Scientific election forecasting has become a thriving enterprise in the leading democracies, and France is no exception. Among the first French efforts was the so-called 'Iowa Model,' a political economy equation predicting the winner on the basis of national economic performance and government popularity. The Iowa Model was applied to the 2007 French presidential contest, and did not fare as well as expected. We explore diagnostics on the Iowa Model, in an attempt to see what went wrong, meanwhile comparing it to rival forecasting efforts. It appears that an important omitted variable may be a direct measure of the quality of the campaign itself.