Theses Doctoral

Essays on Media and Accountability

Groves, Dylan William

Journalism is widely believed to contribute to responsive governance. But rigorous evidence is scarce. This dissertation explores whether journalism improves government responsiveness, how journalism improves government responsiveness, and the conditions under which journalism is supplied. I focus my research on Tanzania, which has experienced a rapid expansion of local and independent media since 1990, but where systems of political accountability and public service delivery remain limited.

In Chapter 1, I evaluate the influence of investigative journalism on government responsiveness using a national-scale randomized controlled trial. I argue that journalism improves responsiveness by strengthening accountability relationships within Tanzania's ruling party bureaucracy. To test the argument, I collaborated with 15 regional radio stations to identify 206 communities experiencing service delivery problems like flooded roads, broken water points, and missing medical supplies. I then randomly assigned half the communities to the treatment group and half the communities to a pure control condition. In treatment communities, journalists investigated the service delivery problem, broadcast their findings on regional radio, and conducted follow up reports several months later. Seven months after the reports were broadcast, independent auditors evaluated the service delivery problem in all 206 communities. I find that treatment communities received higher audit scores on average (coefficient = 0.25 standard deviations, randomization inference p-value = 0.033), amounting to one road or water point repair in every four treated communities. In line with my argument, the reports generated observable responses by un-elected government ministries but not citizens, local government officials, or members of parliament.

In Chapter 2, I evaluate two mechanisms by which journalism influences government responsiveness: informing government officials about the preferences of their constituents and motivating officials with the threat of public exposure. I first draw on surveys of 4,200 citizens and 340 leaders across 109 Tanzanian villages to document whether leaders understand, share, and respond to the policy preferences of their constituents. I then examine the effect of two overlapping treatments, each designed to capture a mechanism of journalism's influence. In the ``information'' experiment, I randomly assigned leaders to receive information about the priorities of their constituents. In the ``motivation'' experiment, I randomly assigned leaders to be contacted by journalists planning reports on a specific development issue in the leader's village. To evaluate outcomes, I developed a behavioral measure of the willingness of village leaders to lobby district council officials for development projects on behalf of their constituents. I find mixed evidence for the role of information, strong evidence for the role of motivation, and no evidence for complimentary between the two mechanisms.

In Chapter 3, I analyze the supply of local journalism in Tanzania. I combine three original data sets: a comprehensive history of radio station ownership in Tanzania, a national survey of local journalists in Tanzania, and a hand-coded data set of every news story published by local radio stations in Tanzania over a five month period. I show that despite a rapid rise in the number of local and independent media stations in Tanzania and a journalistic culture that is generally supportive of ``watchdog journalism,'' the slant of local news coverage in Tanzania remains overwhelmingly pro-government. The most dramatic bias occurs at stations owned by the government and stations owned by individuals with significant business interests outside the radio station, while the bias at radio stations controlled by individuals affiliated with the ruling party is surprisingly muted. Media market competition is also associated with reduced pro-government bias.

Geographic Areas


More About This Work

Academic Units
Political Science
Thesis Advisors
Green, Donald P.
Marshall, John L.
Ph.D., Columbia University
Published Here
July 26, 2023