Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I even say who wrote them. If you have more recent summaries to add to this collection, send them my way I guess. Sorry for the ads; they cover the costs of keeping this online.
Hyde. 2005. Can international election observers deter fraud: Evidence from a natural experiment. Dissertation chapter.
Using a natural experiment, Hyde shows that election observers can prevent election-day fraud.
People argue as to whether election observers matter. After all, most countries that invite observers do so only as a signal to demonstrate that their elections are clean. And even if observers do matter, is it their presence or only their deterrent effect that matters? And perhaps they don't matter--several governments have been known to invite observers but openly commit fraud anyway.
"Since the methodology used in this natural experiment is modeled after medical and epidemiological research, an analogy to this subject may be insightful to explain the reason for the persistent effect of monitoring on election day fraud. Suppose a cold virus circulates among a population and everyone catches the cold. Half of these individuals receive a treatment that also serves as a partially effective vaccination. Later in the year, the same cold circulates among the same population, but it spread much less rapidly, even among those individuals that didn't get the treatment. Those individuals that received the treatment were even less likely to get the cold." (p 36)
The OSCE coordinators assigned each observation team a list of precincts to visit on election day. The people making these lists had only geographic data available. They did not assign observers to the most convenient or interesting places. Instead, their goal was to observe as many precincts as possible, both rural and urban. The assignment, then, was effectively random.
Since no candidate won 50% in the first stage (there were 9 candidates), there was a runoff. Observers were randomly assigned during both stages.
Observation has both transitory and persistent effects.
Urban/rural cleavages and adjacency to Nagorno-Karabakh (incumbent's base of support) do not explain the differences.
It's not clear why observers decrease the incumbent's vote share, only that they do. Perhaps fraud simply ceases while the observers are there. But there are also persistent effects, not just transitory effects, suggesting that workers are also deterred by a fear of being caught a second time. We don't know exactly how much fraud observation eliminates, but we can see that it has significant effects. Indeed, given that the incumbent won barely under 50% of the vote in the first round, it is entirely possible that observation efforts were entirely responsible for causing a runoff to be held.
Research on similar subjects