Socially Desirable Responses in Surveys 

With respect the election poll results, one doubt would be whether the stated preferences would be backed up by actual voting.  This is problematic when one candidate is expected to have a higher turnout rate among his/her supporters.  That is the reason for the various "emergency calls" for the Hong Kong Island Legco By-Election this week.  Another doubt about the poll results has to do with respondents offering socially desirable responses to the interviewers.  Here is some evidence taken from the Lingnan University Public Governance Programme:

Q1.  Will you be voting in on the December 2nd Hong Kong Island Legislative Council by-election?
54.3%: Definitely will
28.0%: Will
  3.4%: Will not
  5.4%: Definitely will not
  8.5%: Undecided
  0.3%: Refused to answer
 
[note: the last Legco by-election had a 40% voter turnout]
 
Q2. If the Legco by-election were held tomorrow, which candidate would you vote for?
39.9%: Anson Chan
22.6%: Regina Ip
  2.2%: Other candidates
23.5%: Undecided
  0.9%: Null ballot
  9.9%: Refused to answer
 
Q3. Did you vote in the 2004 Legislative Council election?
59.9%: Yes
25.7%: No
  2.6%: Not eligible at the time
11.2%: Forgot already
  0.6%: Refused to answer
 
[note: the overall turnout rate for the Hong Kong SAR was 55% in 2004 and the turnout rate for the Hong Kong Island Legco by-election was 40%]
 
Q4. In the 2004 Legislative Council election, which list or political group did you vote for?
35.4%: Democratic Party, Article 45 Concern Group, Front Line, etc.
  8.6%: DAB, Confederation of Trade Unions, etc.
  5.3%: Rita Fan
  0.3%: Liberal Party
  0.8%: Others
  0.3%: Null vote
45.2%: Forgot already
  4.3%: Refused to answer

[note: DAB/CTU/Rita Fan received 39.6% of the votes compared to 59.4% for DP/Article 45/FL]

It is socially desirable to state:

(1) that I am a good citizen who will be voting this time
(2) that I will vote for 'democracy'
(3) that I voted last time
(4) that I voted for 'democracy' last time (or, at least, I have a case of amnesia)

Further evidence for (1) above is given by the Hong Kong Research Association:

Q4.  Will you be voting at the Hong Kong Island Legco by-election?
81%: Yes
  3%: No
16%: Undecided

Based upon the historical record, there is no chance for an 81% voter turnout.  So if this response is dubious, how much credence would you offer to this question from the same poll?

Q2. If the Hong Kong Island Legco by-election were held today, whom would you vote for?
46%: Anson Chan
40%: Regina Ip
  4%: Other candidates
12%: Undecided

Further evidence for (2) above is given in the HKU POP pre-election poll results versus the actual outcomes in the 2004 Legco elections:

• Hong Kong: Ma Lik (DAB): HKU POP 15.5%, actual 21.1%.  Relative shortfall of 27%.
• Kowloon East: Chan Kam-lam (DAB): HKU POP 13.3%, actual 18.8%.  Relative shortfall of 29%.
• Kowloon West: Tsang Yok-sing (DAB): HKU POP 19.5%, actual 27.1%.  Relative shortfall of 28%.
• New Territories East: Lau Kong-wah/Li Kwok-ying (DAB) : HKU POP 17.3%, actual 22.1%.  Relative shortfall of 22%.
• New Territories West: Tam Yiu-chung/Cheung Hok-ming (DAB): HKU POP 18.7%, actual 24.9%.  Relative shortfall of 25.0%.

HKU POP has an analysis of voting intentions by political preferences:

If you have a political preference (either 'pro-democracy' or 'pro-China'), then you are almost 80% or better certain to vote for the candidate of that side.  If you do not have a political preference, then you are approximately 35% to 40% likely to vote for one or the other candidate.  So the outcome of this election depends on the turnout of voters with political preferences, since it will be a wash among those without political preference.  Note that there are two factors: voter turnout by political preferences.

For this analysis, HKU POP does not provide estimates of the sizes of the 'pro-democracy' and 'pro-China' camps.  Even if these size estimates were known, we know from the above that (1) the "pro-China" camp are less likely to identify themselves to interviewers and (2) the voter turnout rates are uncertain.