Social Learning and Momentum Effects: case study 2008 US Democratic Primary

This paper examines whether there are momentum effects and social learning in the 2008 US Democratic primary. In the econometric model, voters are assumed to engage in sincere voting, but they are uncertain about candidate quality. Thus, voters in late states attempt to benefit from sequential voting by updating their voting intentions with information on voting returns from early states. The empirical application uses voting intentions of late voters and through a two-step estimation process shows whether these voting intentions are updated with the release of voting returns from early states. Our findings indicate that there is no social learning in the 2008 Democratic primary. Although Obama and Edwards outperformed in Iowa, voters did not update appropriately and Obama did not enjoy any momentum effect from this surprising victory relative to expectations, with Clinton winning in New Hampshire. Finally, in an additional empirical specification the paper examines the sources of voting behavior and the results show that demographic characteristics such as gender, age, income, education and region of residence have a significant effect on the probability to choose a candidate.