Are the Experts Right about a BJT Victory?

Making Seat Projections from Polling Data

By Joel Sawat Selway

(a shorter version of this article was published on LatitudeTen)

Source: Official Campaign Media from Bhumjaithai website.

In a recent ThaiRath article, four prominent political analysts independently converged on a striking conclusion: Bhumjaithai (BJT) is likely to emerge from the next general election as the largest party in parliament, winning somewhere between 140 and 150 seats.

At first glance, this claim appears puzzling. Throughout January, national polling has consistently placed the People’s Party (PP) well ahead of BJT. Depending on the poll, PP’s support has ranged between roughly 30 and 34 percent, while BJT has polled between approximately 16 and 22 percent. If national vote share alone determined outcomes, PP would seem the clear frontrunner.

Yet Thailand’s electoral system does not translate votes into seats mechanically. Only 100 of the 500 seats in parliament are allocated through national party lists. The remaining 400 seats are decided in single-member constituencies under first-past-the-post rules. In these contests, relative vote shares within each constituency — not national popularity — determine winners. As a result, parties with strong local organization can outperform their national polling, while parties with broad but shallow support can struggle to convert votes into seats.

The key question, then, is not whether experts are wrong to take BJT seriously, but how polling data are being converted into seat projections. Those conversion rules are rarely made explicit. This article seeks to make them transparent by walking through two common approaches to seat projection and showing how different assumptions generate very different outcomes.

Polling Strength versus Electoral Geography

Analysts who project a BJT victory typically point to several structural advantages. BJT is widely understood to possess dense local networks, strong incumbency advantages, and deep ties to provincial political elites. By contrast, the People’s Party has deliberately distanced itself from traditional Thai electoral practices, particularly the recruitment of baan yai(powerful local political families). As a result, strong national polling for PP does not automatically imply dominance in constituency races.

This logic is not without precedent. In the 2023 election, Move Forward (the People’s Party’s predecessor) won nearly 38 percent of the party-list vote, but only about 25 percent of the constituency vote. That gap reflected the party’s limited local penetration in many districts.

The situation in 2026, however, is not identical. Current polling suggests that PP’s support is now much more evenly divided between party-list and constituency vote intentions. This implies a higher ceiling for constituency success than in 2023 — but still leaves open the question of how far national popularity can carry the party under first-past-the-post rules.

To evaluate whether expert seat projections are plausible, we therefore need to be explicit about the assumptions that connect polling data to constituency outcomes.

Method 1: National Imputation

The simplest and most transparent approach to seat projection is the national imputation method. This approach treats national polling shifts as if they apply uniformly across the country.

The method proceeds in five steps:

1. Start with constituency-level results from the previous general election.

2. Calculate each party’s national-level change in support using current polling.

3. Apply that national swing uniformly to every constituency.

4. Identify the party with the highest adjusted vote share in each constituency.

5. Allocate party-list seats separately using polling questions that explicitly ask about party-list voting intentions.

For this analysis, I use NIDA’s national poll conducted between January 5 and January 8.

Party	2023 Vote	2023 Constit-uency Vote Share	2023 National Vote Share	NIDA 2026 vote%	Absolute change	NIDA multi-plier	Projected vote%
Pheu Thai (PT)	46,479	48.59%	24.54%	17.16%	-7.38%	0.699	34.00% (41.21%)
Palang Pracharath (PPRP)	29,118	30.44%	11.00%	<1%	-10.01%	0.090	2.74% (20.43%)
People’s Party (PP)	15,653	16.33%	25.40%	33.19%	+7.79%	1.307	21.38% (24.16%)
Bhumjaithai (BJT)	1,611	1.68%	13.49%	23.97%	+10.48%	1.777	3.00% (12.16%)
UTN	1,573	1.64%	9.48%	2.40%	-7.08%	0.253	0.42% (0.00%)
Democrats (DP)	351	0.37%	5.99%	13.28%	+7.29%	2.216	0.81% (7.66%)
Thai Sang Thai	-	-	2.29%	1.75%	-0.54%	0.763	0.00% (0.00%)
Chart Thai Pattana	156	0.00%	1.54%	0.00%	-1.54%	0.000	0.00% (0.00%)
Economic Party	-	-	2.13%	4.41%	+2.28%	2.071	0.00% (2.28%)
Other			4.14%	3.52%	-0.62%	0.850	0.78% (0.30%)
Total	95,831

Table 1. Nakhon Ratchasima Constituency 5, predicted change in votesNote: percentages in Projected Vote% column do not add up to 100% due to the virtual disappearance of PPRP and BJT having such a small baseline from the 2023 election. Normalizing does not change the outcome of the prediction.

Table 1 displays the calculation for Nakhon Ratchasima Constituency 5. Let’s start with the Pheu Thai vote of 46,479 (48.49%) in the 2023 elections. There are two ways to calculate the national level change. We can compute a multiplier: the ratio of this year’s poll prediction and last year’s nation vote share. For Pheu Thai that multiplier is 17.16/24.54=0.699 (PT are only predicted to get 69.9% of the votes they did last time). Or we can take the absolute change in the vote: PT’s share is 7.38% lower than in 2023—this is the one I report in the LatitudeTen article. We then multiply Pheu Thai’s constituency vote share by 0.699 (or subtract 7.38% from the constituency vote share). This results in a reduction to 34.00% using the multiplier, or 41.21% using the absolute change.

Under this national imputation method, the results already diverge sharply from the dominant expert narrative. BJT performs strongly, winning an estimated 130-141 seats — a figure close to the lower bound of expert forecasts. The larger surprise lies elsewhere. The People’s Party rises well above expert estimates, reaching 199-209 seats, while Pheu Thai (PT) collapses to 60-74 seats. The Democrats (DP) perform modestly lower, with 39-63 seats.

	Constituency seats (multiplier)	Constituency seats (absolute change)	Party list seats	Total seats (multiplier)	Total seats (absolute change)
BJT	120	106	24	144	130
DP	50	26	13	63	39
Other	8	10	4	12	14
PP	176	166	33	209	199
PPRP	0	15	0	0	15
PT	43	57	17	60	74
TST	4	5	2	6	7
UTN	0	9	2	2	11
EP	0	0	5	5	5

Table 2. Seat predictions using the national imputation method

These outcomes highlight an important point: the simplest translation of polling data does not naturally produce a BJT-dominated parliament. Instead, it suggests a fragmented legislature in which PP is the largest party but still short of a majority.

Why Uniform Swings Can Mislead

Despite its transparency, the national imputation method rests on a strong assumption: that vote swings are geographically uniform. In reality, voters switching away from declining parties are not evenly distributed across constituencies. National polling tells us how much support a party gains or loses overall, but it does not tell us where those gains originate.

Under national imputation, if BJT is projected to gain, for example, 10.48 percentage points nationally, that increase is simply added to its previous vote share everywhere. In practice, however, BJT’s gains are likely to come disproportionately from specific parties, and those parties varied greatly in strength across constituencies in the last election. Where those declining parties were strong, many votes are available to be reallocated; where they were weak, far fewer votes can move.

This distinction matters enormously for seat outcomes.

Method 2: Party Transfer

The party transfer method addresses this problem by explicitly modeling where votes are coming from. Instead of applying national swings uniformly, this approach reallocates votes from declining parties to growing parties at the constituency level, based on assumptions about voter movement.

In practice, the method combines three inputs:

1. Constituency-level vote shares from the previous election.

2. National polling data indicating which parties are gaining and losing support.

3. Assumptions about the direction of voter transfers from declining parties to rising ones.

Under this approach, a party’s national gain does not translate into the same constituency-level increase everywhere. In constituencies where a declining party previously performed well, reallocating its lost votes can produce large gains for the recipient party. In constituencies where that party was already weak, the same national swing produces much smaller effects.

The specific transfer assumptions used here are discussed in detail in Table 3 below and will be revised in future articles as regional polling data are incorporated.

Source Party	Total Loss	To EP	To DP	To BJT	To PP	To Others
PPRP / UTN	17.09	2.28	6.89	3.96	2.66	1.3
PT	7.38	–	–	3.52	3.86	–
CTP	1.54	–	–	1.54	–	–
TST	0.54	–	–	–	0.54	–
Liberal Party	0.73	–	–	–	0.73
New Democracy	0.4	–	0.4	–	–	–
CPP	0.78	–	–	0.78	–	–
Pheu Chart	0.12	–	–	0.12	–	–
Thai Civilized	2.13	2.13	–	–	–	–

Table 3. Transfer vote matrix

Let’s apply these transfers to the same constituency as above, Nakhon Ratchasima 5. Table 4 shows these results. All of PPRP's votes are shipped out and they are picked by by BJT, DP, and PP. PT also loses a substantial number of votes, but since no other party came close in 2003, PT still ends up about 8,000 of its biggest rival.

Party	2023 Votes	Outgoing votes	Incoming votes	Projected vote tally	Projected vote %	Projected Result
Pheu Thai (PT)	46,479	11,886	0	34,593	36.10%	Winner
Palang Pracharath (PPRP)	29,118	29,118	0	0	0.00%
People’s Party (PP)	15,653	0	10,994	26,647	27.80%
Bhumjaithai (BJT)	1,611	0	12,781	14,392	15.00%
UTN	1,573	1,573	0	0	0.00%
Democrats (DP)	351	0	12,373	12,724	13.30%
Thai Sang Thai (TST)	0	0	0	0	0.00%
Chart Thai Pattana (CTP)	156	156	0	0	0.00%
Economic Party (EP)	0	0	4,095	4,095	4.30%
Other	1,046	0	2,335	3,381	3.50%

Table 4. Nakhon Ratchasima Constituency 5, predicted change in votes (transfer method)

Comparing the Results

Applying the party transfer method produces results that differ meaningfully from both expert forecasts and the national imputation baselines. The People’s Party performs even better, moving much closer to an outright majority. BJT’s seat total remains relatively stable across both all three calculations. Both Pheu Thai and the Democrats see modest declines under the transfer method relative to the national imputation results.

	Party list seats	Total seats (multiplier)	Total seats (absolute change)	Total seats (transfer method)	Expert Predictions
BJT	24	144	130	145	140–150
DP	13	63	39	46	20–40
Other	4	12	14	14	n/a
PP	33	209	199	239	100–120
PPRP	0	0	15	0	-
PT	17	60	74	45	80–120
TST	2	6	7	4	-
UTN	2	2	11	2	-
EP	5	5	5	5	-
KT	1-2	0	0	0	40-70

Table 5. Seat projections based on NIDA polling, January 5–8, 2026.

These results are not unique to NIDA’s data. Using Suan Dusit’s most recent poll (January 13–16), which places PP roughly four points higher and BJT about six points lower, would almost certainly generate projections even more favorable to PP. Suan Dusit also reports Pheu Thai’s constituency vote share at nearly 19.5 percent — about four points higher than NIDA — a difference that could substantially alter the PT–BJT balance in many marginal districts.

Conclusion

The broader lesson is straightforward. Seat projections are not objective facts derived directly from polling numbers. They are the product of assumptions — about geography, voter movement, and local competition — layered on top of polling data. Different assumptions produce dramatically different outcomes.

Expert forecasts that place BJT at the head of the next parliament are not impossible, but they are far from inevitable. Under transparent and replicable methods, current polling data support a wide range of plausible outcomes, many of which place the People’s Party well ahead of BJT in total seats.

In subsequent articles, I will move beyond national averages by incorporating regional polling, constituency typologies, party switching, and incumbency effects to better identify where expert forecasts are most — and least — robust.