Statistics Behind the July 1 Protest

1
Statistics Behind the July 1 Protest
2
Methods Used to Figure Out the Number

• Estimate by the capacity of the route
• Estimate by head count method
• Estimate by fixed camera record

3
(by Civil Human Rights Fronts)
Estimate by the capacity of the route

• This calculation based on the area of the route,
  density and the total time for the journey
• Area of the journey 72000 thousands sq.meters
• Total time 5-5.5 hours, count at every 1.5 hours
• 1) The Civil Human Rights Fronts (from police
  estimation), density 2.36 _at_1sq.meter
• 5/1.5 720002.36
• 5/1.5170.000
• 530.000(approximately)

4

• 2) By Mingpao,
• the density is 1 person _at_ 1 sq.meter
• So, total people 5.5/1.572000
• 264,000

5
Comments

• density calculation problem
• From Civil Human Rights Fronts (police
  estimation), density 2.36 people _at_ 1sq.meter
  (HIGH)
• By Mingpao, density 1 person _at_ 1sq.meter
• Mingpao calculation of density is more suitable
• The Civil Human Rights Frontss estimation is
  over-estimated

6
Estimate by head count method

• (By Robert Chung Ting-Yiu)
• Estimated 180,000 to 210,000 people
• Method
• six separate head counts
• one-minute count in every five minutes
• total 149,000
• 77 participants walked passed check points

7
Comments

• Inaccuracy of estimate due to
• People who walk in a different and new path are
  not included
• Miscount using head counting (especially when
  people walk very fast)
• The people flow rate may change in different time
  slot. Not suitable to fix one-minute count in
  every five minutes
• The number is underestimated !

8
Estimate by fixed camera record

• Two cameras on Queensway recorded from 330pm to
  730pm
• Study 4 sec. of footage for every 30secs.

9
Assumption

• 1. The march route should be fixed.
• 2. The flow rate in each time period should be
  fixed.

10
Comments

• 1. Hard to determine the position and angle of
  the cameras
• 2. Not scientific enough to study 4secs. out of
  every 30secs.
• 3. Midway joiners not included

11
Survey to include mid-way joiners

• Ask protestors during the rally- dont
  know where they will leave
• Ask Protestors after finishing- some
  midway leavers will be ignored
• a community-wide random telephone poll
• Question 1
• Have you joined the rally? (if yes, ask
  question 2)
• Question 2
• During the march, did you pass through
  Hennessy Road road next to the Southorn
  Playground in Wan Chai? ( the camera position)

12

• Let N be the no. of interviewees join the rally
• n be no of the interviewees that pass the
  checkpt
• p be the sample proportion of people
  
  pass the checkpt
• 1-p is the sample proportion that leave
  before the checkpoint or join the rally after the
  checkpoint
• A be the no. of people counted at the
  checkpt on that day.
• A may either be found by headcount method
  or camera record method.
• To make it sample, we assume we have one checkpt.

13
The best method is
14
The best method

• More accurate
• Include both the people who joined from the
  beginning and in the mid-way

15
A better one

• Find out the no of people aged below 18 as well
• Record the whole rally at a checkpoint. After
  reviewing the tapes later, the sampling frame can
  be more appropriately decided according to the
  rate of people flow. For example, if the rate of
  change is fast, a sample of 1 minute from every 3
  minutes should be taken instead.

16

• Keep getting up-to-date news about the rally on
  that day, so as to get known and reanalyze in
  case there is a new path.
• More points should be set up to get more data so
  as to have comparison between findings.

17
Conclusions

• It seems to be no perfect solution to figure
  out the number of people joining the rally.
• We should not focus on the figure only.
• What we should do is to analyze the constraints
  and reliability of each methods, whether the
  hypotheses are reasonable.