x

Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Using Experiments to Estimate the Effects of Education on Voter Turnout
Last registered on April 16, 2017

Pre-Trial

Trial Information
General Information
Title
Using Experiments to Estimate the Effects of Education on Voter Turnout
RCT ID
AEARCTR-0001097
Initial registration date
April 13, 2017
Last updated
April 16, 2017 12:52 AM EDT
Location(s)
Primary Investigator
Affiliation
Columbia University
Other Primary Investigator(s)
PI Affiliation
U.S. Military Academy, West Point
Additional Trial Information
Status
Completed
Start date
1962-01-01
End date
2002-12-31
Secondary IDs
Abstract
The powerful relationship between education and voter turnout is arguably the most well-documented and robust finding in American survey research. Yet the causal interpretation of this relationship remains controversial, with many authors suggesting that the apparent link between education and turnout is spurious. In contrast to previous work, which has relied on observational data to assess the effect of education on voter turnout, this article analyzes two randomized experiments and one quasi-experiment in which educational attainment was altered exogenously. We track the children in these experiments over the long term, examining their voting rates as adults. In all three studies, we find that exogenously induced changes in high school graduation rates have powerful effects on voter turnout rates. These results imply that the correlation between education and turnout is indeed causal. We discuss some of the pathways by which education may transmit its influence.
External Link(s)
Registration Citation
Citation
Green, Donald and Rachel Sondheimer. 2017. "Using Experiments to Estimate the Effects of Education on Voter Turnout." AEA RCT Registry. April 16. https://doi.org/10.1257/rct.1097-1.0.
Former Citation
Green, Donald, Donald Green and Rachel Sondheimer. 2017. "Using Experiments to Estimate the Effects of Education on Voter Turnout." AEA RCT Registry. April 16. http://www.socialscienceregistry.org/trials/1097/history/16603.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Interventions
Intervention(s)
The study uses data from three previous experiments that introduced random or near-random interventions that generated significant increases in schooling. Two are randomized studies, the Perry Preschool Experiment and the Tennessee STAR experiment. The third is a quasi-experimental study of the “I Have a Dream” (IHAD) scholarship program. In all three cases, voter turnout information of study participants, both treatment and control, was collected from public records.

Perry Preschool experiment: Pre-school children in the treatment group were given 2.5 hours of classroom instruction on weekday mornings and a weekly 90-minute teacher visit to the mother and student at the family’s place of residence during a weekday afternoon. Following the preschool intervention, subjects entered kindergarten as scheduled by local school regulations based on date of birth. No further treatment occurred after the conclusion of the initial intervention.

Tennessee STAR experiment: Students and their teachers were randomly assigned to one of three experimental conditions: small classes of 13–17 students, regular classes with 22–25 students, and regular classes with a full-time teacher’s aide.

I Had a Dream (IHAD) scholarship program: Enrollment in IHAD scholarship was offered to and accepted by all 79 fifth-grade students in Lafayette, Colorado, who qualified for the free or reduced lunch program at three elementary schools. In addition to the promise of college scholarships, students receive tutoring in various academic subjects, work with mentors on establishing and achieving educational and career-oriented goals, and participate in extracurricular activities together from the time of their initial selection in elementary school through high school and sometimes beyond.
Intervention Start Date
1962-01-01
Intervention End Date
2002-12-31
Primary Outcomes
Primary Outcomes (end points)
Voter turnout in any one of the following federal elections: the 2000 primary, the 2000 general, the 2002 primary, or the 2002 general election.
Primary Outcomes (explanation)
Voter turnout for Perry study participants: Gathered from public records, using data supplied by the firm Voter Contact Services, based on each participant’s name, address, and birthday

Voter turnout for STAR study participants: Polimetrix, an independent polling firm, matched the names and birth dates of the Project STAR subjects to their national list of registered voters. The information, however, was sometimes insufficient to generate a unique match. To differentiate among multiple matches, the search process also considered the locations to which subjects were likely to move. For example, individuals born in Tennessee have a higher probability of moving to Kentucky or Missouri than to Connecticut or North Dakota. Using this information, and blind to the treatment or control status of the subjects, Polimetrix searched for each subject in their national voter database and compiled a list of likely matches. A hit was determined by name, date of birth, and state of residence, in that order, and assigned a score indicating the likelihood of a correct match.

Voter turnout for IHAD participants: Since this was not a study but a scholarship program, a control group was created out of students who attended fifth grade during the year preceding and following the IHAD cohort. Only students who were eligible for the free or reduced lunch program were eligible for IHAD, thus the control group also was also made from this group. Boulder Valley School District provided the IHAD Foundation with a list of students that met these criteria. These students were contacted through a phone survey and their participation in the free or reduced lunch program and the IHAD program were checked. Voter turnout information for all those who were part of the free or reduced lunch program was checked from registration rolls.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Perry Preschool experiment: The study drew its participants from preschool-age children who were expected to attend Perry Elementary School in Ypsilanti. Researchers first identified families of low socioeconomic status using a score based on three factors: parents’ educational levels, parents’ occupational levels, and the number of rooms in the family household. Within this group of low SES families, researchers then sought to identify young children with relatively low levels of intelligence. Before randomly assigning the experimental groups, researchers matched students into pairs based on IQ scores from the Stanford-Binet test. A member of each pair was randomly assigned into one of two unmarked groups. Next, some of these ranked pairs were swapped between groups so that both groups would have similar mean socioeconomic status, mean intellectual performance, and gender ratios. A coin flip determined which of these two groups was assigned to the program (treatment) condition and which to the no-program (control) condition. Finally, researchers moved siblings of those students chosen for the treatment condition into this condition as well so as to eliminate spillover effects from the intervention. At the outset, 64 students were each assigned to the treatment and control groups.

Tennessee STAR experiment: Students and their teachers were randomly assigned to one of three experimental conditions: small classes of 13–17 students, regular classes with 22–25 students, and regular classes with a full-time teacher’s aide. Once assigned to a particular type of classroom condition, students remained in those conditions through the remainder of the experiment. The STAR intervention lasted until fourth grade, whereupon students entered regularly sized classes.

IHAD scholarship program: not a randomized experiment and no control group. Students who qualified for free or reduced lunch program in 1992 in three elementary schools were offered the IHAD scholarship program. All those offered accepted.

Voter turnout for the treatment and control participants of the three studies is collected from public records.
Experimental Design Details
Randomization Method
Perry Preschool experiment: coin flip
Tennessee STAR experiment: randomized done by STAR program
IHAD scholarship program: not randomized
Randomization Unit
Perry Preschool experiment: students
Tennessee STAR experiment: students and teachers
IHAD scholarship program: not randomized
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments:
Perry Preschool experiment: 123 students
Tennessee STAR experiment: 1,455 students
IHAD scholarship program: 58 students
Sample size: planned number of observations
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments: Perry Preschool experiment: 123 students Tennessee STAR experiment: 1,455 students IHAD scholarship program: 58 students
Sample size (or number of clusters) by treatment arms
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments:
Perry Preschool experiment: 60 treatment, 63 control
Tennessee STAR experiment: 429 treatment, 1026 control
IHAD scholarship program: 19 treatment, 39 control
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
IRB Approval Date
IRB Approval Number
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
Yes
Intervention Completion Date
December 31, 2002, 12:00 AM +00:00
Is data collection complete?
Yes
Data Collection Completion Date
December 31, 2002, 12:00 AM +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments: Perry Preschool experiment: 123 students Tennessee STAR experiment: 1,455 students IHAD scholarship program: 58 students
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments: Perry Preschool experiment: 123 students Tennessee STAR experiment: 1,455 students IHAD scholarship program: 58 students
Final Sample Size (or Number of Clusters) by Treatment Arms
Sample size numbers refer to what was used for the consolidated study by Sondheimer and Green, not the original experiments: Perry Preschool experiment: 60 treatment, 63 control Tennessee STAR experiment: 429 treatment, 1026 control IHAD scholarship program: 19 treatment, 39 control
Data Publication
Data Publication
Is public data available?
No

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Program Files
Program Files
No
Reports, Papers & Other Materials
Relevant Paper(s)
Abstract
USING EXPERIMENTS TO ESTIMATE THE EFFECTS OF EDUCATION ON VOTER TURNOUT

The powerful relationship between education and voter turnout is arguably the most well-documented and robust finding in American survey research. Yet the causal interpretation of this relationship remains controversial, with many authors suggesting that the apparent link between education and turnout is spurious. In contrast to previous work, which has relied on observational data to assess the effect of education on voter turnout, this article analyzes two randomized experiments and one quasi-experiment in which educational attainment was altered exogenously. We track the children in these experiments over the long term, examining their voting rates as adults. In all three studies, we find that exogenously induced changes in high school graduation rates have powerful effects on voter turnout rates. These results imply that the correlation between education and turnout is indeed causal. We discuss some of the pathways by which education may transmit its influence.
Citation
Sondheimer, Rachel Milstein, and Donald Green. 2010. "Using Experiments to Estimate the Effects of Education on Voter Turnout." American Journal of Political Science 54(1): 174–189.
REPORTS & OTHER MATERIALS