In the lead-up to the July 3 election, BP has blogged on a number of polls as follows:
- The E-san poll, as blogged about here and here, which surveyed all 20 provinces in the Northeast which looked at who those surveyed would cast their party vote for (Puea Thai 63.9%, Democrats, 20.7%, and Bhum Jai Thai, 9.1%),
- Nationwide NIDA poll which looked at who people would cast their constituency vote for (Puea Thai 23%, Democrats 20%, Bhum Jai Thai 3%, undecided 53%),
- Nationwide Suan Dusit poll which showed who people would cast their party vote for (Puea Thai 41%, Democrats 37%, Bhum Jai Thai 4% OR if you remove the undecideds and those who will vote no you get Puea Thai 45%, Democrats 41%, Bhum Jai Thai 4%),
- Bangkok-only DPU poll which showed that Puea Thai would win 19 constituencies, Democrats 5 and the rest were too close to call, and
- Nationwide Suan Dusit poll which showed who people would cast their party vote for (Puea Thai 43%, Democrats 37%, Bhum Jai Thai 3% OR if you remove the undecideds and those who will vote no you get Puea Thai 47%, Democrats 41%, Bhum Jai Thai 3%),
- Poll of the Lower North showing how people would cast their party vote for (Puea Thai 23.2%, Democrats 22.2%, 4.5% for other parties, and 50.1% were undecided)
BP: BP has posted about the accuracy of Thai polls and the problem of using a single poll – hence why we have the above refresher which will be included and updated for future posts on polls. Polls may be poorly worded and not so reflective of those who will vote on July 3, but they are no data points and together with other information we can use them to evaluate.
There is a new nationwide NIDA poll (PDF) that was conducted June 1-2, 2011. The poll surveyed 1,230 people nationwide. The survey data methodology is at the end of the post.*
Q1. Below is the breakdown for the question asking people how they would cast their constituency vote:
BP: This poll unusually includes those won’t vote and won’t say (which also includes unsure whether they will vote – so one can say they are the ‘unlikely’ voters) when providing the party breakdown so BP has created the below chart with those figures excluded:
BP: Then with the Vote No also excluded. All three figures were excluded from last month’s NIDA poll blogged about so hence must also be excluded if BP is to conduct a comparison.
BP: Now, compared with the NIDA poll conducted one month earlier which is below (BP has edited the chart format from the previous post to reflect how NIDA presents their numbers):
BP: Now you can see that Puea Thai has dropped from 23.36% to 22.45%, but the Democrats have dropped from 20.2% to 12.50%. Bhum Jai Thai has also dropped from 2.99% to less than 1%. Undecideds are up from 52.87% to 63.5%.
Q2. Below is the breakdown for the question asking people how they would cast their party vote
Then excluding those won’t say or won’t vote/unlikely to vote.
Then with Vote No also excluded:
BP: The previous NIDA poll did not provide a precise question about the party vote – it asked voters who they expected would win which is different from who people will vote for – so cannot directly compare but as you can see the party vote figures are not that different from the constituency vote EXCEPT that Bhum Jai Thai’s vote count is so small in each region that it is less than 1%.
1. The addition of the “won’t say” figure is interesting. In the Northeast for the party vote, those who won’t say is only 10% whereas it is 25% in the South.
2. BP has no idea why the undecided figure is so high. It is higher than other polls once you take out vote not, those won’t vote etc. It can’t be just because they are afraid to say as they have the ‘won’t say’ option – if they are afraid to say because they fear going against local opinion then choosing ‘undecided’ over ‘won’t say’ doesn’t make that much sense as to go with local opinion you have to choose that party. The % of undecideds has been high in other polls although not as high as this one. Part of the reason could be the wording of the questioning in some polls as if you ask if there was an election held today who would you vote for then the number of undecided falls based on previous polls. Perhaps, a working theory can be that those have decided are those who have indicated are committed to voting for the party + party candidate they have chosen and believe they won’t change their mind (of course, people can later actually change their mind).
3. Okay, Puea Thai has dropped slightly, but the Democrat vote percentage drop is dramatic – this is from NIDA too who would be considered one of the bastions of establishment – so along with the Suan Dusit poll showing Puea Thai gaining more than the Democrats this does show a trend. Another trend is the low numbers of people who have decided to vote for the third parties.
Bangkok and surrounding provinces 198 (16.1%)
Central Region 243 (19.76%)
North 222 (18.05%)
Northeast 392 (31.87%)
South 175 (14.23%)
BP: This seems to be in line with the population of the various regions. Understand that Central Region excludes Bangkok and surrounding provinces.
Under 25, 142 (11.54%)
25-39, 462 (37.56%)
40-59, 536 (43.58%)
60+, 90 (7.32%)
BP: As with most polls, this seems to under-represent those aged over 60. This is probably because these people are more likely to be at home are harder to survey.
Grade 6 or less, 383 (31.14%)
Grade 12 or equivalent, 383 (31.14%)
Vocational Certificate or equivalent, 107 (8.7%)
Bachelor’s degree or equivalent, 316 (25.69%)
Higher than Bachelor’s degree, 24 (2%)
Civil Servant/state enterprise, 162 (13.17%)
Private company employee, 221 (17.97%)
Business owner/freelance (includes self-employed), 313 (25.45%)
Farmers/laborers, 279 (22.68%)
Housewife/house-husband/retired/unemployed, 160 (13.01%)
Students, 95 (7.72%)
BP: Again, the “Housewife/house-husband/retired/unemployed” category probably under-represents this group and the student group is over-represented although this is the nature of most polls (i.e as they are mostly at home and hence more difficult to survey).
None, 243 (19.76%)
Less than 10,000, 552 (44.88%)
10,001-20,000, 282 (22.93%)
20,001-30,000, 78 (6.34%)
30,001-40,000, 35 (2.85%)
More than 40,000, 36 (2.93%)
Unspecified, 4 (.33%)
BP: BP hasn’t able to compare the performance of the poll with actual election results so won’t offer too much comment on the accuracy of NIDA polls, but the survey data is very detailed (more in the actual poll like marital status and religion) and NIDA have been very transparent in releasing it. There is the separate question of how reflective this is of the population – geographically it appears to be and aside from the problem of surveying those aged over 60 and retired/house-wife it appears on face value to be reflective.