I devise a novel approach to evaluate the effectiveness of fiscal policy in the short run with multi-category treatment effects and inverse probability weighting based on the potential outcome framework. This study's main contribution to the literature is the proposed modified conditional independence assumption to improve the evaluation of fiscal policy. Using this approach, I analyze the effects of government spending on the US economy from 1992 to 2019. The empirical study indicates that large fiscal contraction generates a negative effect on the economic growth rate, and small and large fiscal expansions realize a positive effect. However, these effects are not significant in the traditional multiple regression approach. I conclude that this new approach significantly improves the evaluation of fiscal policy.