NearestNeighbor: Old School Social Networks
Overview
•My Question
•Literature Review
•Data Collection and Project Overview
•Observational Data
•Preliminary Analysis
My Question
•Who Votes?
•Two Answers:
–Socioeconomic Status
–Mobilization
•Can Nonvoters be Mobilized?
Project Overview
•Sampling Method: Finding Low SES and Low Propensity Voters
1.Find Lowest Participating Precincts (50% or less turnout in 1992, 1996, and 2000 Presidential elections)
2.Find Predominantly Poor and Uneducated Precincts
3.Find Lowest Participating voters within each precinct
4.Randomly assign low propensity voters into control and treatment groups.
•Canvassing Procedure
–Members in the treatment group received a visit from a volunteer canvasser throughout the month of October 2004.
–If contact was made, the Low Propensity Voter was encouraged to vote in the upcoming election
Field Experiment Results
- Control Group Turnout Rate: 43.28%
- Treatment Turnout Rate: 55.69%
Observational Results
•Turnout Amongst Similar
Precincts
Bakersfield 503—32.44%
Bakersfield 506—23.98%
Bakersfield 507 and 508—27.87%
• Turnout in Control Group
Bakersfield 150 and 160—42.11%
Bakersfield 550 and 560—40.12%
New Question
•Was there a spill over effect?
–Did low propensity voters who were contacted go out and mobilize their neighbors?
•This seems a laughable question
–Alexis de Tocqueville’s “Nation of Joiners”
–Robert Putnum’s Bowling Alone
Old School Social Network
•Huckfeldt and Sprague Citizens, Politics, and Social Communication: Information and Influence in an Election Campaign
–“Partisan organizations heighten the credibility of their messages through the choice of credible messengers, and often the most credible messengers are the friends, and neighbors and co-workers whose viewpoints we take seriously.”
How to Test My Question
•Geocode the treatment and control groups’ addresses.
•Compare the turnout rates of individuals who lived near a contacted voter with the turnout rates of individuals who lived near a non-contacted voter
•Hypothesis:
–Living near a voter who was contacted will increase the likelihood that an individual voted in 2004.
–Living near voters who were not contacted will have no effect on the likelihood that an individual voted in 2004.
What I’ve Done So Far
–Geocoded data: Each address now has a longitude and latitude coordinance
–Created Polygons: This is to separate the precincts into little spaces where I can then add up the number of contacts and the number of votes in 2004 within each polygon. Then, I can compare the turnout rates in polygons with many contacts to polygons with little or no contacts.
–Used Moran’s I Statistic for the Contact Group and their Neighbors’ Turnout Rates
–Used Moran’s I Statistic for the No Contact Group and their Neighbors’ Turnout Rates
