In this paper, we test three methods of estimating ideological bias in news media stories. This forms the basis for the development of a news reading app. We find that WordScores offers the most reliable estimate and reliability is improved when applied after identifying topic.