Institution for Social and Policy Studies

Advancing Research • Shaping Policy • Developing Leaders

Detecting Spillover Effects: Design and Analysis of Multilevel Experiments

ISPS Data Archive: Terms of Use

By using, contributing, and/or downloading files associated with scholarly studies available on the ISPS Data Archive, you agree to these terms and conditions.

Please read the ISPS Data Archive Terms of Use.

Author(s): 

Betsy Sinclair, Margaret McConnell, Donald P. Green

ISPS ID: 
D030
Keyword(s): 
Research design: 
Data type: 
Administrative
Data source(s): 

Authors.

Data source information: 

Authors.
Suggested citation: Sinclair, Betsy, Margaret McConnell, & Donald P. Green (2012), Replication Materials for 'Detecting Spillover Effects: Design and Analysis of Multilevel Experiments,' hdl.handle.net/10079/4b8gtwk; hdl.handle.net/10079/83bk3x3; hdl.handle.net/10079/hmgqnz2; hdl.handle.net/10079/r2280v5; hdl.handle.net/10079/cvdncxk; hdl.handle.net/10079/0k6djw0; hdl.handle.net/10079/mcvdnr9; hdl.handle.net/10079/vt4b8vn. ISPS Data Archive.

Field date: 
April 1, 2009
Location: 
Location details: 
IL - Chicago
Unit of observation: 
Individuals
Sample size: 
64,445
Inclusion/exclusion: 
Individuals eligible to be part of the study were active voters with a permanent address who had registered before 2006. We consider eligible households to be households with between one and three registered voters. We then restricted the study to nine-digit zip codes containing at least two households with two eligible individuals and somewhere between 3 and 15 total households. We mailed postcards only to those residents who had been eligible to participate in both the 2006 and 2008 spring elections.
Randomization procedure: 
We first conducted a randomization based upon each zip code. From the eligible zip codes, we selected a two-voter household at random, which we will call the core household. We then chose 25% of core households to be assigned to control, and 75% were assigned to treatment. For zip codes where core households were assigned to control, all other households in their zip code were also assigned to control. For zip codes where core households were assigned to treatment, zip codes were assigned with equal (1/3) probability into three conditions: First, treatment core households were assigned to be the only household in their zip code receiving treatment. Second, half of the other households in the zip code of the core treatment household were assigned to treatment. Third, all households in the zip code of the treatment core household were also assigned to treatment. Within households assigned to treatment, we randomly chose exactly one individual to receive treatment.
Treatment: 
A mailed postcard reminding voters to do their civic duty and vote in spring elections. This postcard included a reminder about whether or not each individual had participated in the previous spring 2006 and spring 2008 elections.
Treatment administration: 
Mail
Outcome measures: 
Official turnout records for each subject
Archive date: 
2014
Archive contributor: 
Limor Peer
Owner: 
Authors
Owner contact: 

isps(at)yale(dot)edu

Terms of use: 

Academic, non-commercial; see ISPS Terms of Use http://isps.yale.edu/research/data/terms-of-use

Discipline: 
Data file numbersort descending Description File format Size File url
D030F01 Data file Stata (12.0) .dta 8399872 Download file
D030F02 Data file MS Excel .csv 5873664 Download file
D030F03 Program file - annotated replication Stata (12.0) .do 9216 Download file
D030F04 Program file - var labels Stata (12.0) .do 2048 Download file
D030F05 Program file - changes to dataset Stata (12.0) .do 13312 Download file
D030F06 Codebook XML (1.1) .xml 39936 Download file
D030F07 Supplementary materials Adobe acrobat (XI Pro) .pdf 97280 Download file
D030F08 Metadata record Adobe acrobat (XI Pro) .pdf 199680 Download file