How accurate are Thai polls?
Share this on

How accurate are Thai polls?

TAN Network has translated an editorial that appeared in Krungthep Turakit which when translated into English the title is “Opinion Polls Must be Properly Conducted”. Some key excerpts from the translation:

Some politicians have even been trying to push their own agendas by citing the poll results. However by looking at some of their questions, one has to wonder if these opinion polls are even properly conducted or legitimate. Many of the questions asked by the opinion polls seem to have been politically motivated.

We are not sure what exactly the real purpose of the opinion polls is other than to give us some rough assessment of which party is more popular than the other. Nevertheless, opinion polls are being conducted throughout the world; not only in Thailand but also in the rest of the democratic countries. The reputation of the polls whose results come closest to the reality would improve while whose results weren’t so accurate would be trying their best to explain what went wrong. Still, the questions over the accuracy of polling methodology are valid and is rarely discussed.

We support the opinion polls which are properly done and do not support those which are politically motivated and unable to distinguish right from wrong.

BP: Well, basically agree and BP shares concerns that over lack of questions over methodology – and also the wording of questions – and that is something that BP tries to bring when discussing polls, but BP thinks would have been more helpful if examples were provided as opposed to a general statements. Also, what kind of poll can help someone “distinguish right from wrong”? Such a poll would be extremely disturbing as the pollster would be making moral judgments…

Two weeks ago, VOA had a quote from well-known Thai politics scholar Chris Baker. Key excerpt:

Baker says it is still too early to see if the main parties will be able to win an outright majority.

There’s not really enough information available now to make any kind of prediction of the result,” Baker said. “Even if we start to get some better poll data we ought to be fairly sceptical because the polling there is not very professional and people tend to lie to polling officials. It will be very difficult to gauge the likely outcome right down to the day I think.”

BP: BP has the following comments:

1. BP thinks it would be unwise to see a single poll as being reflective of the vote.  A poll is a single data point. Multiple polls are multiple data points and in the last two weeks we have had multiple polls. Nevertheless, BP is of the view that you also need to examine the wording of the questions, look at the breakdown of who was surveyed (and whether this broadly reflects the population), and then lack of the accuracy of this polling company in the past.

2. BP has noted that often there is a slight Democrat-lean for polls – see also question 5 of this poll. The best example is excerpted below from a post last year:

NOTE: In case you are wondering whether there is still any basis for such a disclaimer, you just need to look at the by-election in Constituency 6 in Bangkok in the second half of last year. Exit poll data (over 5,000 were surveyed) indicated the Democrats winning by 12%. The actual result was the Democrats won by 8%. For a variety of reasons, most polls favour the Democrats by around 4-5% points. The discrepancy between the exit poll and the actual result in Constituency 6 has just confirmed BP’s view of this (of course, not every poll will be like this so it will not apply in every case).

However, BP has noted this is fairly consistent over time and BP will be looking forward to evaluating pre-election and exit polls with the actual result.

3. On people lying to pollsters, BP thinks it is common sense that this would normally help the government (i.e if they support opposition they may not admit this). However,  in areas where there are influential, long-established politicians and/or also multiple violent events then people would be less likely to tell the truth regardless of whether the person was an opposition or government MP.

Topics covered: