A man casts his vote in the 2009 Indonesia presidential election. (Flickr/DFAT.)

Indonesians are bracing for a close result as they prepare to go to the polls on Wednesday.

Many of the country's most established polling firms have withheld polls in recent weeks, ostensibly for 'technical issues'. Press reports indicate that firms with ties to the Jokowi campaign withheld these results because they would have showed Jakarta Governor Joko Widodo's campaign losing its longstanding double-digit lead. In a country where politicians and businessmen often seek to back a winner for patronage purposes, such news could have put his rival, retired General Prabowo Subianto, over the top. Twenty-four hours before polls open, those firms remain silent. 

Yet in one of the final surveys we are likely to see from an established firm before election day, the Indonesian Survey Circle (LSI-Lingkaran) yesterday released results (Indonesian) of a survey showing a slight rebound in support for Jokowi, as the governor is known in Indonesia.

LSI estimates Jokowi's support at 47.8% while Prabowo clocks in at 44.2% That 3.6% gap is nearly outside the poll's margin of error of -/- 2%. As importantly, it showed Jokowi gaining in critical demographics where he has previously been outperformed by Prabowo, including Muslims. The improvement there suggests Jokowi's campaign has made progress in beating back smear campaigns that falsely suggested he was a Christian of Chinese descent (he is a Javanese Muslim). The survey's methodology was robust, including face-to-face interviews with a total of 2400 Indonesians in every province over four days from 2 July to 5 July.

LSI-Lingkaran is only one survey firm, and could always be an outlier. We generally look for trends across established survey firms in Indonesia rather than a single data point, because the survey environment is newer and more opaque than in more mature democracies. But the results are in line with results from other, less established and less robust survey firms that show Jokowi gaining on Prabowo in the campaign's final days, and with anecdotal evidence that the governor's disorganised band of volunteers has finally gotten its act together in the countryside. If LSI-Lingkaran is right, Jokowi has arrested Prabowo's momentum, and not a moment too soon.

Moreover, this result would not include reaction to the final debate between the two tickets on Saturday night, where Jokowi showed a marked improvement compared to previous debates, or to his decision to make a minor pilgrimage to Mecca during the 72-hour quiet period between the end of the campaign and voting. Images of Jokowi in Mecca dominated Indonesian media on Monday, and could further extinguish smear campaigns about his faith. 

Still, it should be stressed, none of this points to a wide margin of victory for Jokowi, but to a very close race. That uncertainty has put the country on edge. Unsubstantiated rumours about military interference in the polls have made front pages of major newspapers. The military denies any such interference, but has ominously warned that a margin of less than 5% could lead to disturbances, particularly in key cities in Java like Yogyakarta where hardline Islamic militants have grown more audacious in recent months. Prabowo, addressing concerns that he is not prepared to lose, made an odd show at the conclusion of Saturday's final presidential debate of pledging to respect the results, raising the disquieting question of whether he believes he has the option of not respecting the results.

Against this backdrop, a thin margin of victory for either ticket could lead to prolonged uncertainty about the ultimate victor as the elections commission counts votes and the Constitutional Court hears electoral disputes.

In two previous Indonesian presidential elections, 'quick counts' revealed the winner within a day of the vote. Quick counts are conducted by survey firms observing the vote count at a representative sample of individual polling stations after voting closes, and have historically been accurate within a percentage point. In 2004 and 2009, that margin of error was inconsequential, as incumbent president Susilo Bambang Yudhoyono trounced his opposition by 20% and 34%. The defeated Megawati Sukarnoputri challenged the results over a series of voter list discrepancies in 2009, but most others recognised SBY had won and moved on.

In a close race, however (say, within 2%), neither side is likely to accept quick counts as authoritative. Many Indonesians would wait for the final tally from the General Elections Commission (KPU), which is not to be released until 21-22 July. Even more problematic, challenges to the results over election violations can be filed with the Constitutional Court later that week. The Court will hear the disputes in August, and will not rule on them until 22-24 August.

In a worst case scenario, then, Indonesia could remain on tenterhooks for another seven weeks.