Text Messaging Field Experiment and Survey 2006

1
Text Messaging Field Experiment and Survey 2006
Principal Investigators Allison Dale, U-Mass
Amherst Political Science Aaron Strauss,
Princeton Political Science
2
Background, Motivation, and Goals of the Study
3
New Generation of Voters Require New Outreach
Methods

• Conventional campaign programs rely on mail and
  phone
• Mail campaigns for young people are problematic
• Young people are more mobile
• Urban and college addresses are less reliable
• Landline phone campaigns are problematic
• The Bureau of Labor Statics reports that 20 of
  young people (15-24) only use a mobile phone.

4
If Democrats and Progressives Need to Act Now

• Turnout of young people is unrealiable
• Young voters are underrepresented in the
  electorate
• Educated young voters are especially likely to
  drop off for midterm elections (Achen, 2006).
• but their votes are reliably Democratic
• In 2004, 18-29-year olds were the only age group
  to support Kerry over Bush on average.
• In 2006, 60 of young voters cast a ballot for a
  Democratic candidate for Congress.

5
This Study Helps Determine Whether Text Messaging
Helps Solve the Youth Turnout Problem
Two Goals

• Quantify the benefits of text messaging via a
  randomized experiment with a control group
• Also test appeal type and level of information
  provided
• Detect any sort of backlash by surveying a
  subset of those in the treatment group
• Unique methodology to ensure a representative
  sample

6
Experimental Design
7
Randomized Experiments and a Post-Treatment
Survey Achieve the Goals
Two Components

• Randomized experiments of newly-registered
  voters
• Main experiment (universe of 12,000) tests
  whether receiving a text message reminder induces
  young people to vote
• A pilot study (universe of 1,400) tests whether
  receiving exact polling location and address
  helps people vote.
• A random subset (n300) of individuals who
  received a text message in the experiment were
  surveyed
• Respondents asked whether they found the texts
  helpful
• and what form of communication they prefer

8
Cell Phone Numbers Were Collected Voluntarily and
Verified Independently

• Three sources of cell phone numbers
• Progressive student organization (e.g., campus
  tabling). Collected 8,611 valid mobile numbers.
• Progressive online organization (online
  registration). Collected 5,343 valid mobile
  numbers.
• Mobile-centric organization (text and online
  registration). Collected 353 valid mobile
  numbers.
• With the help of Survey Sampling, Inc. we used
  prefix and 1000-block classifications to
  eliminate landlines from the universe.

9
In the Upcoming Months We Will Check the Voter
Rolls to Determine Whether Individuals Voted

• 20 states comprise 90 of our universe
• Sources of aggregated voter files at our disposal
• Polimetrix, with the help of Young Voter
  Strategies
• Catalist
• Other Potential Sources
• The DNC
• Secretaries of States

10
The Completed Survey Has a Unique Methodology
Problem
The Disposition of Everyone who Picked up the
Phone, Even for a Second, Was Recorded
Those whodid not takethe survey
Those whodid takethe survey
Very Pleasant
Smwt Pleasant
Indifferent/Annoyed
Very Pleasant
Smwt Pleasant
Indifferent/Annoyed
11
Preliminary Survey Results Initial Weighting by
Gender, Weekend (Recall Rate), and Disposition
12
Participants are Young, Democratic, and
Reasonably Diverse
13
While Positive Reactions Far Outnumbered Negative
Feelings, A Distinct Minority Were Bothered
Open-ended What was your reaction to receiving
the text message? (Classified Response)
Did you find the text message helpful or were you
bothered by it?
25Strong
13Strong
Positive
Neutral
Surprised
Negative
14
However, This Small Backlash Did Not Deter People
From Voting
Did receiving the text message make you (much)
more/less likely to vote?
7Much More
15
Young People Prefered to Be Contacted By Text
Message and Email
Imagine that a political organization wanted to
remind you to vote just before Election Day. How
would you prefer that they contact you?
16
Where We Go From Here
17
There Are Several Steps Remaining

• Complete Weighting and Survey Analysis
• Fully-Fund Survey
• Funders/partners get full access to the data
• Collect Data from Voter Rolls
• Design Experiments for 2008!