Banning Exit Polls in Hong Kong

(The Standard)  Pan-democrats seek new shields at the ballot box.  By Una So.  March 11, 2008.

Pan-democrats and the Electoral Affairs Commission chairman will meet next week to discuss a series of violent incidents and irregularities that happened during last year's District Council election and Legislative Council by- election. The pan-democrats want to urge the government to revamp regulations to ensure a transparent and clean Legco election in September.

Thirteen groups conducted exit polls outside voting stations in November, but only Hong Kong University's Public Opinion Programme released results after voting ended. The pan-democrats suspect the pollsters were linked with political groups, using the exit poll results to strategically deploy campaign resources and manpower. According to the pan-democrats, some even claimed they were sent by the government. "We were using the single-vote system," said Frontier legislator Emily Lau Wai-hing, so "publishing data before the election closed will seriously affect the result. It's deplorable that polling officers decided not to do anything about it."

The pan-democrats want only recognized university institutions to conduct exit polls and to make it a criminal offense if results are leaked before the election is closed.

(SCMP)  Democrats urge restrictions on groups taking exit polls.  By Eva Wu.  March 11, 2008.

Pan-democratic lawmakers have proposed that only academic institutions from local universities be allowed to conduct exit polls on voters, to ensure fairness in future elections. Their call follows criticism of some organisations that were given approval by the Registration and Electoral Office to conduct exit polls during the district council elections in November. They were accused of providing the latest information to some candidates to enable strategic canvassing on polling day.

Legislators also pressed the government to make the publication of exit poll results to outsiders before the completion of polling a criminal offence.  "We suspect that some people use the findings before polling ends to strategically plan certain moves to enhance their chance of winning in elections," legislator Emily Lau Wai-hing of The Frontier said.

Democratic Party vice-chairman Sin Chung-kai questioned the background and affiliations of these organisations.  He said some people claimed they had been commissioned by the government to conduct the polls, but pressed further, they had been unable to give any proof that they had government authorisation to do so.

The lawmakers urged the government to amend election guidelines, demanding that those conducting exit polls publicise details of their background and state the reason for data collecting.

Democratic legislators and the political group Power for Democracy jointly released a report yesterday, covering some grey areas in use of exit poll results, the provision of free transport to polling stations and failure of polling officers to handle cases of breached election laws, which took place during November's district council elections and December's Legislative Council by-election. The report said 13 organisations or people, excluding the public opinion programme of the University of Hong Kong, did not release their poll results or academic analysis after the elections. At least five were found to be closely related to the government-friendly camp.

The Hong Kong Development Research Association, one of the approved organisations, was reported to be politically affiliated to the Hong Kong Youths Unified Association. The group is chaired by Benny Yeung Tsz-hei of the Democratic Alliance for the Betterment and Progress of Hong Kong, who won in the district elections. The report also said other organisations deployed 2,100 people to conduct polls outside 300 polling stations, while HKU only used 20 people to cover 27 polling stations.

Civic Party leader Audrey Eu Yuet-mee said under present guidelines organisations which published or broadcast poll results in an attempt to sway electors were merely given a written warning by the Registration and Electoral Office.

(SCMP)  Exit strategy for dubious election pollsters.  By Chris Yeung.  March 12, 2008.

... the publication of a report on alleged violations of election rules in the district council elections and Legislative Council Hong Kong Island by-election last year, and a set of corresponding recommendations, should not be dismissed as a political gimmick.

The report, compiled jointly by democratic legislators and the political group Power for Democracy, has highlighted irregularities and laxity in the existing election rules. It raises concerns about vote-canvassing in restricted zones around polling stations and the provision of free transport for voters on election day.

Referring to alleged abuses of exit polls in the previous two geographical elections, the report suggests banning non-university bodies from conducting such polls and to make it a criminal offence for groups to disclose their results before voting closes.

Concerns over the exit polls grew after the district council elections in November. It was revealed that 13 institutions and individuals were given approval by the Registration and Electoral Office to conduct exit polls. Only one of them, the University of Hong Kong's Public Opinion Programme, is a well-known, long-time pollster.

An investigation by Apple Daily found that five of the other polling agencies had connections with pro-Beijing groups.  Reports of their exit polls have not been made public. According to sources close to the pro-Beijing camp, they were well informed about the latest results of exit polls, particularly in some key constituencies.

For instance, Ip Kwok-him, a vice-chairman of the Democratic Alliance for the Betterment and Progress of Hong Kong, began canvassing for votes for other party candidates in the district polls midway through the day, after knowing he had enough votes in his own constituency. DAB sources also revealed that Regina Ip Lau Suk-yee's campaign team for the Legco by-election in December had access to exit poll results during election day.

The election watchdog urges pollsters not to divulge their exit polls before the end of voting, to avoid influencing voters.  As voting is conducted in strict confidence, there is no hard evidence to assess the impact of any leaked findings from exit polls on voters' behaviour and the results. Under the proportional representation electoral system for geographical constituency polls, it seems certain that political parties will adopt the so-called "multi-list" strategy in the September election.

Previous elections show it is harder to win a second or third seat in a list than to win one seat in each list. Putting candidates on separate lists will also help parties groom second- and third-tier candidates.  The risk, nevertheless, is clear. Votes for a political party may be unevenly spread; more may go to a list headed by a popular figure. For the "multi-list" strategy to succeed, parties may have to explore tactical voting for their supporters, and strive for an even distribution of votes for different lists.  Results of exit polls, if available during the voting period, could therefore become vital for the parties to direct their supporters to cast their ballots accordingly.  Should that happen, it would be unfair to candidates and parties who abide by the rules and to voters who turn out in the belief that the election is fair, open and clean.

(Wong Onyin's Blog)  March 13, 2008.

[in translation]

The democrats held a meeting and they wrote a report in which they offered some proposals to the Electoral Affairs Commission to improve the election process.  The most important one is the banning of polls outside the voting stations on election day with exceptions being granted to recognized university departments.  Any polling data needed to be kept secret until the election has been completed, and any leaking of such data would be a criminal offense.  These proposals are directed at the powerful and resourceful DAB party, although the Liberal Party also has the ability to do so.  In any case, the democrats do not think that they have the money to play this game, which is unfair to they who are the purveyors of truth.  Therefore, they are hollering to ban the exit polls which are helping their opponents.

Exiting polls may or may not be unfair, but it is certainly true that the democrats are being na´ve.  When Chris Patten designed the single seat system as the rules of the game, the pan-democrats with 60% support won a landslide and the DAB with 30% support won only one seat.  Was that fair?  The British obviously said that it was fair, because the single seat system is also used in the United Kingdom.  So what is unfair about it?

Today, who is in charge?  It is the democrats' turn to adapt to the rules of a new game and not hope that the powers-that-be modify the rules to suit them.  The return of Hong Kong to China took place ten years ago, and how can they be still so naive?  It is naivetÚ plus laziness, because they only know to pick up advantages or complain about the unfair rules of the game.  They won't work hard and seek to adapt.  This is the root cause of their failures.

In any election, there will always be one thing or the other that is unfair.  Afterwards, people point fingers and throw accusations at each other.  As for the voters, they just find you very annoying.  For example, the DAB was supposed to have conducted exit polls through five different organizations.  After they got the data, they adjusted the allocation of resources and canvassing efforts.  With these informational advantages, they could work more efficiently.  This is why the democrats are saying that it was unfair and they are now demanding that such activities be banned.  Fairness is when nobody knows anything about voting trends.

But if the DAB can mobilize two or three thousand people, so can you.  What is so unfair?  You complain that they have the money to hire people to carry out the exit polls.  What if they organized unpaid volunteers to do the work?  Can you still criticize them?  You are the ones who are too lazy to organize volunteers.  More than 100,000 people have marched in the streets to support the pan-democrats.  Why is it so hard to organize two to three thousand volunteers?  You need the will to do it.  Instead, it is a lot easier for you to accuse other people of holding advantages which you want to ban.  Why don't you reflect on whether you have done your best?

In the District Council elections, the DAB's electoral tactics were clearly superior to the democrats, and they had several times the manpower at their disposal.  But the pan-democrats lost because public opinion is shifting.  The voters were disappointed with the democrats, who are losing support.  It is regrettable that the democrats will not face up to their defeat and admit that they had let their supporters down.  Instead, they whine and complain, and blame everything on the fact that the DAB has the money and the people.

... Lau Kwong-wah asked how Martin Lee could have issued an emergency appeal for votes during the 2004 Legislative Council elections unless he had exit poll data?  Haven't the democrats also conducted secret polling?  In the end, even if the pan-democrats can achieve the banning of exit polling, does this mean that they will gain the much more important popular support?


ESWN Commentary:

In reading the above articles, the impression was that it requires 2,000+ persons to conduct these exit polls in Hong Kong.  Since many of the political parties and candidates do not have the means of mobilizing this many number of persons (either as employees or volunteers), the situation confers informational advantages to those organizations with the money and manpower at their disposal to do exit polls.  This is the reason why a ban is called for.

But wait!  There is also this statement: "other organizations deployed 2,100 people to conduct polls outside 300 polling stations, while HKU only used 20 people to cover 27 polling stations."  By all accounts, the work of the Hong Kong University Public Opinion Programme is respected and considered valid and useful, to the point that university departments will be exempt under the proposed law.  How come they can get that level of quality data with just 20 people?  How can 20 people do the work of 2,100 people?

Once upon a time, I used to work for a company name named Audits & Surveys.  What is the meaning of the company name?  

Suppose you want to find out about the quality of bookkeeping at a corporation.  The brute-force method is to check every invoice that ever came through.  In practice, this is neither feasible nor necessary.  Instead, you take a sample of the paperwork and you check that sample only.  This is called an audit.  The results are not perfect and are subject to sampling error, but a carefully chosen sample will yield sufficiently precise estimates.  You don't need to know that there were 162 errors out of the 35,924,221 invoices examined; you only need to know that the error rate was 0.01% (plus or minus 0.005%) out of the 2,000 sample invoices.

Suppose you want to determine public opinion on an issue.  The brute-force method is to contact every person in the country and ask.  In practice, this is neither feasible nor necessary.  Instead, you take a sample of the people in the country and you ask that sample only.  This is called a survey.  The results are not perfect and are subject to sampling error, but a carefully chosen sample will yield sufficiently precise estimates.  You don't need to know that 83,238,782 out of 227,684,831 American adults think Saddam Hussein was the mastermind behind 9/11; you only need to know that 39% (plus or minus 2% )of the 1,000 sample adults think so.

The reason why this company exists is that it is neither feasible nor necessary to conduct Enumeration and Census to obtain information.  Instead, Audits and Surveys are precise enough to yield actionable information.

In Hong Kong, there are hundreds (if not thousands) of election stations dispersed across the territory.  It is neither feasible nor necessary for pollsters to send people to work at all of them for the entire day.  It is sufficient to select a sample of election stations and work there.  That is why the Hong Kong University Public Opinion Programme only needs 20 people while still coming up with good quality data.  You don't need to know that Anson Chan was getting 54,585 of the 98,965 votes that were cast by 5:00pm; you only need that she is getting 55% (plus or minus 3%) of the exit interviews so far.

As for the organizations out there which mobilized 2,000+ persons to do exit polls, the kind thing to say about them is that they are being inefficient and the unkind thing is to say that they are being stupid.  If I were in charge of resource allocation over there and I had 2,000+ persons under my direction, I would send only a couple hundred of them to work at a sample of election stations.  They can use mobile telephones to sent the counts back to the headquarters periodically (e.g. every 30 minutes).  I will bet that the information is every bit as good as having 2,000+ workers out there.

What would I do with the 2,000 - 200 = 1,800 workers who no longer have to do exit polls?  They are mine to direct (either because they are volunteers or employees) and I would send them out to visit or make telephone calls to selected people (such as a list of all those who have ever come down to the office of a legislative councilor or district councilor of my party for assistance) to remind them to vote.  This is the so-called Get Out The Vote (GOTV) campaign.  This type of effort will have a much large impact on the final vote than any exit polls will.  Exit polls allow you to allocate your resources more efficiently, but you still have to get out the vote.  If each of the 1,800 workers can make 100 contacts each, that would be a total of 1,800 x 100 = 180,000 contact points.  Even at a 10% conversion rate, it is 18,000 extra votes.

There is no way to ban GOTV efforts, because this is already standard and accepted practice in Hong Kong.  On election day, it is common for the candidates to knock on doors with the media in tow.


(09/05/2008)

(The Standard)  Poll guru in threat to quit.  Diana Lee and Adele Wong.  September 5, 2008.

Hong Kong's top pollster says he might boycott the next election if the professional standards of polling organizations do not improve. Robert Chung Ting-yiu also blamed a lack of mudslinging in the Legislative Council election campaign for putting exit polls in the spotlight this time round.

He was speaking at a forum organized by Hong Kong Human Rights Monitor in Wan Chai yesterday. Representatives of 17 institutions that conducted exit polls in 2004 or at last year's Legco by-election were invited, but only Chung turned up. After saying there had been a "decline in professionalism" among polling groups, Hong Kong University's Chung added: "Maybe there's not much mud-slinging in this election, so the focus is on the exit polls."

This whole exit poll issue is a false issue designed to distract.  First of all, how hard is it to conduct an exit poll?  In (SCMP)  Democrats urge restrictions on groups taking exit polls.  By Eva Wu.  March 11, 2008. (no link, of course): "... organisations deployed 2,100 people to conduct polls outside 300 polling stations."  If true, then this favors the big political parties that have the money and manpower.

What is the reality?  The gold standard here is the Hong Kong University Public Opinion Programme.  Here is their manpower requirements for the 2004 Hong Kong Legco elections (see HKU POP):

20 stations, 20 interviewers: Hong Kong Island
16 stations, 16 interviewers: Kowloon East
13 stations, 13 interviewers: Kowloon West
24 stations, 24 interviewers: New Territories East
28 stations, 28 interviewers: New Territories West

6 more floating interviewers across all districts

101 stations, 107 interviewers across the whole of Hong Kong

By all accounts, everybody believes that HKU POP get good quality and reliable data with these sample sizes.  So who needs 2,100 people to conduct polls outside 300 polling stations?

A candidate runs in one (and only one) of the five districts.  How hard is it to draw a sample of 20 stations and send 20 volunteers/paid workers to work for one day?  This is readily feasible for most of the major political parties.  If a candidate cannot afford to do so personally, isn't it possible to pool resources within the same camp?

The reality is that the answer is a resounding NO.  The candidates within the same camp are mostly bent on cannibalizing each others' votes because they have little or no hope of drawing votes from the other side (unless they mask their true identifications carefully).  None of them are willing to join in a consortium and then agree to transfer votes amongst themselves based upon exit poll results.  Everyone wants every vote for himself.  So rather than deal with this reality of their perfectly understandable selfishness, the candidates attack the exit polls.

Actually, I really don't care about the exit poll results, because I think that this is very much missing the point.

There is an effort to organize voters to either refuse to cooperate with the exit pollsters, or else tell them their vote was for the DAB (see, for example, Those Were The Days).  To my mind, this is very much missing the point.  The issue is not about the exit poll results.  The issue is really about what happens next afterwards.

Let us say that the exit poll data are made available to everybody.  What, if anything, happens?  There are two ways in which something happens. 

First, there are some forces with immense manpower which can mobilize tens of thousands of volunteers who visit homes or make telephone calls to make crisis appeals for their candidates.  "XXX is falling behind and needs your vote!"  The advantage goes to these big parties.  These forces may have multiple candidates in the election and the exit poll boycott may misdirect their allocation of effort (as in voting candidate XXX instead of YYY of the same camp).  But this cannot negate their huge advantage.  And that is the true issue.  With or without the correct exit poll data, tens of thousands of people will be working for them out there.  How can candidates with lesser resource counter-act that?  In Banning Exit Polls in Hong Kong, I argued that it was stupid to send out "2,100 people to conduct polls outside 300 polling stations" because the job can easily be done by a lot fewer people.  But if exit polls are banned or crippled, those 2,100 people would be send out to solicit votes instead and the impact on the election would be a lot greater.

Is it right for some political parties to possess such tremendous resources?  That is hard to say.  Were they bankrolled by the central government? Or 'hostile foreign forces'?  In the United States, it has always been the Democrats and the Republicans, with a lot of minor parties.  It is impossible to regulate party resources to ensure equitable competition among all candidates.

The second way by which things can happen is through the media.  On the day of the 2007 Hong Kong Island Legco by-election, Apple Daily printed a freely distributed extra edition for a crisis appeal on behalf of Anson Chan in order to save democracy.  This has been regarded as a pure political ploy (not based upon any exit poll data) by a media organization, which is not illegal per se.  If and when the exit poll data are actually published for the general public to see, what will happen?  All the candidates will be able to spin their cases.  The leading candidate will say that a few extra votes will get the second member of the list in; the chasers will say that they can take over the lead with a few extra votes, or that they may lose if they miss a few votes; the laggards will say that a few more voters will get them in.  It will be just a cacophony of opinions to suit personal agendas.

When all said and done, the election results will be turned in.  What if the exit poll data are different?  Well, the exit poll could be flawed and they should be shelved forever afterwards because they mislead voters.  Or the exit polls are correct at the time (say, 12 noon) and the voters changed their behavior out of strategic voting afterwards (late in the day).  Neither case is desirable if you want an election in which the voters consider the attributes of the candidates in an objective manner and vote accordingly.

So should exit polls be banned altogether?  I believe that this a total misunderstanding of the purpose of exit polls.  In the western world, exit polls are not used to direct and mobilize GOTV (Get Out The Vote) efforts.  Rather, they are used afterwards to analyze why the voting occurred in certain ways.  For example, candidate XXX won through huge majorities among the women, the middle-class and the politically independents.  This is supposed to provide feedback about the true public concerns of the voters as manifested by their votes.  In Hong Kong, much of those types of analyses are buried in academic reports issued much later than the elections themselves.


(SCMP)  The exit poll as an election tactic.  By Yau Chui-yan and Daniel Sin.  September 7, 2008.

In the last week of a relatively uneventful Legislative Council election campaign, the most controversial topic has not related to any candidate but to the credibility of exit polls.

Exit polls - surveys of voters outside polling stations - are a normal part of polling day in most electoral systems. Voters are asked for whom they have voted and to provide some personal data. Their answers are used to project which candidates might win before voting ends.

These polls - considered valuable tools for studying elections since they may reveal how factors such as demographics and location affect voting patterns and preferences - are not new to Hong Kong. And since the first direct elections for legislative seats in 1991, they have always been run by Robert Chung Ting-yiu of the University of Hong Kong's public opinion programme.

In the past, Dr Chung passed the projections to media sponsors at 9pm to help them prepare their coverage, but they would not be broadcast to the public until 10.30pm, when the polls close.

However, the exit polls suddenly became a bone of contention on Tuesday when Dr Chung announced to the media that he was going to provide results to the five broadcasters sponsoring the poll as early as 12.30pm today.

Although he assured the media that only information about which candidates had more chance of being elected would be released, some political parties and media complained about the possibility of leaks.

They were worried that the exit-poll results could affect voting behaviour and discourage voters from turning out if their preferred candidate was not expected to win or was winning easily. Another concern was that candidates who had access to exit-poll data might be able to adjust their electioneering strategies, and in doing so gain an advantage over those without such access.

Dr Chung said such a scenario was <147,1,0>unlikely as there had never been a leak of data from within the media organisations sponsoring the polls. None of the sponsoring media said they were going to share the results with any candidate, but that did not dampen the controversy.

Pan-democrat groups held a joint news conference on Wednesday to remind voters that they could refuse to talk to the exit pollsters. They wore masks with black tape to urge the public to refrain from taking part in exit polls.

The pan-democrats said they were not targeting Dr Chung, as they had decided on their position long before his announcement.

Democratic Party vice-chairman Sin Chung-kai said: "We cannot accept people masquerading as [workers for] academic institutions, who are actually using data to help parties allocate votes to different candidates."

Under Hong Kong's fragmented electoral system, with parties running several candidates in each seat, it is in the parties' interest not to have one candidate do too well at the expense of his or her colleagues. Exit-poll data allows the parties to switch votes from one candidate assured of being elected to another whose victory may be in doubt. Ironically, the situation backfired for the Democratic Party in the last election, when a "situation critical" message diverted votes to Martin Lee Chu-ming and cost fellow pan-democrat Cyd Ho Sau-lan re-election.

After Wednesday's press conference, Dr Chung announced that a new agreement had been reached with media institutions and he would not provide the information until 8pm. However, the storm did not subside.

Joseph Cheng Yu-shek, a political-science professor at City University and secretary general of the Civic Party, said: "The time of the release of the poll results is not the key to the problem. It is not a solution at all."

The pan-democrat camp asked the electoral authorities to mandate that only academic institutions be allowed to conduct exit polls.

Under the rules on election-related activities issued by the Electoral Affairs Commission, any person or organisation wishing to conduct exit polls must apply to the Registration and Electoral Office at least 10 days before polling day. The office will then "consider the application and issue approval as appropriate".

But the Registration and Electoral Office declined to provide any information on the criteria by which an application is assessed. Seven organisations submitted applications to conduct exit polls on election day. All but one were approved, the office announced on Friday. The seventh had withdrawn its application, a spokeswoman for the office said.

In the 2004 Legco election, seven organisations conducted exit polls. Three of them, the Tsuen Wan Youth Council, the Hong Kong Policy Research Institute and the Hong Kong Youths Unified Association, have confirmed that they will not conduct polls this year.

The rules also specify that exit pollsters may not announce the results of exit polls or make specific remarks or predictions on the performance of an individual candidate of a geographical constituency list before the close of polling.

But according to Mr Justice Pang Kin-kee, chairman of the Electoral Affairs Commission, "announce" means holding a press conference or disseminating the results through the media. The only sanction that can be imposed when an exit pollster contravenes the rules is an open reprimand or censure.

Here is "the key to the problem" about which pan-democrats complain. They believe the regulation of exit polls is too loose and allows candidates or political parties to set up ad hoc research bodies to conduct exit polls so as to collect the data to help them co-ordinate vote-canvassing activities.

Richard Tsoi Yiu-cheong, spokesman for Power of Democracy, said it was impossible to believe that the exit polls were for academic use only. "The HKU team was usually three to four researchers in a few voting stations. As this was for academic research, having about 100 stations and randomly selecting voters was fine for academic use.

"However, other exit pollsters have up to 10 people at all 500 voting stations. They ask all voters and then key in details to a PDA [personal digital assistant] immediately," Mr Tsoi said. This data can be used by the Beijing-loyalist camp, he said, to ensure its candidates win more seats by asking supporters to reallocate votes from one candidate to another.

The biggest pro-government party, the Democratic Alliance for the Betterment and Progress of Hong Kong, denied commissioning any institutions to conduct exit polls. A member of the DAB's New Territories West division, Au Yeung Po-chun, said she was not aware her party had conducted exit polls. "The cost ... would easily exceed the election budget," she said.

The concerns might not be unfounded, however. A former Kwai Tsing district council member, Ting Yin-wah, observed that it was not uncommon for political parties to commission ad hoc organisations to conduct exit polls and to use the poll data to adjust their electioneering activities. "It is almost an open secret," Mr Ting said.

In a letter from Dr Chung to the Legislative Council in March about exit polls, he wrote: "It is no longer a secret that candidates and political parties in Hong Kong use exit polls for their election engineering. As early as December 5, 1993, I saw a political party running a large-scale exit poll at a district board by-election."

Pan-democrats have called for tightening of the current regime and were disappointed when the administration was not forthcoming. Secretary for Constitutional and Mainland Affairs Stephen Lam Sui-lung said the government "should not and will not regulate the internal use of research data" as there was no legal basis to do so.

Dr Chung told the Sunday Morning Post (SEHK: 0583, announcements, news) that people were approaching the problem in the wrong way.

"I do not object to any political parties or other agencies conducting exit polls, but I disapprove of `research agencies' using dishonest means to gauge voters' opinions for their election engineering."

This was very bad from the point of view of professional practice, he said.

Even before today's vote, there was a warning from academics that public trust in election surveys had been compromised by the exit-poll controversy. The response rate to a survey conducted by Chinese University fell by 10 per cent.