The size of the council of a subdistrict administrative organization (TAO) varies a lot, because every administrative village within the TAO has two elected councilors. Only in case there are less than three villages, the number of councilors per Muban is higher since the minimum size of the council is six. Thus there are TAO councils between six and more than 50 members. But not only the size of each TAO council differs, also the size of one TAO council can change - whenever a new administrative village is created, the council will be enlarged by two seats. The councilors from the village which is split are assigned to that part village in which they live, so its no necessarily the seats from the new village which will be filled in the by-election.
Now I add the council term ends in this year and thus the election dates for 2009, I stumble over several cases where the council size has changed between the 2005 and 2009 election. In those cases where I know that a new village was created during the term that change is easily explained - one example can be the TAO Chae Chang (องค์การบริหารส่วนตำบลแช่ช้าง) in San Kamphaeng district of Chiang Main province. In 2005 the council had 14 members, in 2009 it had 16 members. Today, the subdistrict contains 10 administrative villages - thus at first one would expect 20 TAO members. But village 7 is completely covered by the municipality San Kamphaeng and thus is not eligible to send councilors to the TAO council, and for village 8 the area which belongs to the TAO is unpopulated, only for Mu 6 one part of the population belongs to the municipality and another to the TAO - so effectively there are eight Muban for the TAO. And since village 10 was created in 2006 there must have been a by-election to enlarge the council shortly after the new Muban became effective.
This way I could already explain a big number of the council size changes, but sadly no all. Even though those files from where I copy my data are published by the Election Commission, they obviously contain wrong data - for example for the TAO Don Du (องค์การบริหารส่วนตำบลดอนดู่) in Khon Kaen province, according to the EC tables the council had a size of 18 in 2005 and 24 in 2009. But - the subdistrict has 12 villages and no area shared with any municipality, and already in 2002 it had 11 villages. Only one village was added in 2006, so the number for 2005 must have been 22, not 18. There were even a few cases where instead of the number of seats the table showed the number of constituencies (i.e. of Muban). In other cases, there were smaller councils in 2005, and in 2009 the council had two members for every Muban - but part of the subdistrict belongs to a municipality, and there were no boundary changes between TAO and municipality as far as I knew. So I have no idea why those Muban were eligible to have councilors in 2009, but not in 2005.
Apparently, the Excel sheets and PDFs from the ECT were created manually, otherwise such obvious mistakes could not happen. If I could spend more time on the programming part of my data collection project, I would easily have a small tool which could output the same tables from my collected data. But then there's still the problem that I don't have the full source data, I can only reconstruct them from those files I have. Especially whenever a mayor or a council ends its term prematurely I would have to update the data in the XML directly - but I rarely find these information at the ECT provincial websites.