Responding to the COVID-19 pandemic: How municipal government performance affects electoral accountability in Mexico

How to cite

Marshall, John; Enríquez, José Ramón; Larreguy, Horacio; Simpser, Alberto. 2021. "Responding to the COVID-19 pandemic: How municipal government performance affects electoral accountability in Mexico." AEA RCT Registry. AEARCTR-0007789. https://doi.org/10.1257/rct.7789-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

John Marshall* , José Ramón Enríquez , Horacio Larreguy , Alberto Simpser

* Primary investigator

Registry ID
AEARCTR-0007789
DOI
https://doi.org/10.1257/rct.7789-1.0
First registered
Last updated
Status
On going
Trial dates
2021-05-28 – 2021-09-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/7789

Abstract

Political accountability is regarded as critical for responsive government, economic development, and equitable policy-making. However, effective accountability is often constrained by voters lacking the information required to evaluate the performance of elected politicians. Moreover, conditional on receiving information, political polarization and selective engagement with content may lead voters to reject credible information. Policy responses to the COVID-19 pandemic, which is both salient while also polarizing public opinion on policy responses, provides an opportunity for voters to learn about the priorities and competence of governments. To understand how voters hold elected politicians accountable for their crisis management, we conduct a randomized evaluation of the effect of Facebook ads---sent by Data Cívica, a local civil society organization---documenting the policies and COVID-19 case and death counts of Mexican municipalities, relative to other municipalities in the same state, ahead of the June 2021 elections. This project examines the effect of such information, as well as an additional randomized message encouraging consumers to view the information through a nonpartisan lens, on the beliefs, attitudes, and voting behaviors of Mexican citizens in medium-sized municipalities across the country. To this end, we will examine the effects of localized Facebook ad campaigns on electoral returns and the effects of the ads on citizen beliefs and intended vote in the context of a pre-election online survey.

Intervention

In collaboration with local civil society organization, Data Cívica, we examine the effects of providing Mexican voters with information about their municipal government's response to COVID-19 and relative COVID-19 cases and deaths in their municipality. In particular, Data Cívica's informational video ads benchmark COVID-19 cases and deaths per capita in their municipality against other municipalities within the same state. Variants of the ad further inform viewers of policies implemented by the municipal government in response to the pandemic and---given the increasingly polarized political space in Mexico---seek to increase internalization of the information by encouraging viewers to consider the information in a non-partisan way.

Experimental design

The impact of the ads will be evaluated by two parallel experiments. First, a two-level randomization process is used to assign municipalities to receive different Facebook ad campaigns and then to assign different geographical segments within those municipalities to receive these ads. This will enable the research team to measure the direct and (within-municipality) spillover effects of the ads on precinct-level voting behavior. Second, to better understand how voter beliefs and vote intentions are influenced by the video ads and the mechanisms driving any effects, the research team will conduct a pre-election online survey that includes a battery of questions about voter preferences and randomizes exposure to the ads within the survey itself.

Design details

Randomization method
Facebook ad intervention: computer in office. Survey-level intervention: Qualtrics.
Randomization unit
Facebook ad intervention: --- Higher-level randomization: municipality. --- Lower-level randomization: segments within municipalities. Survey intervention: individual.
Clusters
Facebook ad intervention: 500 municipalities, each divided in 5 segments.
Observations
Facebook ad intervention: 2,500 segments (this randomization unit in turn contains electoral precincts). Survey-level intervention: ~2,400 individual respondents.
Arms
Facebook ad intervention: 300 control municipalities (1,500 control segments); 100 basic treatment municipalities (100 basic treatment, 100 basic + policies segments, and 300 control segments); 100 polarization treatment municipalities (100 polarization treatment, 100 polarization + policies segments, and 300 control segments). Survey-level intervention: ~800 control individuals; ~400 basic treatment individuals; ~400 polarization treatment individuals; ~400 basic + policies treatment individuals; ~400 basic + policies treatment individuals.
Primary outcomes
Precinct-level electoral outcomes: turnout, incumbent vote share. Survey outcomes: beliefs about municipal governments and vote intentions.

Cited by

No citation data available for this registration yet.

Keywords