Can Civil Society Organisations Sway Elections? – Field Experiment

Last registered on July 19, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

Can Civil Society Organisations Sway Elections? – Field Experiment

RCT ID

AEARCTR-0004622

Initial registration date

August 24, 2019

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

August 29, 2019, 8:46 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated

July 19, 2021, 3:09 AM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Country

Germany

Region

Primary Investigator

Name

Johannes Wimmer

Affiliation

LMU Munich

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Leonhard Vollmer

PI Affiliation

LMU Munich

Contact Investigator

Additional Trial Information

Status

In development

Start date

2019-08-24

End date

2022-03-31

Keywords

Governance

Additional Keywords

Civil Society Society Organizations; Elections; Populism

JEL code(s)

Secondary IDs

Abstract

This document describes the pre-analysis plan for a field experiment evaluating the effective-ness of civil society campaigns in shaping election outcomes. This field experiment will be embedded in the Facebook advertising campaign of a German civil society organization dur- ing the run-up to three state elections in Germany in autumn of 2019. The main objective of this field experiment is to test the hypothesis whether civil-society organization can shape elec- tion outcomes via online campaigns. To pursue this objective, we design a randomized field experiment which experimentally varies which postal districts in these three states are exposed to the organization’s Facebook campaign and which are not. In this document we describe the motivation for conducting this experiment, the data and experimental design as well as our main hypotheses and the corresponding empirical analyses.

External Link(s)

Registration Citation

Citation

Vollmer, Leonhard and Johannes Wimmer. 2021. "Can Civil Society Organisations Sway Elections? – Field Experiment." AEA RCT Registry. July 19. https://doi.org/10.1257/rct.4622-1.4000000000000001

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention Start Date

2019-08-24

Intervention End Date

2019-10-27

Primary Outcomes

Primary Outcomes (end points)

Our main outcomes are election outcomes taken from official sources. We consider the following two primary outcomes:
1. Vote share of "Alternative für Deutschland" in the state elections in Brandenburg, Sachsen, and Thüringen in autumn 2019.
1. Turnout in the state elections in Brandenburg, Sachsen, and Thüringen in autumn 2019.

For more details on the outcomes employed in our study please see our pre-analysis plan.

Primary Outcomes (explanation)

We take our primary outcomes directly from official statistics. The only transformation we conduct is aggregation to the postal district level.

Secondary Outcomes

Secondary Outcomes (end points)

Besides our main outcomes, we employ a set of additional outcomes:
1. social media data (e.g. posts by location and sender)
2. further political outcomes (e.g. demonstrations)
3. internal data from our project partner (e.g. performance metrics of Facebook campaign)

For more details on our set of additional outcomes please refer to our pre-analysis plan.

Secondary Outcomes (explanation)

Experimental Design

We design a randomized field experiment which experimentally varies which postal districts in the three states of Brandenburg, Sachsen, and Thüringen are exposed to the Facebook campaign of our project partner. Our experimental design has the following key features: first, postal districts constitute the level of randomization. Second, we employ a stratified design as to reduce differences in pre-treatment characteristics between postal districts in the treatment and the control group. To be precise, we construct strata of four postal districts which are similar with respect to their past election outcomes as well as population. Within each of these strata we assign 50 percent of postal districts to the treatment group and the remaining 50 percent to the control group.

We specify further details of our design in our pre-analysis plan.

Experimental Design Details

Randomization Method

Randomization done in office by a computer (DeclareDesign package for R)

Randomization Unit

Postal districts

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

812

Sample size: planned number of observations

In case of the cluster-level analysis (postal districts): 812 In case of the unit-level analysis (electoral-precincts): ca. 9500 (we will know the exact number only once the official election results are available because the number of electoral precincts varies over time)

Sample size (or number of clusters) by treatment arms

406

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Our power calculations yield that we are able detect treatment effects starting at approximately 3 percent of a standard deviation

Supporting Documents and Materials

IRB