The Hong Kong 7/1 March: Crowd Size Estimates

Since I am professionally a statistician, my long comment will be on the crowd size estimates.  This is not a trivial issue since it has become highly controversial on account of the huge divergence among various estimates.

There is a vast amount of information available, so the details are placed on separate pages.  Each page includes the estimate coming from a particular source together with the methodology, news reports, technical analyses and commentaries.

Sources #1, 5:  Civil Human Rights Front : 530,000 and Human Rights Monitor: 160,000 (updated August 4, 2004)

Source # 2: Hong Kong Police: 200,000 (updated: July 20, 2004)

Source # 3: Ming Pao aerial photo analysis: 192,000 (updated: July 21, 2004)

Source # 4: Hong Kong University Department of Statistics & Actuarial Science (Yip Siu-fei): 165,000 (updated: July 11, 2004)

Source # 6:  Hong Kong University Social Sciences Research Center (John Bacon-Shone): 112,000 (updated: July 15, 2004)

Source # 7:  Chinese University of Hong Kong-Hong Kong University-Hong Kong Polytechnic University: 195,000 (updated: July 17, 2004)


Originally, I just reported the numbers.  Then someone wrote to me and insisted that I help sort out this mess.  That is, they wanted to know what 'my' estimate is.  Here is how I went about doing this based upon the known data:

Source #1, the Civil Human Rights Front number is 530,000, which is corrected down to 389,000 by reason of physics and a 545pm final start.  However, I believe that the route density of 170,000 is overstated, because (1) this year's march was less dense than last year's, and (2) because the tail end of any march will always be less dense than the front.  So I will discount it by 10% to 350,000.  That is the first estimate.

Source #2, police estimates are typically under-statements, since they don't directly count people joining in and/or leaving in the middle.  But to make sure that they will not be short of resources, the police had already applied a 20% inflation factor to arrive at 200,000.  I will leave the estimate at 200,000.  That is the second estimate.

Source #3, the Ming Pao aerial photo analysis yielded 264,000 which I corrected down to 156,000 by reason of physics.  This method does not consider people who join in the middle of the march, so I will factor the point estimate upwards by 15% to get 179,000.  That is the third estimate.

Source #4, the Hong Kong University Department of Statistics and Actuarial Science used human counters at two locations at Causeway Bay and Admiralty.  This is supplemented by intercepts at Adimiralty to ask people if they had passed the Causeway Bay station.  The estimate is 165,000.  That is the fourth (and best) estimate.

Source #5, the Hong Kong Human Rights Monitor used human observers and counted 160,000.  This method does not produce cumulative size.  I will take the estimate of 160,000 and give it a 15% increase to 184,000.  This is the fifth estimate.

Solution #6, the Hong Kong University Social Sciences Research Center conducted a frame-analysis of video tapes taken at Admiralty and estimated between 105,000 and 120,000 marchers.  This is a strong validation of Source #5, which counted 110,000 by hand.  The number of marchers is estimated by using the other ingredients in Solution #5.  This is the sixth estimate.

Source #7, a team from Hong Kong University, Chinese University of Hong Kong and the Polytechnic University counted the number of marchers midway and then used a telephone survey to estimate the number of people who "departed before" or "joined after" the point.  The estimate is 195,000.  This is the seventh estimate. 

All seven published methods have some degree of plausibility as well as uncertainty.  I will not include any other estimate that have no disclosed reasonable basis.  For example, statements such as "I've been to music concerts with 200,000 people and this looked larger" have no place here --- what makes you think that the concert promoters weren't inflating their attendance numbers?  I will give these six estimates equal weight.

The average of the seven numbers is (350,000 + 200,000 + 179,000 + 165,000 +165,000 + 184,000 + 195,000) / 7 = 205,000.  The disclaimer is that this is an unscientific estimate that is every bit as bad as its components.

My methodology is similar to those used to forecast economic performance such as corporate earnings based upon pooling estimates from a large number of financial analysts.  Different analysts have access to different information, they have different opinions and attitudes and therefore they make different forecasts.  This doesn't make the average value the 'true' number, but it reflects the average opinion even though it not truly a consensus (in the sense that everyone agrees).

Some people have misunderstood that this number is somehow 'my' estimate.  Well, it is not 'my' estimate.  I wasn't out there counting, I did not conduct any surveys and I did not analyze photographs or videos, so I don't have my own estimate of crowd size.  The number above is a pooled average of other people's estimates.  If you don't like 'my' estimate, then you don't like one or more of these other people's estimates.  If you think that I am deceitfully understating the number, then you are thinking that one or more of these people are deceitfully understating their number(s).  You are going to have to decide for yourself if you trust these sources.

Please note that the Civil Human Rights Front's number now stands out at odds with the others which average (200,000 + 179,000 + 165,000 + 165,000 + 184,000 + 195,000) / 6 = 181,000.  This is no longer a situation of the Front vs. the police, for it is the Front versus the whole world.  And there is no reason to impugn the motives and/or professionalism of all of the other parties.

MY COMMENTS (July 8, 2004)

At this point, it is my turn to have a say.  What is the true number of marchers?  I don't know for sure and I don't really care.  Everything that I need to know is captured by photos such as these.  The exact number is unimportant.


So why did I spend so much time compiling this huge post?  Every mass event generates multiple estimates of the number of participants, and the differences among these estimates often reflect ideological biases.  That is my greater interest.  And this has always been a subject of interest to me and I treat these events as Rorschach tests in which commentators reveal their true colors.

In the Mexican march, the estimates varied from 25,000 to 2 million.  I don't know what the real number is, but I tried to track down who said what.  In the Venezuelan march, the government claimed 3 million, the fire department estimated 500,000 but Associated Press said 'tens of thousands' which leads to me to wonder what was going through the AP reporter's mind.  In all those posts, you should get the sense that I was really not concerned about the exact number and I did not say which number was the truth except I knew that there were lots of people out there.  

In Hong Kong, the estimates ranged from 530,000 down to fewer than 200,000.  The range is less than the Mexican and Venezuelan instances.  At this point, as I have promised you, I won't talk about what the 'real' number is.  But something different happened here.  Nobody else ever discloses in detail how they did their calculations.  This time, all but one estimation methodologies were disclosed and some unusual developments occurred.  And we have moved beyond the number itself to something more problematic.  

So what did I find this time in Hong Kong?  Let me begin with the principal actor on center stage: the Civil Human Rights Front.  Here is a review of the events:

At 920pm on July 1st, the Civil Human Rights Front issued a statement that claimed a total of 530,000 marchers.  The calculation method was later disclosed: the Front had obtained a police estimate of 170,000 persons on the route between the start and finishing points, and observed that the trip took 90 minutes and the march began at 230pm and ended at 800pm.  The Front then assumed that there would be enough time for 3 waves of marchers for a total of 3 x 170,000 = 510,000 and they added another 20,000 for people who joined in the middle of the march to get the final figure of 530,000.

Subsequently, a number of people (including me) pointed out that the Front had committed an arithmetic error.  There was not enough time for three waves of people.  The first person left at 230pm, but the last person must leave at 630pm in order to finish at 800pm.  Therefore, there was only 4 hours and not the 5.5 hours as the Front assumed.  This meant the number should be 4/1.5 x 170,000 = 454,000 and not the 3 x 170,000 = 510,000 that was used.

On July 6th, the Civil Human Rights Front went into a meeting.  They came out and declared, "There is no possibility of overturning the 530,000."  Instead, they came up with a new calculation.  Indeed, there was time for 4/1.5 x 170,000 = 454,000 persons.  However, they have decided that there were 76,000 persons who joined in the middle instead of the 20,000 that they said previously.  The final number is therefore 454,000 + 76,000 = 530,000 as before.  This number is said to be reliable, accurate, precise and all the other wonderful things too.

The Civil Human Rights Front 'cooked' the new formula to get the politically desirable result.  They needed to stay with 530,000, they were caught with making an arithmetic error and they 'fudged' another part of the formula to get there.  For me, the stock of integrity of the Civil Human Rights Front is nil.

It was no surprise that politicians would continue to use this fudged number for advantage.  Case in point:  Legislator Emily Lau published an article in the Taiwan Times on July 17, several days after the myth of the number was exploded.  It is impossible for her not to cognizant of the situation.

On July 1, half a million people braved intense heat and humidity to march for hours demanding direct elections in 2007 and 2008. The peaceful and dignified demonstration exploded the myth that Hong Kong people do not care about politics and democracy and that they are very pragmatic, meaning if a decision has been taken, particularly by the central government, they will not press the demands anymore.

Many people were stunned by the overwhelming turnout because the march had the single objective of fighting for direct elections, which Beijing has categorically rejected. Tung not only has a duty to reflect the people's concerns to the central government, but should persuade the leaders in Beijing to heed the Hong Kong people's wishes and aspirations.

To our dismay, Tung said he has checked with the central government and was told he has no power to reopen the issue, so he cannot make further representation to the National People's Congress Standing Committee (NPCSC).  This incident reinforces the widely held impression that he merely does what he is told by Beijing.

For me, the shining stars of this debate have been the academics.  The team from Hong Kong University ran the best methodologically designed study on their own time.  They did it out of intellectual curiosity and seeking the truth.  Many others made comments to the media and posted on the Internet.  These discussions were focused on the technical issues and relatively free of political bitterness.  The responses of the Civil Human Rights Front to these comments have sometimes been bitter, such as spokesperson Jackie Hung Ling-yu charging the critics with betraying the marchers.  The interview of HKU researcher Yip Siu-fei, who is a pan-democrat at heart, makes for heart-rending reading about someone who gets lambasted for only wanting to seek the truth.

For the media, all of them were quite careful in their initial reports by quoting both the 530,000 number as well as the police estimate of 200,000.  Some newspapers put the 530,000 on the front covers of their next day's editions and I don't blame them because the controversy had not surfaced at the time.  After the doubts were raised by various people, the only newspaper to actively cover this issue is Sing Tao Daily.  The pro-democracy cheerleader Apple Daily avoided the subject, and most other media dropped the ball.

On July 6th, there was an opinion column in Sing Tao Daily.  The author said that he had discussed the issues with friends in the media industry.  These people were skeptical of the 530,000 number initially but they went along with it anyway.  "This way, I won't get attacked by the pan-democratic camp or receive complaints from readers."  If this is true, then this is a rapid road to a tyranny of self-censorship.  This is every bit as bad as the late night phone calls that democrats supposedly receive.  In this sort of atmosphere, freedom of media is illusionary.

This issue is not going away, because there will continue to be references to this march for a long time.  Now that the Civil Human Rights Front has custom-fitted a new formula to preserve the old number, it is much more difficult for any self-respecting media worker to rationalize the 530,000 number.  Any media worker who continues to quote the 530,000 number is either lazy, or intimidated, or partisan.

As for the people of Hong Kong, there is no broad information available as yet.  It would be an interesting survey to ask them whether they thought the number of marchers was 530,000 as claimed or that it is less.  This is an important question because democracy is not just going into a booth to cast a vote on the basis of some campaign slogans.  Democracy involves more than that.  Democracy also requires that the citizens have to make an effort to understand the issues and hold the politicians according to standard of ethics.  If the people choose to ignore a blatant lie because of willful ignorance or because the fabricators have politically correct credentials, then they are descending down the road to fascism already.

The Civil Human Rights Front said that the people should decide which is the right number.  I happen to think the people don't have to decide which is the right number, because it doesn't really matter anyway.  We all know that there were lots of people out there.  But  I think that the people should decide whether or not this Civil Human Rights Front is a good representative of the people based upon their conduct in this case.  That is the more relevant test of democracy.

And there is also another political test.  The Standing Committee of the National People's Congress is no doubt watching.  Do you think that they would want to be seen as caving in to a falsely claimed demonstration of 530,000 people?  And if they see that the people of Hong Kong are willing to ignore a highly publicized blatant fabrication by the pro-democracy camp leaders, would they believe that the people of the Hong Kong SAR are politically mature and responsible enough for direct elections?

As in the Hong Kong Legco SARS report, I ought to name someone for special condemnation.  My nominee is Law Yuk-kai of Human Rights Monitor.  On July 1st, the HRM people did a physical count of the marchers and came up with a figure of 160,000.  This is consistent with the numbers from the police, the HKU study and the Ming Pao aerial photograph analysis.  Coming from the HRM, this is a significant validation of the numbers from these other souces since HRM is a member association of the Civil Human Rights Front and cannot be said to be biased against the Front.  When asked about how he could reconcile the HRM count of 160,000 with the CHRF claimed number of 530,000, Law's response was that he did not want to get tangled up in the controversy.  This was the same person who was so confident of his work at a previous march and dismissed an alternate estimate haughtily with "People like us who have worked on this regularly think this is hilarious.".  This is the united front mentality of Leninist/Stalinist/Maoist/Trotskyist parties --- nothing (not even the truth) must disrupt the appearance of unity, and integrity, conscience and the Hong Kong core values be damned.  To his credit, though, Law Yuk-kai did not attempt to claim that there were 370,000 people who joined the march at some point beyond their counting station.

Sing Tao Daily, July 8, 2004
Human Rights Monitor manually
counted 160,000 marchers that day