Presidential Influence Versus Agency Characteristics in Explaining Policy Implementation

Abstract

We posit three rival models of policy implementation in the realm of civil rights: presidential influence, bureaucratic discretion, and prior activities. The first model, presidential influence, works better than the bureaucratic discretion or prior activities models. In fact, as modified in the analysis, presidential influence explains half of the adjusted variance in civil rights cases filed but just under a quarter of adjusted variance in cases closed. A combined model using the most powerful variables from each of the three models improved predictability only slightly and only for cases filed. The findings were encouraging, particularly when one considers the major problems that occur when analyzing data at the agency level.