Methodology
The analysis in Diane Hessan’s columns are based on research activities conducted with
a 500-person online panel, comprised of American voters. The primary purpose of the
panel is to engage with voters on an ongoing basis about their views of the country, the
presidency, our politics, and the future. Although the size of the respondent base is
sufficient for quantitative analysis, the primary objective of the project is qualitative – to
understand the “why” behind more quantitative methodologies like surveys and polls.
In this particular situation, the design of the panel and its methodology had special
considerations. First, candor was a critical factor: people are so used to being polled
that many admit to not telling the truth to pollsters. And, given the divisiveness in our
country, many voters are accustomed to avoiding controversy and keeping their
opinions to themselves. Although it is not possible to guarantee that people would tell
everything, we wanted to significantly increase the chances that this would happen.
Panelists were recruited via a variety of methods, including social media (recruiting
from Facebook groups), in-person recruiting at political events, and purchases from
firms that sell address-based sampling. In addition, an incremental group of millennials
was added in the summer of 2017 via work with The Blue Lab at the Liberty Square
Group, a political consulting firm. The panel is refreshed on a continuous basis, mostly
to replace non-participants or to add segments of particular interest; for instance, in the
summer of 2017, we added 50 additional millennials in order to do special work with
this segment of the population. Or, we added additional voters from Alabama in the
midst of the Senate campaign with Roy Moore and Doug Jones.
As of January 31, 2019, Panelists include:
Democrats: 39%
Republicans: 41%
Independents: 10%
Trump Voters (2016): 45%
Clinton Voters (2016): 42%
Non-voters (2016): 4%
Voted for other than Trump or Clinton (2016): 9%

 Panelists represent all 50 States, with most states having 10 participants. Exceptions to
this include:
Alabama: 11
Alaska: 7
Hawaii: 6
Kansas: 9
Massachusetts: 13
Michigan: 14
Mississippi: 6
North Carolina: 12
New Hampshire: 12
North Dakota: 7
Pennsylvania: 13
Texas: 14
Wyoming: 6
Men: 231
Women: 279
White: 405
African-American or Caribbean-American: 40
Hispanic: 41
Other: 16
College graduates +:229

 Non-college grad: 271
18-29: 104
30-49: 143
50-64: 130
65+:123

The decisions about numbers in each category above were made based on many
considerations. However, the primary driver was to get a critical mass of voters (focus
group size) in each state, and to generally reflect the other demographics of voters in the
U.S.
To qualify for the panel, each prospective member received a detailed description of the
work, including guidelines ensuring that their privacy and confidentiality would be
protected. In this document, Hessan revealed that she had worked on an unpaid project
for the Clinton campaign and that she is self-funding the research; moreover, she noted
that no organization was paying her for her data – to ensure for respondents that their
opinions were not being “sold to the other side”. If panelists wanted to continue, the
next step was a 45-minute telephone interview with Hessan, with the goal of verifying
detailed demographic information (age, gender, occupation, location, voter registration,
past presidential votes from 2008, 2012 and 2016, and more.) The second goal of the
interview was to increase trust – so that participants would know that an actual person
– who would not judge them, only listen – was on the other end of their comments. The
third goal of the phone interviews was to make a determination about where along the
political spectrum a respondent should be placed: liberal, moderate, or conservative.
Moderates were tagged based on their views of where we are as a country, and their
reactions to certain questions about policy and values including views on immigration,
the media, healthcare, and gun control. All participants needed to have access to the
internet. Approximately 90% of those interviewed qualified for the panel; those
eliminated included non-U.S. citizens, and people who decided not to take the time.
Once on the panel, participants are given weekly projects, via email. These projects vary
intentionally, to make participation fun and interesting, given the longitudinal nature of
the work. Typical projects might be a True/False quiz, a video for people to view and
respond to, a fill-in-the-blanks exercise (like Mad Libs), an open-ended question, or
even a request that they ask questions of friends, or watch a speech and send reactions.
Additional techniques have included image annotation, voter journeys, mobile
ethnography, and heat mapping. In rare circumstances, participants are given a more
quantitative survey, but surveys are typically used to begin a conversation on a

 particular topic; for instance, people might be asked about whether they support
Obamacare, but the meat of the response is in why they responded the way they did.
Other than bias that occasionally arises from sampling, the primary bias in this panel is
that most of the respondents are more engaged politically than the average US citizen.
Thus, for instance, they are much more likely to read a newspaper daily – or to be aware
of world events -- than a typical voter.
The responses are content coded and sorted, using spreadsheets and technology that
was originally created at C Space for the purposes of coding qualitative content. Themes
are ranked on a scale of 1-5, with 1 being “only said by 1-2 people” and 5 being “a major
recurring theme.” Quotes chosen for Boston Globe articles are typically rated a 4 or 5.
From time to time, if Hessan is unclear about a particular comment or response, she will
follow up with a phone call to the respondent.
None of the respondents are paid for their time – and on average, a respondent will take
5-10 minutes each week to do the work of the panel. Response rates are almost always
greater than 50% and often significantly higher. The highest response rates come from
Fill-in-the-blanks/Mad Libs exercises.
The techniques used by Hessan were derived during her tenure as founder and CEO of
Communispace Corporation, now C Space, which was acquired by Omnicom in 2011.
Communispace was the first company to use online communities for market research,
which is now an accepted and proven method of leveraging the internet for insight. In
online communities, respondents are also able to have conversations with each other.
In online panels, called communipanels, the respondents are only in conversation with
the researcher. Hessan chose panels because of the emotional nature of the topic, and
the risk of voter conversations turning unproductive. She covers all of the expenses for
the research personally.
More information about online communities and the methodology can be found at
https://cspace.com/customerinside/