Back

Note: This article represents the view of the author and not the University of Hong Kong.

To Hong Kong people, participating in processions is not only a channel to express opinions, but also a kind of culture formed gradually, with an emphasis on rationality and order. The Central Policy Unit has commissioned a professional research agency to count the number of participants in the April 11 Procession. This has aroused public concern and discussions. The way used in counting the number of participants was one of the questions in the discussion. Although the motive and the release method of the research may bear political considerations, the counting method itself is a scientific topic, which should be handled by objective and scientific methods.

Since the route, the speed and the in-and-out flow of the mass will vary according to the changes of the environment and time, this poses the greatest difficulties in counting the number of participants. If a random sampling method is used without mastering the above information well, the estimation will become rough. In fact, the even more basic question is, how to define "participants of processions"? Some people left after waiting for too long at the starting point. Some people joined in the mid-way to avoid the crowd. Some people left the mass due to many other different reasons. Should they be counted in the number of participants? In fact some discrepancies among different counting figures might have arisen simply due to different definitions of "participants". For example, if "fixed-point" counting is used in counting the number of participants, it has actually filtered out those who left before passing, or joined in beyond, certain counting stations. Increasing the number of counting stations will also not solve the problem arising from people moving in and out between check points. This method only produces under-estimates.

As a matter of fact, the procession itself is only a channel to express opinions. If we, only for the sake of convenience in counting, eliminate those participants who have not passed the counting stations or reached the destination, or assume the number of such kind of participants at will, they are both unreasonable and unscientific. Of course, how to adjust the number depends on the size of such group of people, and the size of such group of people in turn depends on the total number of participants in the processions, the arrangement of the routes, the crowd control measures, as well as the actual situation at that time. Take the July 1 Rally as an example, the writer believes that appropriate adjustment is necessary in order to arrive at the total number of participants.

How can we count the number of participants in processions moving in and out from the mass? Take the commonly-used approach of "fixed-point" counting as an example, the total number of participants can be calculated by multiplying the number of participants passed the counting station by an adjustment factor. The factor actually represents the total number of participants in relation to each participant passing the counting station. For example, if the study reveals that for every 100 interviewees who participated in a procession, 80 of them passed a particular counting station, the adjustment factor would be 1.25 (or 100/80). The problem is, if the survey is conducted during the procession, the interviewees cannot predict where they will leave. Even if the survey is conducted at the end-point, only those who join the procession in the mid-way and leave at the destination can be covered. Those who join or leave before reaching the destination will still not be covered. If we need to count the number of such kind of participants, a random sampling survey will be inevitable. Then, the question is whether the statistics obtained are reliable. We will worry that the interviewee will give us false information, for example, those who have not participated in the procession will claim they have. However, when being asked if one has passed a particular counting station, there should not be any systematic bias because some liars will say they have while some will say they have not. Even if there is a systematic bias due to untrue answers, based on a common-sense deduction, those who lied they have passed a particular station should be more than those who lied they have not. Therefore, this will only lead to a smaller multiplier and make the estimation of participants conservative.

One academic thinks that it is necessary to provide the estimated error in calculating the number of participants. The writer of course does not object to this. But if the Poisson Process is used to simulate the flow of people, we will need concrete evidence to support the assumption that the number of people passing through a checkpoint within a fixed period of time is in fact equal to the variance adopted by the model. As to the method of "capture and recapture" suggested by the academics, to estimate the number of participants moving in and out of the mass, it has a lot of technical problems, and its feasibility needs to be further examined.

In fact, the choice of different counting methods depends very much on the total number of participants in the procession, the arrangement of the routes and the utilization of resources put on statistics. We should understand fully the limitations of each method and be open-minded to compare and discuss different methods. It is really unnecessary to focus merely on the figures, and neglect the real meaning behind.