In the 2000 presidential runoff election in Romania, voters had a choice between a left-wing candidate (Iliescu) and an extreme right candidate (Tudor). This was unlike the two previous runoffs (1992 and 1996) and subsequent runoffs (2004-2019), when a left candidate faced a center-right candidate. So the 2000 runoff provides the setting for a natural experiment. I show that the impact of regional development on the vote for left in 2000 was different from other runoffs. Unlike in all other runoffs, when development was a strong predictor of voting, its impact in 2000 was insignificant. Why? The 2000 runoff was indeed peculiar, but that is half the answer. It would be tempting to assume that, if development had no impact at the macro level, it did not have an impact at the individual level, either. But that conclusion would be wrong. Exit polls show that development (education) was strongly correlated with voting in the 2000 runoff, but the effect was non-linear; in the aggregate, we saw no effect. Thus, in addition to providing an interesting substantive story, I also offer an example of a potential ecological fallacy.
Keywords: Left-right, extreme right, natural experiment, presidential elections, Romania, ecological fallacy