Questions have been raised about Suan Dusit polls and whether they have deliberately biased in favor of the government – see here and here. The exit polls in the 2011 election, particularly for Bangkok constituencies, were completely wrong, but less so for the party vote. Interestingly for the party vote in the 2011 election,the pollsters overestimated the Puea Thai vote, but they got the Democrat party vote quite accurate.*

For this post, BP will look at the final Suan Dusit polls for the 2000, 2008, and 2009 gubernatorial elections as well as exit polls to see how accurate they were. BP has focused on Suan Dusit as they have come under criticism, but also as they have an archive of all their polls dating back more than 10 years. For other pollsters, particularly ABAC, they don’t.

NOTE: The reason that BP hasn’t included polls for 2004 was that the Election Commission banned them! See Suan Dusit note on this here.

For the July 23, 2000 election, the last Suan Dusit Poll was conducted on July 16.

Microsoft Excel

Source: Actual Results from BMA; Suan Dusit Poll

NOTE: Weighted = undecideds and won’t say removed and apportioned equally to other candidates.

BP: Generally accurate when you consider the number of undecideds although the major candidates increased their vote total from the other candidates.

For the October 5, 2008 election, the last Suan Dusit Poll was conducted on September 18-21.

Microsoft Excel

Source: Actual results from Wikipedia; Last Suan Dusit Poll; Exit Poll

BP: This time the last poll had the Democrat candidate getting more than 10 points than they got in the final result and had the pro-Thaksin party candidate getting 10 points less than they did (predicted a 44 point gap versus an actual 20 point gap).  The exit poll was closer, but still overestimated the Democrat candidate by just over six points and underestimated the pro-Thaksin party candidate by two points.

For the January 11, 2009 election, the last Suan Dusit Poll was conducted on December 26-29.

Microsoft Excel

NOTE: Should be 45.41 for Sukhumbhand; not 45.47

Source: Actual results from Wikipedia; Last Suan Dusit Poll; Exit Poll

BP: Again, the Suan Dusit poll underestimated the support for the pro-Thaksin party candidate by nine points (last poll) or three points (exit poll). However, in this case the last Suan Dusit poll also underestimated the Democrat party’s candidate by five points (last poll), but got the exit poll roughly accurate.

Some comments:

1. As you can see, the Suan Dusit poll have previously always underestimated the pro-Thaksin party. The Democrats fared better particularly in 2000 and 2008 last polls and 2008 and 2009 exit polls.

2. One reason for underestimating the pro-Thaksin party support in the past may have been demographics. Traditionally, the pro-Thaksin party candidate’s support has been from lower socioeconomic groups which can be difficult to poll but post-2009 you have the rise of the reds (which in Bangkok includes large numbers of the middle-class) so while there long seemed a problem in accurately polling the votes for the pro-Thaksin party candidate this doesn’t seem to have been a problem in the 2011 election.

3. Now, Suan Dusit may have realized by 2011 that they were getting the pro-Thaksin party candidates vote inaccurate so needed to adjust their demographics they were choosing so not problems in 2000, 2008, and 2009, but they over-corrected and got problems in 2011 not taking into account what BP notes in the rise of the reds and more middle class for Puea Thai.

4. If Sukhumhand is within five of the final polls he can certainly win as the polls have only proved roughly accurate.If  it is more than then points, it will be difficult for him as he can only depend on the polls being very wrong.

NOTE: It seems fair to criticize pollsters for getting demographics wrong, but this is different from implying they have been paid to be biased.

*For the party vote in the 2011 election, the four pollsters said:

Puea Thai would win 64, 66 (Suan Dusit), 65, and 67 seats respectively for an average of 65.5 seats.  The actual number of seats won was 61.

Democrats would win 40, 43, 45 (Suan Dusit), and 45 seats respectively for an average 43.3 seats. The actual number of seats won was 44.

Suan Dusit’s large poll surveying over 100,000 people between June 4-18 had (when excluding the ‘no vote’ and undecideds) Puea Thai winning 53.58% and the Democrats 35.39%. The actual result was Puea Thai 48.42% and the Democrats 35.15%.

As you see, they overestimated Puea Thai by around 5 %, but got the Democrats fairly accurate.