This paper introduces a dataset of over 100,000 pages of discussion about how U.S. federal budget line items ought to be used and explores how well a variety of methods commonly used to identify similarity, difference, and influence among texts travel to this new context. Ultimately, it aims to test theories about the role of the executive branch in influencing the policy agenda for congressional appropriations and the role of Congress in influencing the policy agenda of executive agencies. This involves two methodological exercises. First I contrast several approaches to measure similarity and differences in policy texts were we have strong intuitions about their relationships. For example, we expect presidential transitions to affect the content of the president’s annual budget. Different text analysis methods give us different perspectives on the magnitude and substance of change. Second, it outlines an approach for estimating the relative influence of texts on a change in a policy document, for example, the influence of the president’s budget on congressional appropriations committee reports and vice versa.
Measuring Change and Influence in Budget Texts
