They studied vote-buying, told stories about cliques and personalities, and described the intricate webs of money and influence needed to win a poll. But they paid little attention to the numbers, and the patterns in the numbers. The assumption was that elections were very local affairs. They depended on money, personality and networks. It was long considered a waste of time to apply sophisticated statistics of party performance and electoral behaviour in more settled democracies.

But this is changing a bit. Voting behaviour in the last three polls (2001, 2005, 2006) has been much more consistent than in the past.

Ordinary people's attachment to political parties has taken on some new colouring. People in the South say they support the Democrats because of "ideology". People interviewed in the Northeast during Thaksin's recent trip described themselves as "TRT people", and could not imagine voting for a Democrat candidate.

These are emotional attachments. People are identifying themselves with a party. Of course, such sentiments may easily be overwhelmed by personal connections, local pressures or money. But there are clear signs that party attachment has become more important. Also, there are hints that this kind of attachment can shift (by a few percentage points this way or that) on a national or regional scale. So the numbers are beginning to matter a little bit. What kind of swing does the opposition need to dislodge TRT? Here, it only makes sense to look at the Democrats. The new super-dinosaur party should achieve near-extinction; Mahachon will probably do no better than last time; and Chat Thai is planning to slither into bed with whoever wins. Compared to the results in 2005, what percentage of voters do the Democrats have to win over from TRT?

If 5 per cent of voters switch from TRT to Democrat, then the Democrats get only 14 more seats, giving a total of 86 out of 400. If 10 per cent, then 134. If 15 per cent, then 164. If 20 percent, then 197. In sum, the Democrats need to change the allegiance of one fifth of the voting electorate to approach an absolute majority. That's steep.

Where might they do it? The map shows the percentage of all votes cast polled by TRT on April 2, 2006. (The map is based on unofficial returns, misses a few constituencies counted late and fudges a few boundary changes, but the pattern is still clear) The stronger the colour, the tighter the TRT grip. Looking at this map, where should the opposition concentrate their limited firepower?

First off, forget the Central North from Sukhothai up to Chiang Rai, and the Northeast above Korat, except the provincial capitals. Don't waste a single billboard. This is a tough decision, given this area represents over a quarter of the electing nation, but a wise one. Winning this area would take years of investment, not weeks. TRT's soft spots are elsewhere. Greater Bangkok's electoral sentiment changes from the bull's eye to the outskirts. The central areas are secure opposition territory, especially since Thaksin called city people deluded idiots. Concentrate efforts on the outskirts which could go either way.

The western fringe has always been opposition territory, and TRT has never established itself. The Eastern Seaboard dramatically defected from TRT at the April poll. TRT has never looked solid in the bottom right-hand corner of the Northeast, and there are opportunities for the opposition all along the underbelly of this region. These are the areas where the opposition must invest.

The intriguing area is the central plain. TRT looks much less secure here than five years ago. Perhaps this is the area where this election will be lost or won.

Cang noi