月23日新聞公報之附加資料)
A: Due to various reasons, the characteristics of a survey sample may differ from those of the target population. For example, a survey may have captured 1,029 Hong Kong residents, 43.2% of which are males. However, government statistics may show that 47.0% of Hong Kong adults are males. In order to increase the representativeness of the sample, the research organization may choose to scale up the importance of each male subject by 0.088 time, and compress the importance of each female subject by the same proportion, in order to match the sex ratio of the target population. This is called "statistical weighting". HKUPOP's general practice is to weight all raw data according to the latest figures obtained from the government regarding gender-age distribution of the Hong Kong population, and to report both the raw and weighted demographic profiles of the respondents in all releases. (First release: Supplementary section of our press release of 23 March 2006)
Q: How can opinion polls avoid asking supporters or opponents of certain political groups?
答：民意調查毋須避免訪問傾向支持或反對某某政黨的人士。重要的是調查樣本中持不同意見的被訪者，比例要與所有目標對象相同。因此便要使用科學隨機抽樣方法。民研計劃一般都是從住宅電話簿中隨機抽樣，然後再從住戶成員中抽取其中一人進行訪問。這個方法是當今社會調查常用的方法，比街頭訪問優勝很多。 (初版：2006年3月23日新聞公報之附加資料)
A: Opinion polls do not need to avoid asking supporters or opponents of political groups. The important point is to make sure that the proportions of respondents in favour of certain opinions are the same as those in the target population. We therefore need to use scientific random sampling method. POP normally randomly samples telephone numbers from telephone directories, and then samples one respondent from the target household for interview. This method is often used in contemporary social surveys, and is much better than street interviews. (First release: Supplementary section of our press release of 23 March 2006)
Q: How can people know the contact information of each survey?
答：民研計劃在每次的民調發放中，都詳細交待該調查的訪問對象和調查方法，包括抽樣方法、被訪者背景資料、問卷全文和其他樣本資料。根據國際標準，交待有關資料是發放和報導民調機構的基本責任。(初版：2006年5月9日新聞公報之附加資料)
A: POP's practice is to give detailed contact information of every survey whenever the survey is released, including the definition of target population and its research design, which includes the sampling method, the demographic background of the respondents, the questionnaire in full, and other contact information. According to international standards, providing such information is the responsibility of any organization when releasing or reporting the survey results. (First release: Supplementary section of our press release of 9 May 2006)
Q: In measuring the popularity of political parties, would it be better to measure hypothetical choice rather than popularity ratings?
答： 臨近選舉，當被訪者可以肯定甚麼人士代表甚麼政黨參選時，假設性的問題如「假如明天投票，你會選擇支持哪個政黨」便能提供很有用的指標。不過，如果某些政黨鼓吹配票或策略性投票，情况便會變得相當複雜。(初版：2006年5月11日新聞公報之附加資料)
A: Hypothetical questions like "which party you would choose if there is an election tomorrow" give very useful indicators near elections, when respondents are quite sure which party is competing against which, and who is representing which party in the election. Needless to say, when parties advocate strategic voting, or split voting among supporters, the situation would be much more complicated. (First release: Supplementary section of our press release of 11 May 2006)
Q: Can we deduce respondents' choice of voting based on popularity rating questions?
答： 可以，我們亦有使用。道理很簡單：當某被訪者給某政黨打上最高分數，我們就可以假設如果即時進行選舉的話，該被訪者就會投票給該政黨。「最高分數」是相對的，可以代表最喜歡或最不討厭的選擇。此外，針對個別政黨評分會比單項選擇更加有用，儘管比較昂貴，但所得到的結果卻可以用不同的組合方法分析。(初版：2006年5月11日新聞公報之附加資料)
A: Yes, we can and we have. The conceptual framework is simple: If a respondent gives the highest marks to Party A among a group of parties, we can assume that he/she would vote for Party A if there was an election right now. "Highest" is defined on relative term, which may mean one's most favourite or least disliked choice. Moreover, rating parties individually is more useful but costly than using one question to map single choice, because the results could be analyzed with a large number of combinations. (First release: Supplementary section of our press release of 11 May 2006)
Q: How can systematic biases be eliminated in sample surveys?
答： 我們可能永遠無法避免偏差，但我們可以盡量把它減到最低。民研計劃的電話調查，會先從住戶電話號碼簿中抽取電話號碼，然後再從目標住戶中以出生日期抽取一人接受訪問。理論上，這個抽樣架已經函蓋了所有香港居民。不過，測試樣本代表性的最好方法，就是比較樣本和母體的人口特性，通常包括性別、年齡、教育程度、房屋類型、職業、地域分佈等等。民研計劃在所有調查報告和發放中，都會交代有關資料。(初版：2006年5月16日新聞公報之附加資料)
A: We may never be able to eliminate all biases, but we can minimize them. For HKUPOP telephone surveys, we randomly sample telephone numbers from household telephone directories, and then select one respondent from a target household using the "next birthday rule". Theoretically, our sampling frame covers everyone in Hong Kong. The best way to examine the representativeness of a sample is to compare its demographic profile with that of the target population, usually in terms of gender, age, education attainment, housing type, occupation, geographical distribution and so on. Such profiles are always given in HKUPOP survey reports and releases. (First release: Supplementary section of our press release of 16 May 2006)
Q: Why not interview experts and professionals instead of members of the general public?
答： 我們有時也會調查專家的意見。不過，專家和巿民的意見始終屬於不同層次，不能互相替代。在民主社會中，重要事情往往最終是由人民投票決定。因此，無論專家意見如何，總要不時探討巿民的意見和需要。(初版：2006年5月16日新聞公報之附加資料)
A: We sometimes do. However, expert and public opinions belong to different levels and they cannot replace each other. In democratic societies where important issues are ultimately decided by the popular vote, it is important to study from time to time what the public wants, in spite of what the experts say. (First release: Supplementary section of our press release of 16 May 2006)
Q: Does HKUPOP provide comments on surveys conducted by other organizations?
答： 除非情況特殊，否則盡量不會，以免有利益衝突。不過，民研計劃絕對不會迴避公民教育的責任。如果民研計劃有充裕的時間和空間去解釋研究方法和各項問題，我們會樂於效勞；如果只是在公眾面前簡單評論某些調查結果，我們認為應該交由沒有利益衝突的獨立學者專家進行。作為長遠目標，民研計劃一直建議所有機構在進行及發放民意調查時，應該遵守國際性的專業守則，以及在公佈調查結果時，一併公佈整份調查報告。政府部門在引述其內部調查結果時，也要同時公佈調查報告，以示公允。(初版：2006年5月23日新聞公報之附加資料)
A: We try not to, in order to avoid any conflict of interest, unless there is a desperate necessity for us to do so. However, we do not avoid taking the responsibility of educating the general public whenever needed. This occurs when we can discuss research methodology in detail, rather than giving short comments to individual studies. We hope this task could be taken up by independent academics or professionals who do not have any conflict of interest. As a long term solution, we have been advocating for a long time for the adoption of international standards in the conduct and release of public opinion surveys, and the release of the full research report whenever a survey is published and whenever the government cited a study it conducted or commissioned. (First release: Supplementary section of our press release of 23 May 2006)
Q: Does POP initiate surveys upon the suggestion of members of the general public?
答： 通常不會，日後可作檢討。不過，我們會接受委托調查，條件是委托人士和機構不能干預我們的獨立性和獨佔調查數據的版權。換言之，所有調查的操作、數據收集和分析工作都是由民研計劃獨立進行，不受外界影響。(初版：2006年5月23日新聞公報之附加資料)
A: It is not our general practice, but we will look into this possibility in future. However, we do accept commissioned survey, but even then, we require research autonomy and shared copyrights, meaning that all fieldwork operations and data analyses would be conducted independently by the POP Team without interference from any outside party. (First release: Supplementary section of our press release of 23 May 2006)
Q: Political parties are not legally defined in Hong Kong, how does HKUPOP define an organization as a political party?
答： 有一定困難，但可以用三種方法解決。第一，由團體自己界定；第二，由巿民界定；第三，採取最廣義的方法界定政治團體，迴避政黨和政團的分別。民研計劃進行的十大政團評分調查，便是結合了第二和第三種方法，先由巿民自由提名，然後篩選出巿民最熟悉的政治團體，再以支持評分排名。(初版：2006年6月6日新聞公報之附加資料)
A: It is somewhat difficult, but we could still solve the problem in three ways. First, let them be defined by the organizations themselves; second, let them be defined by the people; third, study political groups from the widest perspective in order to avoid the distinction between political parties and political groups. In our survey of top 10 political groups, we combined the second and third methods, and let people first nominate the political groups they know, and then rank the most well-known groups according to their supporting rates. (First release: Supplementary section of our press release of 6 June 2006)
Q: If a political party or group is disbanded to set up a new political party or group, should we treat the two groups as the same?
答： 不應該。有關政團或者政黨之所以重組，或者重新命名，就是希望有個新的開始。新成立的政團或政黨除了在名稱上不同外，在組織架構、會籍安排、發展方向等也會不同。況且，舊有政團或政黨亦不一定會解散。從民望分析的角度看，新舊組織可以互相比較，但不能視作同一組織。(初版：2006年6月6日新聞公報之附加資料)
A: No. Whenever a political party or group is re-organized or re-named, it signifies a new start. Apart from the difference in names, the old and new political party or group would also differ in terms of organization structure, membership arrangement and development objectives. Moreover, the old party or group may not be disbanded after all. From the perspective of opinion studies, the new and old organizations can be compared, but cannot be treated as the same entity. (First release: Supplementary section of our press release of 6 June 2006)
Q: Certain survey findings are at record high or low, but they are not significantly different from those recorded last time. How should they be reported?
答： 最好清楚說明「最新數字錄得歷史新高或新低，但在統計上與上次調查數字沒有明顯差異」。「歷史新高或新低」是包涵名次和排序的概念，統計差誤是關乎絕對數值的問題，兩者屬於不同層次，各有理據。以日常生活舉例，某學生今次以一分之微奪得全班第一名，從統計學分析，他可能只是幸運，因為他的成績與第二名的學生沒有顯著分別。但以常人的用法，第一就是第一，一次就是一次。如何描述這個學生的成就要視乎學校的制度，和長期的觀察。(初版：2006年6月13日新聞公報之附加資料)
A: It is better to say that "the latest figures are at record high or low, but they are not statistically different from those registered last time." "Record high or low" encompasses the concepts of ranking and ordering, whereas statistical differences relate to absolute values. They belong to different levels, each with its own logic and rationale. Take a real life example - if a student won the first position by only one mark's difference, from a statistical point of view, he is simply lucky, because there is no significant difference between his/her results and his/her peer in the second place. But from a normal person's perspective, first is first, once means once. How to describe this student's results depends on the system of the school, and some long-term observations. (First release: Supplementary section of our press release of 13 June 2006)
Q: Will HKUPOP reprimand newspapers for inaccurate reporting?
答： 可能會，但我們會極之小心。就以我們於6月1日發表的「六四事件的最新週年調查結果」為例，翌日報章報導的手法各有不同。有說「平反六四支持率新高」，有說「反對平反六四回歸新高」，有些報章側重報導巿民對國內人權狀況的意見，有些索性沒有報導。民研計劃的策略，是在尊重事實的原則下，鼓勵傳媒百花齊放。在沒有專業守則規範之下，報章的報導肯定有所偏頗。我們鼓勵所有報章讀者到《民意網站》閱讀詳細的調查結果。(初版：2006年6月13日新聞公報之附加資料)
A: Maybe, but we will be extremely careful. Take our latest release of "June Fourth Annual Survey" as an example. Different newspapers had different ways of reporting the survey. Some said "record high for supporting a reversion of the official stand on June 4", some said "record high since handover for objecting a reversion of the official position". Some chose to report people's views on China's human rights only, while some simply neglected the survey. HKUPOP's strategy is to encourage free and diversified reporting, provided that everybody respects the truth. In the absence of a professional code of practice, there are bounded to be biases in newspaper reports. We encourage all newspaper readers to go to our POP Site to study the details of all our surveys. (First release: Supplementary section of our press release of 13 June 2006)
Q: When the proportion of positive answers to a tracking question drops, does it mean that the proportion of negative answers will rise?
答： 未必，除非問題的答案只有「正」與「反」兩個選項。如果採用五等量尺或是其他有中間數的平衡量尺，部分正面或負面意見可能是變得中性，而非轉向極端，分析時要格外小心。(初版：2006年6月20日新聞公報之附加資料)
A: Not necessarily, unless there is only one positive and one negative answer. If we are using a five-point scale, or any balanced scale with a mid-point, some of the positive or negative answers might have just turned neutral, instead of taking sides. We therefore need to be very careful when reading these findings. (First release: Supplementary section of our press release of 20 June 2006)
Q: Is there an "ideal" support rate?
答：本欄從今天開始會因應調查數據不時討論這個問題。特首曾蔭權的最新民意支持率是66%，正是討論這個話題的好時機。特首不是普選產生，民望支持率的概念與一般民主社會不同。暫且不談民意支持率與民主選舉結果的關係，如果民意支持率可以直接化成選票，則三分之二絕對支持率，即67%，應可視為一般憲制要求的最高標準。特首曾蔭權的最新支持率剛剛跌出這個標準，在分析上有一定意義。當然，統計上的偶然結果、「棄權」意見的作用、「支持度」與「支持率」的關係等因素，都會影響有關分析，本欄日後會跟進討論。(初版：2006年7月25日新聞公報之附加資料)
A: We will tackle this question every now and then in this section in light of survey findings. The latest support rate of CE Donald Tsang is 66%, which is opportune moment to discuss this question. To start with, our CE is not returned by universal suffrage, so our idea of support rate differs from that in normal democratic societies. Brushing aside the connection between support rate in opinion polls and actual election results, and assuming that our support rate figures can be converted to vote shares, then two-thirds majority, or 67%, can be taken as the ultimate test for all constitutional requirements. The fact that CE Donald Tsang's support rate has just dropped behind that standard has a special meaning conceptually. Of course, whether this is just a statistical coincidence, the meaning of "abstention", and the connections between "support ratings" and "support rates", are also important factors to be considered. We will discuss them here in future. (First release: Supplementary section of our press release of 25 July 2006)
Q: Is there a "successful" support rate?
答：本欄在2006年7月25日的公報中討論了「理想」支持率的問題，指出如果民意支持率可以直接化成選票，則三分之二絕對支持率，即67%，應可視為一般憲制要求的最高標準。退而求其次，半數以上的絕對支持率，即50%以上，如果可以化成選票，都會在一般選舉制度中勝出。因此，51%的支持率可以視為「成功」的民望基準。以民研計劃今日發表的數字而論，只有李少光的民望達到「理想」，曾蔭權、黄仁龍、廖秀冬和葉澍堃的民望屬於「成功」而未及「理想」。當然，統計上的偶然結果、「棄權」意見的作用、「支持度」與「支持率」的關係等因素，都會影響有關分析，本欄日後會繼續討論。 (初版：2006年8月8日新聞公報之附加資料)
A: We discussed the concept of "ideal" support rate in our release of July 25, 2006. We pointed out that if support rate figures can be converted into real votes, then two-thirds majority, or 67%, can be taken as the ultimate test for all constitutional requirements. Likewise, in most constitutional systems, any simple majority of over 50% would return a candidate by popular election. We can, therefore, consider 51% to be the benchmark for a "successful" support rate. Referring to the popularity figures we release today, only Ambrose Lee’s popularity rate could be considered "ideal". The popularity of Donald Tsang, Wong Yan-lung, Sarah Liao and Stephen Ip can all be considered as "successful" but less than "ideal". Of course, the factor of statistical coincidence, the meaning of "abstention", and the connections between "support ratings" and "support rates", are also important factors to be considered. We will discuss them here in future. (First release: Supplementary section of our press release of 8 August 2006)
Q: What would be the negative counterparts for "ideal" and "successful" support rates?
答：本欄在2006年7月25日和8月8日的公報中分別討論了「理想」和「成功」支持率的問題，指出如果民意支持率可以直接化成選票，則三分之二絕對支持率，即67%，應可視為「理想」的支持率，而半數以上的絕對支持率，即50%以上，則可視為「成功」支持率。反過來說，如果民意罷免率達到50％或67％以上，則分別可以視為「很差」和「極差」的民望基準。與「理想」和「成功」呼應，可以用「拙劣」和「失敗」來形容。當然，統計上的偶然結果、「棄權」意見的作用、「支持度」與「支持率」的關係等因素，都會影響有關分析，本欄日後會繼續討論。 (初版：2006年8月29日新聞公報之附加資料)
A: We discussed the concepts of "ideal" and "successful" support rate in our releases of July 25 and August 8, 2006 respectively. We pointed out that if support rate figures can be converted into real votes, then support rates over 67% or two-thirds majority can be taken as "ideal", while support rates over 50% or simple majority can be taken as "successful". Conversely, disapproval rate over 50% and 67% can be considered respectively as "very bad" and "extremely bad" benchmarks of popularity. To contrast them with "ideal" and "successful" support rates, we may describe them as "disastrous" and "depressing". Of course, the factor of statistical coincidence, the meaning of "abstention", and the connections between "support ratings" and "support rates", are also important factors to be considered. We will discuss them here in future. (First release: Supplementary section of our press release of 29 August 2006)
Q: Are there other benchmarks of popularity other than "ideal" and "successful", "depressing" and "disastrous"?
答：本欄在2006年7月25日、8月8日和8月29日的公報中討論了以上四個基準，在此不贅。今回討論的，是倘若有官員長期不被巿民認識，民意支持率和罷免率合計起來都低於50％，亦即是說以上四個基準無一達到，他是屬於怎麼樣的官員。根據我們以民意支持率化成選票的推理，在不少憲制中，低於50％投票率的全民表決會被視為表決無效。以此推理，在民望調查中，如果某官員的「棄權」、「不知道」等比率合計超過50％，則不論他的「支持」與「罷免」比率如何分佈，他都是一個「不顯得稱職」的官員。 (初版：2006年9月12日新聞公報之附加資料)
A: We discussed the concepts of these four benchmarks in our releases of July 25, August 8 and 29, 2006, so they are not repeated here. Today's discussion is how to benchmark some officials' performance if they are not known by the people over a long period of time, meaning that the summation of their popularity support and disapproval rates is lower than 50%, and none of the 4 benchmarks has ever been achieved. According to our reasoning regarding the conversion of support rate figures into real votes, in many constitutions, less than 50% turnout in a referendum would make it invalid. We therefore deduce that if the total proportion of an official's "abstention" or "don't know" rates exceeds 50% in an opinion survey, then whatever the distribution of his/her "support" versus "disapproval" rates, he/she is not seen to be a capable official. (First release: Supplementary section of our press release of 12 September 2006)
Q: In describing the support rates of government officials, there are benchmarks like "ideal", "successful", "depressing", "disastrous" and "inconspicuous". Are there similar benchmarks for describing a government's popularity?
答：本欄在2006年7月25日、8月8日、8月29日和9月12日的公報中討論了以上五個民望基準，不過都是以政府官員的民望為基礎，和以民意支持率化為選票比率作為推理。政府民望調查方面，如果採用的提問為「假設明天進行全民投票，表決現任政府應否繼續執政，你會如何投票？」則或可產生同樣功效。不過，民研計劃現時採用的提問方法是「你對政府整體表現是否滿意」，而答案亦分五等。這種提問比較容易掌握巿民的一般感覺，但就難於推論政府應否倒台，上述五項民望基準亦不易應用。不過，我們可以大膽假設，如果特區首長的民望長期處於「失敗」、「拙劣」或「不彰」的狀態，巿民更換政府的訴求，就會不言而喻。(初版：2006年9月26日新聞公報之附加資料)
A: We discussed the concepts of these five benchmarks in our releases of July 25, August 8 and 29, and September 12, 2006, but they are related to the popularity figures of government officials, and they are all derived from our logic of converting support rate figures into actual votes. In mapping the popularity of a government, if we use questions like "Assuming there is a referendum tomorrow to decide whether the government should remain in power, how would you vote?", we may well come up with similar benchmarks. However, the wordings used by HKUPOP in gauging government popularity is "Are you satisfied with the overall performance of the government?" and we use a 5-point scale to register the answers. We manage to grasp people's feeling better using this question, but it makes it difficult to project whether a government should be removed. The five benchmarks we discussed become inapplicable. Nevertheless, we may boldly assume that if the popularity of our supreme leader remains to be "depressing", "disastrous" or "inconspicuous" for a long time, it will go unspoken that people wants the government to be removed. (First release: Supplementary section of our press release of 26 September 2006)
Q: In HKUPOP's latest survey, judging from government officials' support rates alone, how can we describe the popularity of the CE and the principal officials using various benchmarks?
答：關於官員民望基準的問題，本欄在2006年7月25日、8月8日、8月29日、9月12日和9月26日已經討論了五個基準，包括「理想」、「成功」、「失敗」、「拙劣」和「不彰」，在此不贅。今日討論的，是如何把以上基準應用在最新的調查上。數據顯示，在10月初，保安局局長李少光的支持率超過66％，屬於表現「理想」；律政司司長黃仁龍、特首曾蔭權、財政司司長唐英年、及環境運輸及工務局局長廖秀冬的支持率超過50％，屬於表現「成功」；公務員事務局局長俞宗怡、工商及科技局局長王永平、及政務司司長許仕仁的支持和反對率合計不足50％，屬於表現「不彰」；其餘官員的表現介乎「成功」與「失敗」之間，屬於表現「一般」，包括經濟發展及勞工局局長葉澍堃、衛生福利及食物局局長周一嶽、教育統籌局局長李國章、財經事務及庫務局局長馬時亨、民政事務局局長何志平、房屋及規劃地政局局長孫明揚、及政制事務局局長林瑞麟。以10月初計，沒有官員的表現屬於「失敗」或者「拙劣」。(初版：2006年10月10日新聞公報之附加資料)
A: In our press releases of July 25, August 8 and 29, September 12 and 26, 2006, we discussed five benchmarks of popularity, namely, "ideal", "successful", "depressing", "disastrous" and "inconspicuous". We will not repeat the discussion here, but we will apply them to our latest survey findings. In early October, the support rate of Secretary for Security Ambrose Lee exceeds 66%, his performance can be labeled as "ideal". The support rates of SJ Wong Yan-lung, CE Donald Tsang, FS Henry Tang and Secretary for the Environment, Transport and Works Sarah Liao all exceed 50%, their performance can be labeled as "successful". The combined support and disapproval rates of Secretary for the Civil Service Denise Yue, Secretary for the Commerce, Industry and Technology Joseph Wong and CS Rafael Hui do not reach 50%, their performance can be labeled as "inconspicuous". The performance of all other officials range between "successful" and "depressing", they can be labeled as just "mediocre". They include Secretary for Economic Development and Labour Stephen Ip, Secretary for Health, Welfare and Food York Chow, Secretary for Education and Manpower Arthur Li, Secretary for Financial Services and the Treasury Frederick Ma, Secretary for Home Affairs Patrick Ho, Secretary for Housing, Planning and Lands Michael Suen, and Secretary for Constitutional Affairs Stephen Lam. In early October, no official falls under the categories of "depressing" or "disastrous". (First release: Supplementary section of our press release of 10 October 2006)
Q: There are a number of column articles recently, on the functions of polls and public opinion. Does HKUPOP have any comment?
答：民研計劃注意到近日起碼有五篇評論文章討論上述主題，分別有王家英：〈特首民望的政治學〉，《新報》，11月4日；曾鈺成：〈民意何來？民調何用？民望何價？〉，《明報》，11月6日；余若薇：〈曾蔭權不再孤獨〉，《明報》11月7日；王家英：〈波動的民意與嗜血的政治〉，《新報》，11月11日；和曾鈺成：〈選舉、民望與施政〉，《明報》，11月13日。由於有關文章只是討論各類民意調查的作用，而未有質疑民研計劃的專業水平和操守，民研計劃認為沒有回應的必要。民研計劃歡迎更多學者專家參與討論，集思廣益。民研計劃希望有關人士在討論有關問題時，會同時關注香港社會缺乏民意研究的專業守則，以至不少團體和傳媒在發放和報導有關調查時，沒有區分調查的優劣。更有甚者，不少人士在批評民意調查之時，又同時參與或鼓勵不專業不科學的民意調查，例如音頻互動調查和隨意式的街頭訪問，而又拒絕發表所謂調查的詳細報告。事實上，在電腦網絡發達的今天，只要在互聯網上稍作搜查，便不難發現世界各地的民意研究專業守則。略為一讀，便可知道傳媒和調查機構的專業責任，即時和定期民意調查的功能和局限等。民研計劃在此向讀者推薦以下網址，豐富討論：
- 世界民意研究學會 http://www.unl.edu/wapor
- 世界民意及市場研究專業人員協會 (前稱「歐洲市場及民意研究學會」) http://www.esomar.org
- 美國民意研究學會 http://www.aapor.org
- 美國全國民意調查評議會 http://www.ncpp.org
- (初版：2006年11月14日新聞公報之附加資料)
A: We notice that there are at least five newspaper column articles recently on the above-mentioned topic, written separately by Timothy Wong, Jasper Tsang and Audrey Eu, between November 4 and 13. Because those articles contain no criticism on the professional standards and ethics of HKUPOP, we do not see the need to respond to them. We actually welcome more experts and scholars to join the discussion. We also hope during such discussions, attention would also be drawn to the lack of professional standards for public opinion research in Hong Kong. As a result, the media and the research organizations often neglect the distinction between professional and unprofessional polls in their dissemination and reporting. What is more, quite a number of those who criticizes opinion polls themselves produce or encourage sub-standard polls, like IVR polls and casual street polls, and refuse to disclose methodological details of such polls. As a matter of fact, with the proliferation of the internet these days, it is so easy to search and find professional standards for opinion polling around the world. A quick glance through them would easily show us the responsibility of the media and the researchers, as well as the functions and limitations of instant and tracking polls. HKUPOP hereby recommends the following websites to the reader, in order to enrich the discussion -
- World Association for Public Opinion Research (WAPOR) at http://www.unl.edu/wapor
- World Association of Opinion and Marketing Research Professionals (ESOMAR, previously called "European Society for Opinion and Marketing Research") at http://www.esomar.org
- American Association for Public Opinion Research (AAPOR) at http://www.aapor.org
- National Council on Public Polls of USA (NCPP) at http://www.ncpp.org
(First release: Supplementary section of our press release of 14 November 2006)
Q: Would HKUPOP consider including think tanks and political forums into the survey?
答： 香港沒有政黨法，論政和參政團體未必容易分辨。至於智庫組織，有時又與論政團體和利益團體分不開。因此，民研計劃只會按照目前的調查方法，以巿民用最廣義方法界定的政治團體為主軸，進行政治團體評分調查。(初版：2006年11月30日新聞公報之附加資料)
A: Political parties are not legally entities in Hong Kong, it may not be easy to distinguish political forums from political parties. Likewise, it may not be easy to distinguish think tanks from political forums or interest groups. We will therefore continue our current practice of using the widest definition adopted by the people themselves, in order to measure the popularity rating of different political groups. (First release: Supplementary section of our press release of 30 November 2006)