The Hong Kong 7/1 March: Crowd Size Estimates (CUHK-HKU-HKPU)
SOURCE #7: CHINESE UNIVERSITY OF HONG KONG-HONG KONG UNIVERSITY JOINT PROJECT
(Ming Pao) July 3, 2004.
[translation] The 7/1 march is over, but the number of marchers continues to be a topic of discussion. This year, the Chinese University of Hong Kong and Hong Kong University joined together to estimate the crowd size. They sent workers to the pedestrian overpass at Wan Chai to count the number of marchers. According to Hong Kong University Department of Statistics and Actuarial Science lecturer Chan So-kuen, the most difficult part of the task is to estimate the number of people who joined in the middle. Therefore, they have added an interview phase to estimate the proportion of people who joined in the the middle in order to accurately estimate the number of participants.
The Journalism and Communications School in Chinese University of Hong Kong and the Statistics Department of Hong Kong University colloborated this year to estimate the 7/1 crowd size. Chan So-keun said that they sent six workers to a pedestrian bridge between Wan Chai and Admiralty. Each worker counted the number of marchers on one of the six traffic lanes for one minute out of every 5 minutes. In addition, they also took photos of the roadway in order to adjust for different human flow densities.
Chan So-kuen pointed out that last year's experience showed that a number of people joined the march in the middle. Therefore, they interviewed people at the end of the march to estimate the proportion of people who joined in the middle. For example, last year, they counted about 260,000 people at the observation point, but only 56% of the surveyed finishers said that they had gone past that point. Therefore, they estimated the total number of marchers to be 460,000 last year. The two universities are still processing this year's study.
今年七一大遊行雖結束，但遊行人數繼續成為討論議題。中文大學及香港大學今年進行有關七一遊行的人數統計，派出工作人員在灣仔行人天橋按跳表計算遊行者。主持研究的港大統計及精算系講師陳素娟指出，統計中最難計算中途加入的遊行人士，故研究加入問卷環節，了解中途加入的遊行人士比例，以便科學化地推算出參與人數。
中大新聞與傳播學院及港大統計及精算系，今年合作統計七一遊行人數。陳素娟說，七一派出6名人員，在灣仔至金鐘一條的行人天橋上，每一人統計一條行車線上的遊行人數，每5分鐘計算1分鐘。此外，統計亦攝錄路面片段，以便調整不同時段人流疏密度。
陳素娟指出，去年經驗發現不少遊行者中途加入，故遊行後用問卷調查遊行人士途經點算地點的比例，以便計算全部曾參與的遊行人數，例如去年點算發現約26萬人參與七一遊行，而問卷調查顯示，只有56%受訪遊行人士途經點算地點，故推算參與遊行的整體人數約46萬。兩大今年的人數調查仍在進行。
(Sing Tao) July 8
[translation] According to Chan So-kuen, the CUHK-HKU research team is presently finishing the write-up of their paper. The paper will be released next week. She said that several researchers have already published their numbers and methodologies. She has no intention of creating any more trouble, so she does not intend to hold a press conference. The methodology and its theory will be described in the paper.
On one hand, it is true that most academic research papers are published without a press conference. On the other hand, is the mob hysteria that accompanied the number of marchers an inhibitor for free exchange of research data and ideas?
(Ming Pao) HKU-CUHK-HKPU Scholars: 210,000 Marched on 7/1. July 16, 2004.
[translation] Based upon their experience in counting last year's 7/1 march, a group of scholars from Hong Kong University, Chinese University of Hong Kong and the Polytechnic University formed a "march size research team." Based upon counting the number of marchers by observation and a telephone survey to estimate the number of people who departed or joined the march midway, they have estimated that the number of marchers at this year's march to be between 180,000 and 210,000.
The number of marchers was estimated by various people to range from more than 100,000 to 530,000 (from the Civil Human Rights Front organizers). Certain sociologists say that the public is interested in the number of marchers precisely because Hong Kong lacks democratic elections to reflect public opinion. Therefore the public was more interested in the "head count" than in western societies as a means of expressing public opinion.
Chinese University of Hong Kong Department of Sociology Associate Professor Chan Kin-man suggested that a standard methodology should be derived from all the different methods from various scholars. Once this methodology is publicly accepted, it will be the basis for estimating the number of participants in mass events in the future.
Hong Kong University Public Opinion Project Director Chung Ting-yiu, Hong Kong University Department of Statistics and Actuarial Science Lecturer Chan So-kuen and Chinese University of Hong Kong Department of Journalism and Communications Associated Professor So Luen-kei formed a research team that had 12 research members who counted the number of marchers with mechanical hand clickers at the pedestrian bridge near Hennessy Road and Arsenal Street.
The research team assigned one person per car lane (six persons for six lanes). They counted for one minute out of every four minutes. Three other people were responsible for double checking. One person was responsible recording the situation for future reference and being the back-up person. The other two people took notes of the conditions. It was estimated that the total number of marchers was 149,000.
Furthermore, the research team interviewed by telephone to interview 231 citizens who participated in the march. They found that 77.4% of them had passed the pedestrian bridge. Therefore, the total number of marchers was estimated to be between 180,000 and 210,000.
This same group of Hong Kong University-Chinese University of Hong Kong researchers estimated the number of marchers in last year's march and came up with an answer between 420,000 and 500,000. That was close to the 500,000 announced by the organizers.
As of today, a total of six organizations or scholars have announced their estimates of the number of marchers for this year's 7/1 march. The numbers varied from 105,000 to 530,000.
On the Internet, there are a number of statistics that are alleged to be the number of people exiting the subway stations at Causeway Bay and Tien Hou. The numbers were supposedly for 7/1 and the same day the week before. So it was suggested that the Octopus card should be used to estimate the number of marchers next year.
According to the MTR spokesperson, the company does not have passenger numbers at individual stations. The spokesperson said that the MTR is in the business of providing transportation services and it is not easy to determine the number of persons coming in and out of individual stations. The Octopus spokesperson said that they don't have any data on individual MTR stations, and therefore cannot consider this proposal for publishing such data.
擁有點算去年七一遊行人數經驗的港大、中大及理大一群學者，自發組成「遊行集會人數研究隊」，透過人手抽樣點算人流、電話調查評估插隊及離隊人士的方法，點算出今年七一遊行人數介乎18萬至21萬。
就各界對今年七一遊行人數的統計由10餘萬至53萬（主辦者民陣）不等，有社會學者指出，公眾重視今次的遊行數字，正因為本港缺乏透過民主選舉以反映民意，才令公眾遠較西方社會重「點人頭」結果，希望藉此表達群眾的意向。
中文大學社會學系副教授陳健民建議，隨不同學者提出各種點算方法，未來可合作訂出一個較具標準及共識的方案，一旦確立公眾認可的人數估算方法，便可作為日後大型遊行活動人數的估算基礎。
港大民意研究計劃主任鍾庭耀、港大統計及精算系講師陳素娟、中大新聞與傳播學院副教授蘇鑰機等多名學者組成的研究隊，於七一派出12名研究人員，在灣仔進入金鐘的軒尼詩道與軍器廠街交界的行人天橋，以手動點算器點算遊行人數。
研究隊伍在每條行車線派1人（6條行車線即共6人），每4分鐘點算1分鐘，另外3人負責複核點算、1人負責同步攝錄作後備複核，另外2人同步記錄路面情，結果估計遊行人流總數約14.9萬人。
此外，研究隊伍以電話調查231名曾參與七一遊行的市民，發現有77.4%人曾經過點算的天橋，故將遊行總人數調整至18萬至21萬之間（詳見A32、A33論壇版）。
上述港大與中大學者群，去年亦有以同樣方法點算七一遊行人數，得出結論，去年遊行人數介乎42萬至50萬，與主辦單位公布的50萬相若。
截至昨日，逾6個團體或學界就今年七一遊行人數統計及公布結果（見表），結論由10.5萬至53萬人不等。
就網上流傳一些地鐵銅鑼灣及天后站於七一當日及對上一個星期日的出站人數，以評估遊行人數，亦有建議明年用八達通統計人數等。
地鐵發言人說，該公司並沒有個別車站出入閘人次的數字，發言人表示，該公司主要任務是提供暢順的交通服務，故很難公布個別車站的出入站人次。八達通發言人回覆，並無地鐵出站數據，故無法考慮公布出站人次的建議。
(Sing Tao) 7/1 Marchers Not More Than 210,000. July 16, 2004.
[translation not provided since it is similar to the above]
繼 港 大 統 計 及 精 算 系 高 級 講 師 葉 兆 輝 及 港 大 社 會 科 學 研 究 中 心 後 ， 再 有 大 學 學 者 公 布 七 一 遊 行 的 獨 立 點 算 結 果 。 由 港 大 、 理 大 和 中 大 多 所 大 學 的 學 者 就 七 一 遊 行 人 數 進 行 的 研 究 推 算 ， 約 有 十 八 至 廿 一 萬 巿 民 參 與 今 年 的 遊 行 。 這 推 算 與 警 方 的 點 算 結 果 相 若 ， 但 與 民 陣 計 算 的 五 十 三 萬 就 有 極 大 差 距 。
今 次 點 算 ， 是 由 港 大 統 計 及 精 算 系 講 師 陳 素 娟 、 理 工 大 學 應 用 數 學 系 助 理 教 授 李 亮 坤 、 中 文 大 學 的 蘇 鑰 機 和 郭 婉 鳳 、 港 大 的 鍾 庭 耀 聯 合 進 行 的 研 究 計 畫 ， 有 關 研 究 結 果 並 會 在 今 天 一 份 報 章 上 刊 登 。
他 們 是 在 七 一 遊 行 當 天 ， 以 定 點 點 算 的 方 式 ， 派 出 六 名 學 生 ， 從 軒 尼 詩 道 至 軍 器 廠 街 ， 點 算 遊 行 人 數 ， 然 後 以 遊 行 時 間 ， 再 配 合 民 意 調 查 統 計 中 途 離 隊 人 士 ， 推 算 出 遊 行 數 目 約 為 十 八 萬 至 廿 一 萬 。 這 個 推 算 結 果 與 警 方 估 計 的 二 十 萬 相 若 。
陳 素 娟 等 學 者 亦 就 今 年 六 四 燭 光 晚 會 的 集 會 人 數 進 行 點 算 ， 結 果 發 現 ， 連 同 足 球 場 外 看 台 及 草 地 的 參 加 者 ， 參 加 六 四 集 會 的 人 士 約 有 五 萬 一 千 人 ， 這 數 字 同 樣 遠 低 於 支 聯 會 聲 稱 的 八 萬 二 千 人 ， 而 與 警 方 估 計 相 若 。
陳 素 娟 公 布 的 ， 是 第 三 個 由 學 者 進 行 的 七 一 遊 行 數 目 統 計 結 果 ， 在 這 之 前 ， 港 大 統 計 系 的 葉 兆 輝 透 過 人 流 計 算 方 式 ， 推 算 出 七 一 遊 行 的 人 數 約 為 十 六 萬 五 千 人 。
另 外 ， 港 大 社 會 科 學 研 究 中 心 在 金 鐘 天 橋 透 過 錄 影 後 慢 格 點 算 的 方 式 ， 點 算 得 行 經 金 鐘 的 遊 行 人 士 有 十 一 萬 二 千 人 ， 並 由 此 推 算 出 遊 行 人 數 為 十 萬 五 千 至 十 二 萬 人 ， 不 過 ， 由 於 社 科 中 心 的 點 算 未 能 計 算 中 途 離 隊 人 士 ， 中 心 主 任 白 景 崇 承 認 當 中 存 在 漏 洞 ， 但 他 相 信 這 類 人 不 會 有 十 萬 之 眾 。
連 同 陳 素 娟 等 學 者 公 布 的 最 後 點 算 結 果 ， 除 了 主 辦 團 體 民 陣 外 ， 其 他 進 行 人 數 點 算 的 機 構 ， 點 算 的 結 果 都 是 在 二 十 萬 或 以 下 。
基 於 錯 誤 算 式 而 錯 計 遊 行 人 數 為 五 十 三 萬 的 民 陣 ， 至 今 堅 持 五 十 三 萬 這 數 字 ， 有 民 陣 成 員 說 ， 現 在 遊 行 人 數 也 是 「 人 言 人 殊 」 ， 無 必 要 作 出 更 改 。
中 大 政 治 系 高 級 導 師 蔡 子 強 認 為 ， 由 多 位 學 者 獨 立 進 行 的 點 算 結 果 都 比 較 接 近 ， 二 十 萬 或 以 下 的 遊 行 人 數 應 較 為 可 信 。 他 認 為 ， 民 陣 以 後 應 考 慮 由 獨 立 人 士 點 算 遊 行 數 字 ， 這 會 增 加 數 字 的 公 信 力 。
Hong Kong University Public Opinion Programme links:
Methodologically, this CUHK-HKU study is not as good as the HKU study led by Yip Siu-fei. While the counting portion is meticulously executed and documented, the telephone survey is vulnerable to criticisms. First, the technical documentation says that the response rate to the telephone survey was 63.8%. Thus, the survey data are valid only to the degree that the non-responders are not different from the responders. By definition, this is unknowable since the non-responders provided no information.
Next, it was noted that a total of 3,512 telephone interviews were completed with persons age 18+. This study therefore excludes marchers under the age of 18. Of the survey respondents, 231 said that they had participated in the march. This would have been an enormous expenditure normally to conduct so many interviews, so I believe that this work must have been in conjunction with the regular polling done by the Hong Kong University Public Opinion Programme. There are about 5.5 million adults in Hong Kong. The estimated number of marchers is therefore 5,500,000 x 231 / 3512 = 362,000. This is obviously not consistent with the human traffic count.
The problem is that the survey was conducted after the march and the people were 'contaminated' by the brouhaha over the number of marchers. Thus, the respondents may or may not be telling the truth at this time. It is also salient to answer 'yes' to this question. For these reasons, the telephone survey should not be used to estimate the number of marchers. The only saving grace is that the key question about whether the person had passed by a special street intersection is usually not understood by non-technically inclined people. That is, there was no obvious way of figuring out whether a "yes" or "no" answer is 'better' for democracy.
This study also provided detailed data that can be used to adjust other estimates. For example, the following table shows the traffic counts by the half hour. The data is also available on a lane-by-lane basis. The entire march lasted 3 hours 50 minutes, but there was a drop off after the third hour. This is proof that the march flow was not constant at the maximal rate, as assumed in the Civil Human Rights Front estimate and the Ming Pao aerial photo analysis.
| Start | End | Total |
| 1525 | 1559 | 22,615 |
| 1600 | 1629 | 22,280 |
| 1630 | 1659 | 22,165 |
| 1700 | 1729 | 22,795 |
| 1730 | 1759 | 24,030 |
| 1800 | 1829 | 20,380 |
| 1830 | 1859 | 12,330 |
| 1900 | 1915 | 2,462 |
| TOTAL | 149,057 |
(SCMP) Survey revises rally turnout to 210,000. By Jimmy Cheung. July 17, 2004.
Disputes over the turnout of the July 1 rally took a new twist after a group of academics said the size of the crowd ranged from 180,000 to 210,000.The latest study was conducted by prominent pollster Robert Chung Ting-yiu and fellow academics from University of Hong Kong, Chinese University and Hong Kong Polytechnic University. Unlike studies that only conducted spot headcounts or used aerial photos to estimate the crowd, the researchers added a survey to find out the proportion of marchers who might not have made it to the headcount points.
The Civil Human Rights Front has been challenged for claiming 530,000 people joined the pro-democracy march from Victoria Park to government headquarters.
Police put the figure at 200,000, assuming a maximum capacity of 170,000 for the streets along Causeway Bay to Central.
With the help of satellite pictures, studies for the Chinese-language newspaper Ming Pao estimated the crowd at 192,000.
Separate research by John Bacon-Shone - a former member of the government's Central Policy Unit think-tank and now with the University of Hong Kong's Social Sciences Research Centre - indicated turnout of 105,000 to 120,000.
In his latest article on the HKU Public Opinion Programme website, Dr Chung said the turnout should not be exploited for political ends. According to headcounts conducted by his team at the junction of Hennessy Road and Arsenal Street, 149,000 people passed through. Subsequent surveys later found that not all who participated in the rally had made it to that point. Of the 231 participants polled, 77.4 per cent said they had reached the area.
Assuming about 20 per cent of the participants had not been included in the headcounts, researchers came up with the final figure of 180,000 to 210,000. But the academics conceded the methodology still had limitations as the survey excluded respondents aged below 18.
Co-researcher Jennifer Chan So-kuen, lecturer at the Hong Kong University department of statistics and actuarial science, said the size of the crowd did not necessarily show the strength and quality of people's aspirations.
Dr Chung added: "Only in the absence of democracy and the lack of opportunity for people to express their aspirations through the ballot box would the turnout become such an important matter." He said the government and rally organisers should entrust headcounts to independent authorities and stay clear of such political disputes in future.
Jackie Hung Ling-yu, spokeswoman for the rally, reiterated that the organisers would not respond to individual turnout counts. She stressed there was no plan to revise the figure. "I think it's up to individual bodies to come up with their own estimate," she said.
I am sorry to say, but this article is totally 'out of it.' I won't say that this reporter is biased, but there is this problem of the innumeracy of the press, which has also exhibited itself elsewhere during this episode. I will begin with the headline: "Survey revises rally turnout to 210,000." Fair enough. In the first paragraph, I read: "a group of academics said the size of the crowd ranged from 180,000 to 210,000." Ahem, sorry, this does not allow you to put in 210,000 in the headline, because you cannot cite the top end of the margin of error as the point estimate. The proper number in the headline is 195,000, which is the midpoint value.
Imagine that you are looking at economic forecasts from a bunch of macroeconomists whose growth estimates range from 4% to 15%. Would you put "Growth estimated at 15%" in your headline? A decent editor would have slapped some sense into the reporter.
Let me review the numbers for you.
Do you see what is going on now? All six other studies are within the margin of error of each other but the Civil Human Rights Front's absolute number of 530,000 is nowhere near. There was no new twist as such, just a slow twising in the wind as one study after another validated each other and made the 530,000 figure ever more improbable.
The Hong Kong 7/1 March: Crowd Size Estimates (CUHK-HKU-HKPU)
SOURCE #7: CHINESE UNIVERSITY OF HONG KONG-HONG KONG UNIVERSITY JOINT PROJECT
(Ming Pao) July 3, 2004.
[translation] The 7/1 march is over, but the number of marchers continues to be a topic of discussion. This year, the Chinese University of Hong Kong and Hong Kong University joined together to estimate the crowd size. They sent workers to the pedestrian overpass at Wan Chai to count the number of marchers. According to Hong Kong University Department of Statistics and Actuarial Science lecturer Chan So-kuen, the most difficult part of the task is to estimate the number of people who joined in the middle. Therefore, they have added an interview phase to estimate the proportion of people who joined in the the middle in order to accurately estimate the number of participants.
The Journalism and Communications School in Chinese University of Hong Kong and the Statistics Department of Hong Kong University colloborated this year to estimate the 7/1 crowd size. Chan So-keun said that they sent six workers to a pedestrian bridge between Wan Chai and Admiralty. Each worker counted the number of marchers on one of the six traffic lanes for one minute out of every 5 minutes. In addition, they also took photos of the roadway in order to adjust for different human flow densities.
Chan So-kuen pointed out that last year's experience showed that a number of people joined the march in the middle. Therefore, they interviewed people at the end of the march to estimate the proportion of people who joined in the middle. For example, last year, they counted about 260,000 people at the observation point, but only 56% of the surveyed finishers said that they had gone past that point. Therefore, they estimated the total number of marchers to be 460,000 last year. The two universities are still processing this year's study.
今年七一大遊行雖結束，但遊行人數繼續成為討論議題。中文大學及香港大學今年進行有關七一遊行的人數統計，派出工作人員在灣仔行人天橋按跳表計算遊行者。主持研究的港大統計及精算系講師陳素娟指出，統計中最難計算中途加入的遊行人士，故研究加入問卷環節，了解中途加入的遊行人士比例，以便科學化地推算出參與人數。
中大新聞與傳播學院及港大統計及精算系，今年合作統計七一遊行人數。陳素娟說，七一派出6名人員，在灣仔至金鐘一條的行人天橋上，每一人統計一條行車線上的遊行人數，每5分鐘計算1分鐘。此外，統計亦攝錄路面片段，以便調整不同時段人流疏密度。
陳素娟指出，去年經驗發現不少遊行者中途加入，故遊行後用問卷調查遊行人士途經點算地點的比例，以便計算全部曾參與的遊行人數，例如去年點算發現約26萬人參與七一遊行，而問卷調查顯示，只有56%受訪遊行人士途經點算地點，故推算參與遊行的整體人數約46萬。兩大今年的人數調查仍在進行。
(Sing Tao) July 8
[translation] According to Chan So-kuen, the CUHK-HKU research team is presently finishing the write-up of their paper. The paper will be released next week. She said that several researchers have already published their numbers and methodologies. She has no intention of creating any more trouble, so she does not intend to hold a press conference. The methodology and its theory will be described in the paper.
On one hand, it is true that most academic research papers are published without a press conference. On the other hand, is the mob hysteria that accompanied the number of marchers an inhibitor for free exchange of research data and ideas?
(Ming Pao) HKU-CUHK-HKPU Scholars: 210,000 Marched on 7/1. July 16, 2004.
[translation] Based upon their experience in counting last year's 7/1 march, a group of scholars from Hong Kong University, Chinese University of Hong Kong and the Polytechnic University formed a "march size research team." Based upon counting the number of marchers by observation and a telephone survey to estimate the number of people who departed or joined the march midway, they have estimated that the number of marchers at this year's march to be between 180,000 and 210,000.
The number of marchers was estimated by various people to range from more than 100,000 to 530,000 (from the Civil Human Rights Front organizers). Certain sociologists say that the public is interested in the number of marchers precisely because Hong Kong lacks democratic elections to reflect public opinion. Therefore the public was more interested in the "head count" than in western societies as a means of expressing public opinion.
Chinese University of Hong Kong Department of Sociology Associate Professor Chan Kin-man suggested that a standard methodology should be derived from all the different methods from various scholars. Once this methodology is publicly accepted, it will be the basis for estimating the number of participants in mass events in the future.
Hong Kong University Public Opinion Project Director Chung Ting-yiu, Hong Kong University Department of Statistics and Actuarial Science Lecturer Chan So-kuen and Chinese University of Hong Kong Department of Journalism and Communications Associated Professor So Luen-kei formed a research team that had 12 research members who counted the number of marchers with mechanical hand clickers at the pedestrian bridge near Hennessy Road and Arsenal Street.
The research team assigned one person per car lane (six persons for six lanes). They counted for one minute out of every four minutes. Three other people were responsible for double checking. One person was responsible recording the situation for future reference and being the back-up person. The other two people took notes of the conditions. It was estimated that the total number of marchers was 149,000.
Furthermore, the research team interviewed by telephone to interview 231 citizens who participated in the march. They found that 77.4% of them had passed the pedestrian bridge. Therefore, the total number of marchers was estimated to be between 180,000 and 210,000.
This same group of Hong Kong University-Chinese University of Hong Kong researchers estimated the number of marchers in last year's march and came up with an answer between 420,000 and 500,000. That was close to the 500,000 announced by the organizers.
As of today, a total of six organizations or scholars have announced their estimates of the number of marchers for this year's 7/1 march. The numbers varied from 105,000 to 530,000.
On the Internet, there are a number of statistics that are alleged to be the number of people exiting the subway stations at Causeway Bay and Tien Hou. The numbers were supposedly for 7/1 and the same day the week before. So it was suggested that the Octopus card should be used to estimate the number of marchers next year.
According to the MTR spokesperson, the company does not have passenger numbers at individual stations. The spokesperson said that the MTR is in the business of providing transportation services and it is not easy to determine the number of persons coming in and out of individual stations. The Octopus spokesperson said that they don't have any data on individual MTR stations, and therefore cannot consider this proposal for publishing such data.
擁有點算去年七一遊行人數經驗的港大、中大及理大一群學者，自發組成「遊行集會人數研究隊」，透過人手抽樣點算人流、電話調查評估插隊及離隊人士的方法，點算出今年七一遊行人數介乎18萬至21萬。
就各界對今年七一遊行人數的統計由10餘萬至53萬（主辦者民陣）不等，有社會學者指出，公眾重視今次的遊行數字，正因為本港缺乏透過民主選舉以反映民意，才令公眾遠較西方社會重「點人頭」結果，希望藉此表達群眾的意向。
中文大學社會學系副教授陳健民建議，隨不同學者提出各種點算方法，未來可合作訂出一個較具標準及共識的方案，一旦確立公眾認可的人數估算方法，便可作為日後大型遊行活動人數的估算基礎。
港大民意研究計劃主任鍾庭耀、港大統計及精算系講師陳素娟、中大新聞與傳播學院副教授蘇鑰機等多名學者組成的研究隊，於七一派出12名研究人員，在灣仔進入金鐘的軒尼詩道與軍器廠街交界的行人天橋，以手動點算器點算遊行人數。
研究隊伍在每條行車線派1人（6條行車線即共6人），每4分鐘點算1分鐘，另外3人負責複核點算、1人負責同步攝錄作後備複核，另外2人同步記錄路面情，結果估計遊行人流總數約14.9萬人。
此外，研究隊伍以電話調查231名曾參與七一遊行的市民，發現有77.4%人曾經過點算的天橋，故將遊行總人數調整至18萬至21萬之間（詳見A32、A33論壇版）。
上述港大與中大學者群，去年亦有以同樣方法點算七一遊行人數，得出結論，去年遊行人數介乎42萬至50萬，與主辦單位公布的50萬相若。
截至昨日，逾6個團體或學界就今年七一遊行人數統計及公布結果（見表），結論由10.5萬至53萬人不等。
就網上流傳一些地鐵銅鑼灣及天后站於七一當日及對上一個星期日的出站人數，以評估遊行人數，亦有建議明年用八達通統計人數等。
地鐵發言人說，該公司並沒有個別車站出入閘人次的數字，發言人表示，該公司主要任務是提供暢順的交通服務，故很難公布個別車站的出入站人次。八達通發言人回覆，並無地鐵出站數據，故無法考慮公布出站人次的建議。
(Sing Tao) 7/1 Marchers Not More Than 210,000. July 16, 2004.
[translation not provided since it is similar to the above]
繼 港 大 統 計 及 精 算 系 高 級 講 師 葉 兆 輝 及 港 大 社 會 科 學 研 究 中 心 後 ， 再 有 大 學 學 者 公 布 七 一 遊 行 的 獨 立 點 算 結 果 。 由 港 大 、 理 大 和 中 大 多 所 大 學 的 學 者 就 七 一 遊 行 人 數 進 行 的 研 究 推 算 ， 約 有 十 八 至 廿 一 萬 巿 民 參 與 今 年 的 遊 行 。 這 推 算 與 警 方 的 點 算 結 果 相 若 ， 但 與 民 陣 計 算 的 五 十 三 萬 就 有 極 大 差 距 。
今 次 點 算 ， 是 由 港 大 統 計 及 精 算 系 講 師 陳 素 娟 、 理 工 大 學 應 用 數 學 系 助 理 教 授 李 亮 坤 、 中 文 大 學 的 蘇 鑰 機 和 郭 婉 鳳 、 港 大 的 鍾 庭 耀 聯 合 進 行 的 研 究 計 畫 ， 有 關 研 究 結 果 並 會 在 今 天 一 份 報 章 上 刊 登 。
他 們 是 在 七 一 遊 行 當 天 ， 以 定 點 點 算 的 方 式 ， 派 出 六 名 學 生 ， 從 軒 尼 詩 道 至 軍 器 廠 街 ， 點 算 遊 行 人 數 ， 然 後 以 遊 行 時 間 ， 再 配 合 民 意 調 查 統 計 中 途 離 隊 人 士 ， 推 算 出 遊 行 數 目 約 為 十 八 萬 至 廿 一 萬 。 這 個 推 算 結 果 與 警 方 估 計 的 二 十 萬 相 若 。
陳 素 娟 等 學 者 亦 就 今 年 六 四 燭 光 晚 會 的 集 會 人 數 進 行 點 算 ， 結 果 發 現 ， 連 同 足 球 場 外 看 台 及 草 地 的 參 加 者 ， 參 加 六 四 集 會 的 人 士 約 有 五 萬 一 千 人 ， 這 數 字 同 樣 遠 低 於 支 聯 會 聲 稱 的 八 萬 二 千 人 ， 而 與 警 方 估 計 相 若 。
陳 素 娟 公 布 的 ， 是 第 三 個 由 學 者 進 行 的 七 一 遊 行 數 目 統 計 結 果 ， 在 這 之 前 ， 港 大 統 計 系 的 葉 兆 輝 透 過 人 流 計 算 方 式 ， 推 算 出 七 一 遊 行 的 人 數 約 為 十 六 萬 五 千 人 。
另 外 ， 港 大 社 會 科 學 研 究 中 心 在 金 鐘 天 橋 透 過 錄 影 後 慢 格 點 算 的 方 式 ， 點 算 得 行 經 金 鐘 的 遊 行 人 士 有 十 一 萬 二 千 人 ， 並 由 此 推 算 出 遊 行 人 數 為 十 萬 五 千 至 十 二 萬 人 ， 不 過 ， 由 於 社 科 中 心 的 點 算 未 能 計 算 中 途 離 隊 人 士 ， 中 心 主 任 白 景 崇 承 認 當 中 存 在 漏 洞 ， 但 他 相 信 這 類 人 不 會 有 十 萬 之 眾 。
連 同 陳 素 娟 等 學 者 公 布 的 最 後 點 算 結 果 ， 除 了 主 辦 團 體 民 陣 外 ， 其 他 進 行 人 數 點 算 的 機 構 ， 點 算 的 結 果 都 是 在 二 十 萬 或 以 下 。
基 於 錯 誤 算 式 而 錯 計 遊 行 人 數 為 五 十 三 萬 的 民 陣 ， 至 今 堅 持 五 十 三 萬 這 數 字 ， 有 民 陣 成 員 說 ， 現 在 遊 行 人 數 也 是 「 人 言 人 殊 」 ， 無 必 要 作 出 更 改 。
中 大 政 治 系 高 級 導 師 蔡 子 強 認 為 ， 由 多 位 學 者 獨 立 進 行 的 點 算 結 果 都 比 較 接 近 ， 二 十 萬 或 以 下 的 遊 行 人 數 應 較 為 可 信 。 他 認 為 ， 民 陣 以 後 應 考 慮 由 獨 立 人 士 點 算 遊 行 數 字 ， 這 會 增 加 數 字 的 公 信 力 。
Hong Kong University Public Opinion Programme links:
Methodologically, this CUHK-HKU study is not as good as the HKU study led by Yip Siu-fei. While the counting portion is meticulously executed and documented, the telehone survey is vulnerable to criticisms. First, the technical documentation says that the response rate to the telephone survey was 63.8%. Thus, the survey data are valid only to the degree that the non-responders are not different from the responders. By definition, this is unknowable since the non-responders provided no information.
Next, it was noted that a total of 3,512 telephone interviews were completed with persons age 18+. This study therefore excludes marchers under the age of 18. Of the survey respondents, 231 said that they had participated in the march. This would have been an enormous expenditure normally to conduct so many interviews, so I believe that this work must have been in conjunction with the regular polling done by the Hong Kong University Public Opinion Programme. There are about 5.5 million adults in Hong Kong. The estimated number of marchers is therefore 5,500,000 x 231 / 3512 = 362,000. This is obviously not consistent with the human traffic count.
The problem is that the survey was conducted after the march and the people were 'contaminated' by the brouhaha over the number of marchers. Thus, the respondents may or may not be telling the truth at this time. It is also salient to answer 'yes' to this question. For these reasons, the telephone survey should not be used to estimate the number of marchers. The only saving grace is that the key question about whether the person had passed by a special street intersection is usually not understood by non-technically inclined people. That is, there was no obvious way of figuring out whether a "yes" or "no" answer is 'better' for democracy.
This study also provided detailed data that can be used to adjust other estimates. For example, the following table shows the traffic counts by the half hour. The data is also available on a lane-by-lane basis. The entire march lasted 3 hours 50 minutes, but there was a drop off after the third hour. This is proof that the march flow was not constant at the maximal rate, as assumed in the Civil Human Rights Front estimate and the Ming Pao aerial photo analysis.
| Start | End | Total |
| 1525 | 1559 | 22,615 |
| 1600 | 1629 | 22,280 |
| 1630 | 1659 | 22,165 |
| 1700 | 1729 | 22,795 |
| 1730 | 1759 | 24,030 |
| 1800 | 1829 | 20,380 |
| 1830 | 1859 | 12,330 |
| 1900 | 1915 | 2,462 |
| TOTAL | 149,057 |
(SCMP) Survey revises rally turnout to 210,000. By Jimmy Cheung. July 17, 2004.
Disputes over the turnout of the July 1 rally took a new twist after a group of academics said the size of the crowd ranged from 180,000 to 210,000.The latest study was conducted by prominent pollster Robert Chung Ting-yiu and fellow academics from University of Hong Kong, Chinese University and Hong Kong Polytechnic University. Unlike studies that only conducted spot headcounts or used aerial photos to estimate the crowd, the researchers added a survey to find out the proportion of marchers who might not have made it to the headcount points.
The Civil Human Rights Front has been challenged for claiming 530,000 people joined the pro-democracy march from Victoria Park to government headquarters.
Police put the figure at 200,000, assuming a maximum capacity of 170,000 for the streets along Causeway Bay to Central.
With the help of satellite pictures, studies for the Chinese-language newspaper Ming Pao estimated the crowd at 192,000.
Separate research by John Bacon-Shone - a former member of the government's Central Policy Unit think-tank and now with the University of Hong Kong's Social Sciences Research Centre - indicated turnout of 105,000 to 120,000.
In his latest article on the HKU Public Opinion Programme website, Dr Chung said the turnout should not be exploited for political ends. According to headcounts conducted by his team at the junction of Hennessy Road and Arsenal Street, 149,000 people passed through. Subsequent surveys later found that not all who participated in the rally had made it to that point. Of the 231 participants polled, 77.4 per cent said they had reached the area.
Assuming about 20 per cent of the participants had not been included in the headcounts, researchers came up with the final figure of 180,000 to 210,000. But the academics conceded the methodology still had limitations as the survey excluded respondents aged below 18.
Co-researcher Jennifer Chan So-kuen, lecturer at the Hong Kong University department of statistics and actuarial science, said the size of the crowd did not necessarily show the strength and quality of people's aspirations.
Dr Chung added: "Only in the absence of democracy and the lack of opportunity for people to express their aspirations through the ballot box would the turnout become such an important matter." He said the government and rally organisers should entrust headcounts to independent authorities and stay clear of such political disputes in future.
Jackie Hung Ling-yu, spokeswoman for the rally, reiterated that the organisers would not respond to individual turnout counts. She stressed there was no plan to revise the figure. "I think it's up to individual bodies to come up with their own estimate," she said.
I am sorry to say, but this article is totally 'out of it.' I won't say that this reporter is biased, but there is this problem of the innumeracy of the press, which has also exhibited itself elsewhere during this episode. I will begin with the headline: "Survey revises rally turnout to 210,000." Fair enough. In the first paragraph, I read: "a group of academics said the size of the crowd ranged from 180,000 to 210,000." Ahem, sorry, this does not allow you to put in 210,000 in the headline, because you cannot cite the top end of the margin of error as the point estimate. The proper number in the headline is 195,000, which is the midpoint value.
Imagine that you are looking at economic forecasts from a bunch of macroeconomists whose growth estimates range from 4% to 15%. Would you put "Growth estimated at 15%" in your headline? A decent editor would have slapped some sense into the reporter.
Let me review the numbers for you.
Do you see what is going on now? All six other studies are within the margin of error of each other but the Civil Human Rights Front's absolute number of 530,000 is nowhere near. There was no new twist as such, just a slow twisting in the wind as one study after another validated each other and made the 530,000 figure ever more improbable.