Political information is a central variable for the study of mass behavior; numerous theories argue that voters with more information behave fundamentally differently from those with less. Nearly all of the empirical support for these theories, however, comes from cross-sectional data. As a result, these findings are typically biased, and systematically overstate the effect of information on behavior. I demonstrate how to minimize these biases and more accurately estimate the effects of information using several different analytical techniques. These adjustments cause the estimated effect of information to shrink dramatically, often falling to one-half to one-quarter of its former size. I conclude by discussing the implications of my results for the study of political information and political behavior more generally.