Interesting article from The Economist reviewing the cases and the causes of disputed elections worldwide, including recent presidential election in Armenia.
Why more and more ballots are inconclusive
[...] Recent months have seen an increasing number of elections in which the vote itself is only a small factor in the eventual outcome. The Kenyan poll in December 2007 is the best case. There was a vote and, in principle, a new government. But the link between the two was tenuous. Tribal violence, internationally sponsored power-sharing talks and constitutional amendments had as much to do with the shape of the deal as the election did.
This was an extreme case, but not a unique one. Two recent polls in the Caucasus failed to bring stability: after Georgia's presidential race in January, the opposition staged a hunger strike to contest the results. A far bloodier outcome followed Armenia's ballot in February: eight people were killed when the police laid into supporters of an aggrieved opposition.
[...] of 21 countries which have elected new governments in the past four months, the result of the vote itself was less than decisive in at least six.
The number seems to be rising. In 2006, four or possibly five elections fell into the “inconclusive” category: parliamentary votes in Thailand and Fiji were both overridden by the army; a general election in the Czech Republic produced a long stalemate; the presidential election in the Congo was disputed, though eventually accepted. The loser also disputed Mexico's presidential result and staged street protests, though the Federal Electoral Tribunal confirmed the outcome. There were 70 national votes for president or parliament that year (excluding referendums).
The year 2000 saw roughly the same number: the American presidential election, plus five other such votes, out of 64 in total (the others were in Thailand again, Peru, Côte d'Ivoire, Ethiopia and Serbia). Going back further, all but three of the 48 national elections held in 1990 had clear, accepted results (exceptions were Myanmar, Grenada and Suriname). In 1980, all the national elections were decisive.
So the number of inconclusive elections seems to be rising. Why? One simple answer is that there are more elections now, and that some go off at half-cock. According to Freedom House, an American think-tank, the number of electoral democracies has risen from fewer than 70 in the 1980s to almost 100 in 1992 and to 121 in 2007. Many recent polls took place in new democracies where those in power are reluctant to step down (because ceding power risks losing everything) and opponents balk at accepting the result because they (rightly) mistrust their rulers.
Another factor: the prevalence of election monitors may have changed the way elections are rigged. Instead of claiming to have won by 99% before lunch, new democracies put on a show of sophistication and claim modest victories by, say, 53% to 47%. (Sceptics note that 53% was the winning share claimed in both Armenia and Georgia; but in the Georgian case observers did agree that the incumbent, Mikheil Saakashvili, had clearly topped the poll.)
When the margin is slim, counting disputes increase and challenges are more likely. Sometimes this produces dramatic outcomes: the uprisings in Georgia, Ukraine and Kyrgyzstan in 2003-05 all sprang from disputed polls. And sometimes it merely prolongs political machinations.
Elections, in fact, are just one part of the network of institutions (like honest courts) that need to be in place for democracy to work properly. Without those institutions, voting sometimes seems, at least in the short term, to make things worse.