Exit polls epic fail: Human error or divine intervention?
It's "Phew!" for the Democrats, "Aaagh!" for the Pheu Thai Party and "Nooo! Not again!" for those in charge of exit polls. The ruling camp is the loser, all right, but eggs are on the face of the pollsters who, many say, have never learned their lessons.What went wrong? To be fair, in Thailand most political surveys were once quite accurate. They could predict how many seats each party would get before a general election. In some cases, popularity surveys could tell how many parties would form a post-election coalition, or even name them.
To say that Bangkokians "fooled" the pollsters last Sunday may be partly right. The whole story, however, could be more complicated. I have talked to someone actively involved in one of the exit polls and my initial contempt turned to sympathy. It seems exit pollters had taken into account past mistakes but they didn't expect cruel divine intervention that came in the form of heavy downpours in various parts of the capital around noon.
"We extended the sampling period to around 1.30pm, and we asked those surveyed to tick questionnaires instead of verbally telling us who they voted for," said my source. That addressed two key issues - pro-Democrat middle-class voters who have a habit of going to the polls at or after lunchtime, and the shyness or reluctance of many Thais when it comes to speaking out loud about their political choices.
But who would have expected heavy rain in early March? I didn't, and neither did my source and her colleagues. Worse still, the showers came around noon and lasted for long periods in several areas. This meant that the benefit of extending sampling times went out the window. Either the polling stations were empty then, or the young volunteers were forced to take shelter, or both.
It became too late, too quickly. When voters re-emerged after the rain, exit-poll headquarters' were already processing what data they had, which was basically how the early birds made their choices. The rest, as they say, is history.
This rain theory, however, flies in the face of some reports that a certain exit poll did find that Sukhumbhand Paribatra would be the winner. It's the poll that, instead of declaring Pongsapat Pongcharoen the winner, like all others, told the public the race was "too close to call". This particular poll must be tearing its hair out right now for not being initially more forthright about what it had found.
If one exit poll, despite the obstacles, did find Sukhumbhand the victor, what does this say about the other polls? My source said it was possible some voters surveyed "cheated", either out of embarrassment about their choice or just for kicks. But were there so many cheaters out there that they could give a totally false impression of the eventual outcome?
I asked my source if it was also possible that some volunteers cheated as well. Her 500 university students were paid Bt500 each for the day's work, which was to ask 20,000 people leaving the polling stations to tick the questionnaires. Getting Bt500 to get answers from 40 voters is a fair deal, but is it possible that some got bored after 20 voters and headed off to Starbucks instead?
My source giggled. "I have friends who did that when they were young," she said. I told her all those cheating volunteers needed was prior knowledge of pre-election rating surveys, which had Pongsapat in either small or comfortable leads. They could have simply forged data based on those surveys and sent it back to headquarters.
Or some volunteers could have violated methodology. Instead of spreading out sampling periods as instructed, they may have opted for an easier way. People may have been interviewed in big clusters rather than a more random manner. Instead of sampling 40 people in five hours, some volunteers may have done so in two hours. A volunteer may have gone for a friendly-looking group and come back with a solid 10 votes for one candidate, for example.
Or the pollsters' methodology is far from perfect. There are totally pro-Pheu Thai zones, pro-Democrat zones and mixed zones. My source suggested that some academics and statisticians involved in the exit polls were reluctant to give too much importance to the national divide. They wanted to treat all zones more or less equally. That could mean instead of seeking 20 samples from an unpredictable mixed zone and five each from a "surefire Pongsapat" red zone and a "surefire Sukhumbhand" blue zone, they got 10 from each.
We shall not take anything away from Sukhumbhand. Most of the media had not been on his side. Many Democrat supporters had threatened to rebel against his nomination. The social media had ridiculed him. Popularity surveys had all but written him off. Even exit polls had apparently discounted him. He won against all odds.
But as far as Pheu Thai is concerned, whether Sukhumbhand triumphed on his own or because of a "fear factor" doesn't make much difference. Both theories sound gloomy for Pheu Thai. If Sukhumbhand won on his own, that says a lot about the government's attempts to make inroads into city management. If he won "in spite of himself", the message should scare Pheu Thai even worse.