NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Back to History
When do media stations support political accountability? A field experiment in Mexico
Initial registration date
April 19, 2016
April 19, 2016 1:32 PM EDT
Other Primary Investigator(s)
Additional Trial Information
The importance of an informed electorate for electoral accountability is widely recognized. However, while a large literature has focused on voter access to news media, little is known empirically about when media outlets choose to provide voters with indicators of their incumbent party's performance in office. This project seeks to explain the relatively low supply of incumbent performance information in Mexico using a three-year and six-period clustered design exploiting differential treatment intensity within media market clusters. We will scrape newspapers and radio stations to generate a massive corpus of text and audio news reports, before applying recent advances in text analysis and machine transcription to measure what the media does and does not report. We will first identify the extent to which search costs affect whether radio stations and newspapers report the results of independent audit reports detailing mayoral malfeasance in office. We will then identify how the effects of providing media outlets with information varies with media market competition and proximity to elections. Finally, we will identify the extent to which outlets learn to acquire information for themselves after receiving a prior treatment or local spillover. We intend for our findings to shed light on an essential but understudied condition required for voters to be able to hold governments accountable for their performance in office. This pre-analysis plan registers our experimental design and how our hypotheses will be tested.
Larreguy, Horacio, Christopher Lucas and John Marshall. 2016. "When do media stations support political accountability? A field experiment in Mexico." AEA RCT Registry. April 19.
We seek to evaluate when media outlets broadcast or publish performance information about municipal and federal incumbents. To identify when media outlets support electoral accountability by reporting incumbent performance information, we conduct a three-year and six-period field experiment in Mexico. We randomize whether local radio stations and newspapers are sent easy-to-digest scorecards detailing two sources of incumbent performance each year: (1) independent audit reports documenting municipal mayoral malfeasance; and (2) scorecards detailing the Congressional activity of federal deputies.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Our main outcomes are:
(1) Media outlet reporting of municipal and federal incumbent performance. (2) Media report sentiment toward municipal and federal incumbents. (3) Municipal and federal incumbent party vote share. (4) Voter turnout at municipal and federal elections. (5) Municipal and federal incumbent performance in office.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
We follow the randomized saturation design proposed by Baird et al. (2015) to capture the average treatment effect and the spillover effect on the non-treated of providing media outlet with easily accessible incumbent performance information. Within blocks of 3 similar media markets, we first randomize 1 media market into a pure control condition and 2 media markets into a partial saturation treatment condition within which half of media outlets receive our information treatment. The experiment will be conducted across 6 periods over 3 years. Treatment assignment will be fixed within years (across both intervention periods in each year), but will vary across years.
Experimental Design Details
Randomization done in office by a computer.
We employ two levels of randomization: the media market and the media outlet. For each year of the intervention (2 intervention periods per year), 2/3 of media markets will be block randomized into treatment. Within treated media markets, 1/2 of media outlets will be treated and 1/2 will be spillover outlets.
Was the treatment clustered?
Sample size: planned number of clusters
292 media market clusters.
Sample size: planned number of observations
9,624 media outlet reporting period observations (6 observations per media outlet).
Sample size (or number of clusters) by treatment arms
In expectation, 535 outlets will be assigned to treatment, 535 to spillover, and 535 to pure control each year.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
Committee on the Use of Human Subjects
IRB Approval Date
IRB Approval Number