How do we make sense of the 43 million or 1.3 billion dollar budget cuts that are being made in Congress? Large quantities such as these are common in our political discourse, yet recent studies demonstrate substantial and systematic biases in evaluating them (Landy, Silbert & Goldin 2012). We explore how the integration of numerical and political information affects voters’ evaluation of political scenarios, and more specifically the effect of number training on this evaluation. The current study investigates the effects of a number training intervention on the numerical estimation task and evaluation of deficit-reducing proposals using a within-subjects design and a typical voting population. Participants in the training condition completed a number estimation task (Siegler & Opfer 2003) and were shown the accurate location of 1 million on a number line from 10 thousand to 1 billion. Line estimation and situation evaluation were assessed before and after the intervention.