Embed Size (px)
Citation preview
Getting Out the Vote in the Social Media Era: Are Digital Tools changing
the Extent, Nature and Impact of Party Contacting in Elections?
John Aldrich
Duke University, [email protected]
Rachel K. Gibson
University of Manchester, [email protected]
Marta Cantijoch
University of Manchester, [email protected]
Tobias Konitzer
Stanford University, [email protected]
Paper prepared for presentation at the workshop “Voter Mobilization in Context”, Manchester, 8th November 2013
Abstract: The arrival of the internet and email, and more recently social media or web 2.0 technologies has provided parties and candidates with a range of new and more personalized ways to engage in voter contacting. This paper seeks to measure the extent of these new online forms of contact in the U.S. 2012 Presidential election and UK 2010 General Election and compare them with more traditional offline methods. In a second step we seek to measure the effect of these new forms of contact on individual political behaviour, specifically in terms of whether they increased the likelihood of voting and engagement in other campaign activities. We do so using a combination data sources that include the American National Election Study, an original survey of online political behaviour conducted in the UK 2010 campaign and the 2010 British Election Study. Our findings show that online contact constitutes a small proportion of that received by voters in both countries, particularly in the case of the UK. Offline methods still remain most popular. Furthermore indirect online contact (i.e. mediated through social networks) is more common than that coming directly from the campaign. In terms of mobilization, online contacting (direct or indirect) makes little difference to turnout although it does appear to be linked with undertaking a range of wider campaign activities.
Introduction:
The arrival of the internet and email, and more recently social media or web 2.0 technologies
has provided parties and candidates with a range of new and more personalized ways to
engage in voter contacting. Through the strategic development of email lists, twitter feeds
and Facebook networks parties now have a suite of tools through which they can directly
reach individual citizens in their homes and workplaces and also ‘on the move’ via smart
phones. To date the research into the occurrence and effects of these new types of voter
contact, and particularly that taking place via social media platforms has been limited. What
has been done has presented mixed findings. Studies of email GOTV and registration
campaigns have reported little discernible effects of online contacting (Krueger, 2008;
Nickerson, 2007; Stollwerk, 2006). However, a recent highly publicised study of Facebook
showed that GOTV messages mediated through one’s social network have some small but
significant effects akin to phone contacting or direct mail (Bond et al, 2012). Other studies
have shown positive effects of unsolicited sms or text message reminders on younger voters
(Dale & Strauss (2007).
This paper advances the growing debate over the effectiveness of online voter contact
in three main ways. First we develop a wider classificatory framework in which this new
mode can be located and understood in relation to existing offline modes. This allows us to
better understand and specify its mobilizing properties. We then undertake an empirical
analysis in which we use election study data from the U.S. and UK to measure the extent of
different types of voter contacting over the past decade and locate online contact within those
trends. Finally we examine the impact of these various types of contact on voters to see how
far the online mode actually can be seen to produce results in terms of voter mobilization,
both for turnout and for other more intensive forms of campaign participation. We conclude
by comparing the rates and impact of online contacting across the two countries and
speculating on the macro or contextual factors that may be affecting the use and effectiveness
of online contacting across countries. The study is important in terms of increasing our
understanding of voter mobilization in the digital era. Not all types of contact are created
equal and while face to face methods are generally seen as the ‘gold standard’ for prompting
behavioural shifts, perhaps the vast informal contacting opportunities that Facebook and
Twitter present give campaigns a new and potentially rich channel for more indirect and
personalized two-step communication?
Literature Review
Political science has a fairly long history in the study of the voter contacting. Much of the
work has been conducted in the U.S. and focused on its effects, particularly on voter turnout.
The research has covered a range of elections over different time periods and employed a
variety of methods and data from self-reported contacting via surveys to controlled
interventions using field experiments. Despite this variation a key finding has been that voter
contact, and particularly direct face-to-face canvassing, matters (Blydenburgh 1971, Beck
2002; Bochtel and Denver 1971, Bergen et al. 2005; Cutright 1963, Gerber and Green 2000,
2008; Gosnell 1927, Katz and Eldersveld 1961, Kramer 1971, Merriam and Gosnell 1924;
Nagel 1987; Panagopoulos, and Francia 2009; Rosenstone and Hanson 1993). Convinced by
their evidence of the impact of party mobilization on turnout in U.S. Presidential elections,
two of the leading scholars on the topic even argued that all else being equal, had political
actors been as active in mobilizing voters in the 1980s as they were in the 1960s then turnout
the U.S. would have fallen by less than three percent compared to the 11.3 percent drop that
actually occurred (Rosenstone and Hansen 1993; 218). The experimental work of Gerber and
Green (2000; 2004; 2008) demonstrated this point in even more precise terms with face to
face contacting found to be considerably more effective in prompting turnout than direct mail
and telephone. Using randomized control trials across several elections they estimated the
cost of a canvassing operation to be around $29 for each new voter, compared with $38 for a
phone bank operation and $67 for a direct mail campaign (2008). Subsequent research has
only served to reinforce their argument with the upturn in voter turnout in recent elections
being attributed to the parties improving their ability to target potential voters based on the
increasingly complex array of demographic, socioeconomic and consumer, or “big,” data that
they can now assemble (Panagopoulos and Francia, 2009; Wielhouwer, 1994).
Moving outside the U.S. context, comparative work by Beck (2002) and by
Magalhães (2007, 2010), using survey data from the CNEP study, has echoed American
scholars’ findings that being contacted prior to an election matters in terms of whether one
votes. Indeed Magalhães found that the effect amounted to as much as a five percent increase
across the fifteen nations analysed (Magalhães, 2010). Beyond measuring its effects, other
comparative work examining contact has also used country differences to examine the factors
that drive party mobilization efforts. Work by Karp et al. (2007) using CSES data has shown
sharp variation in contact rates across countries. The presence of a personalized ‘candidate-
centered’ election campaign, rather than the more nameless party campaigns found in PR
systems, was found to be an important driver.
To date, therefore, there appears to be substantial agreement within the literature that
contacting works and that mode matters with face-to-face contact proving to be the most
effective method. The emergence of newer forms of digital communication has presented a
wholly new set of contacting opportunities and raised fresh questions for researchers about
their mobilization potential. Certainly the adoption and increasing importance of websites in
party and campaign communication has been well documented from the late 1990s (Margolis
and Resnick, 1997; Gibson and Ward, 1998; Gibson et al, 2003; Foot and Schneider, 2002;
Kluver et al, 2007; Ward, Nixon and Taras, 2008; Vergeer, 2013). Studies of the effects of
these efforts have been limited but revealed some interesting if inconsistent results. A small
number of studies have looked directly at the impact of candidates’ websites and levels of
electoral support, controlling for a range of other obvious sources of success (D’Alessio,
1997; Gibson and McAllister, 2011; Sudulich and Wall, 2010). The results have consistently
shown a small but significant boost associated with having a web presence. The campaign of
2008 by Barack Obama was also seen by many to demonstrate unequivocally the importance
of the web and particularly social media for success in modern elections (Harfoush, 2009;
Gibson, 2013). The ‘pull’ nature of these technologies, however, means that voters are
required to seek out the information rather than have it ‘pushed’ to them through their
mailboxes and unsolicited phone calls. This requirement clearly reduces their capacity for
mobilization in terms of bringing people to the polls who might otherwise have stayed at
home. Their effects are thus seen to be conversions of the undecided rather than mobilization.
Websites are of course not the only type of online stimuli that voters can receive
during an election. Email is perhaps most comparable to offline voter contact in terms of
direct mailing and campaigns are increasingly developing extensive email lists to drive their
GOTV efforts. Williams and Trammell (2005) for instance note that in the 2008 Presidential
campaign the Obama team sent 3 emails per day and built up a list of 13 million recipients.
This compares to the paltry efforts of Bush and Kerry in the 2004 U.S. presidential election
who sent a total of 50 and 28 e-mails, respectively. Studies of their effectiveness, however,
have been even less conclusive than have studies of websites. The results of a randomized
field experiment conducted by Stollwerk (2005) in conjunction with the Democratic National
Committee during municipal elections in 2005 reported no significant effects of an email
GOTV campaign on voter turnout. Nickerson (2007) reported little discernible effect of email
GOTV and voter registration campaigns from a series of field experiments conducted
between 2002 and 2004 that involved over 200,000 people. Krueger (2008) took a slightly
different approach and used survey data to first identify the number individuals who had
reported receiving any unsolicited versus solicited or ‘opt-in’ political emails. He found that
around 1 in 7 had received some form of email contact but that the unsolicited emails had
little to no effect on people’s level of online or offline political engagement. This was in
direct contrast to the effect of offline unsolicited contact (e.g. in person, by phone or mail)
which he did conclude was positively related to online and offline participation.
On a more positive note, studies of the impact of texting by Dale & Strauss (2007a;
2007b) have pointed to stronger effects. Using a combination of field experiments, a mobile
phone survey and vote validation during the 2006 Congressional elections the authors sought
to determine whether text messaging could be used to encourage young people to vote. The
findings showed a highly positive (4%) but not quite statistically significant effect of sending
a text message as a reminder to vote. In practical terms the authors calculated that despite
their weak effect, the marginal costs of text-based get-out-the-vote reminders made them a
highly useful tool with one voter recruited for every 25 people contacted (for an overall cost
$1.50 per new voter).
Even less well documented than email is the mobilizing impact of the more informal
and indirect forms of online contact made possible through social media platforms. Given
how widely used these social media spaces now are (Obama had 33 million supporters on
Facebook in 2012) this type of contact is expected to be a significant new channel for voter
communication. It is particularly important, however, to make a distinction between the
political messages that are mediated through friends and family on platforms such as
Facebook and Twitter and those that come direct from the campaign. Pioneering work on this
latter topic was reported by Bond et al (2012) who ran a randomized controlled trial of
political mobilization messages delivered to 61 million Facebook users during the 2010 U.S.
congressional elections. Key findings from the study were that those who received a neutral
informational message on their Facebook profile to remind them to vote were not affected by
it. However, those who received the message endorsed by a random selection of their friends
were significantly (2%) more likely to click the 'I voted' button and more importantly were
also 0.3% more likely to seek information about a polling place and 0.4% more likely to have
voted (vote validation was carried out on a sample). Although the effects were small in size,
the authors contend that the scale of the coverage means that potentially millions of people
could be affected. Indeed for this experiment they estimate that the more socially constructed
message increased turnout directly by about 60,000 votes and indirectly through accidental
exposure by a further 280,000.
Given that the conclusions about the marginal effects of email reported above have
been drawn largely from studies of official campaign messages or those sent by third parties
keen to stimulate turnout, drawing any clear lessons about the limited impact of online
contact it seems may be premature. The effects of viral emailing or those messages that are
passed on by a friend or acquaintance in this two-step manner have not as yet been the
subject of a great deal of systematic investigation but may be a source of significant effects.
Certainly if one consults the related body of literature on the impact of “social resources” it is
clear that informal discussion networks can act as important prompts or shortcuts in voting
decisions (Beck et al. 2002; Huckfeldt et. al. 1995; Leighley, 1990; Magalhães 2007;
Partheymüller and Schmitt-Beck, 2012; Popkin, 1991; Schmitt-Beck, 2008; Sniderman et al.,
1991). Contexts of social interaction tend to be characterized by a similarity of interests and
mutual trust, increasing the likelihood that the cues received in this way are seen as credible
and consequential (Huckfeldt and Sprague, 1995). This can be particularly consequential for
individuals with lower levels of political information and awareness (Zaller 1992; Beck et al.
2002). If partisan messages are mediated through online social networks, therefore, then one
might expect to see a similar mobilizing effect.
Research Questions
In summary, the literature on voter contacting has shown fairly conclusively that it is
consequential to turnout, but that the type of contact matters. The ‘personal’ touch appears to
be the critical element to achieving an impact offline, and although less well documented it
appears that this effect also applies to online forms of contact. Comparative studies have
demonstrated that there is variance in the extent of contacting across systems and that this is
affected by institutional macro level factors. As such online contacting poses a number of
interesting questions. Firstly what are the rates of online contact compared with offline
methods and how much online contact is mediated or indirect through social networks versus
being direct from the party? Also how consistent are these patterns across countries? In
addition, to what extent is online contact most likely to prompt political engagement and
again does this vary according to mode used or type of contact i.e. direct or indirect?
In order to address this first set of more descriptive questions it is necessary to
separate or classify modes of contact. The classification is fairly standard for offline modes
which generally are divided into whether they use face to face methods versus other more
impersonal communication channels, i.e. mail, leaflet, phone call. The latter can be treated
individually or aggregated together. For online methods while one can separate specific
channels for communication such as email, websites, Twitter, a more important first
distinction seems to be whether the source of the contact is direct or indirect, i.e. whether it
comes via official channels or is an informal prompt passed on through friends and family.
Combining these various distinctions in mode, medium and source yields a two by two table:
Table 1 about here
Much of the literature to date has focused on quadrants 1 and 2 with investigation of quadrant
2 focusing largely on the role and impact of political discussion and social network influence
noted above. More limited attention has been given to online modes of contact, particularly
that occurring in quadrant 4. An interesting gap to fill, therefore, first of all lies in mapping
the extent of these various types of contact in relation to one another in recent national
elections.
Once we have compared the use of online contact with other types, a second
important question is how well it works in comparison with other more traditional types of
contact. Given how the online mode is disposed to a greater ‘personalising’ of the campaign
message through social media platforms such as Facebook and Twitter, and we know a
personalization of mobilization messages through friends and family can significantly
increase voter turnout,1 then one might expect that online tools could provide an important
new channel for direct mobilization, but might be particularly important for indirect modes of
contact or two-step mobilization, in which direct mobilization attempts are mediated through
friends or family.
As a first step we seek to populate this table with data for U.S. and UK national
elections over time and track changing patterns in the frequency and types of voter contact
being engaged in as we enter the era of digital elections. In looking at these trends we also try
to assess whether online tools actually might be extending political actors’ reach during an
election campaign in terms of size of the voter pool that are contacted. We then turn to
questions of impact of the various modes of contact on individual voting behaviour and wider
1 See Gerber, Alan S.; Donald P. Green. 2000. “The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment.” The American Political Science Review. Vol. 94, No. 3. pp. 653-663; Green, Donald P. and Alan S. Gerber. 2001. “Getting Out the Youth Vote: Results from Randomized Field Experiments.” Unpublished Manuscript, Yale University. December 29, 2001; Eldersveld, Samuel. 1956. “Experimental Propaganda Techniques and Voting Behavior.” The American Political Science Review 50(1): 154-165; Patterson, Samuel C. and Gregory A. Caldeira. 1983. “Getting Out the Vote: Participation in Gubernatorial Elections.” The American Political Science Review 77(3): 675-689.
involvement in campaign activities. While we expect face to face methods to be extremely
important, as noted above our central focus is on the impact of online contacting, particularly
in the mediated or indirect form. Its viral nature means that it radically expands the direct
contact opportunities beyond what is possible through face to face and even by phone or
direct mail. We hypothesise that this will increase the extent of contacting received and
particularly that by friends and family as a whole. We further hypothesize that indirect online
contact will significantly affect individuals’ likelihood of voting and also their engagement in
campaign activity more generally.
We think it is particularly interesting to expand the scope of the analysis to examine
the impact of contact on participation beyond turnout or the vote, per se. This is especially so
in the U.S., where presidential campaigns are not only vastly longer than the national
campaign in the UK, but they are also extraordinarily expensive, in both money and effort.
Especially in the presidential nomination system, both money and labour are raised by the
candidates and their campaign teams (and, more recently by the parties and by unaffiliated
groups of various sorts). The 2008 Obama campaign was considered especially effective in
the spring in terms of fund raising and securing the commitment of volunteer campaign
workers. This network of people and resources were deployed particularly effectively in the
spring in getting people to attend state and local party caucuses in support of Obama, and in
the fall, efforts were deployed particularly effectively in GOTV campaigns, particularly
among minority communities and the young, such as on college campuses. Assuming that
2012 continued in this vein then it should offer one of the most fruitful recent cases for
examining the impact of online-based mobilization efforts.
Data and Methods
The data used to test our research questions are taken from different election surveys. The
UK data come from two sources. The first one is a post-election, face-to-face survey
conducted by BMRB, a UK polling company2. The survey included indicators of three forms
of contact: online direct contact, offline direct contact, and online indirect contact3. The
2 The fieldwork was conducted between the 20th and 26th of May, 2010. Control of quotas affecting likelihood of being at home (age and working status within sex) was applied following a one stage ACORN and region stratification. The data was weighted to ensure that demographic profiles matched those for all residents in Great Britain aged eighteen or over. The overall N was 1,960. 3 The question wordings were: for direct contact – “In the course of the recent election did anyone from a political party, campaign or political organisation contact you to ask about how you were planning to vote through any of the following methods? Online or internet-based contact (i.e. through email or any internet/web-related technology); By telephone, mail or face to face; I have not been contacted; Don’t know or not sure”. For indirect online contact: “In the course of the recent election, did you receive any campaign-related political
second dataset used was the post-election cross-section face-to-face sample of the 2010
British Election Study (BES)4. This survey included detailed indicators of direct contact5 and
a battery on contact by friends, family, fellow workers or other acquaintances6. The US data
are taken from the face-to-face survey of the 2012 time series study of the American National
Election Study (ANES).7 The survey included indicators of all four forms of contact: online
and offline direct contact and offline and online indirect contact8. The basic frequencies for
each contact item obtained in the three samples are presented in Table 2. These are then
compared with prior contact figures for each country in Tables 3 and 4.
Table 2 about here
The results in Table 2 show that offline direct contact was very common during the British
campaign, in which around half of the electorate reported having been contacted by a party,
campaign or political organisation through methods such as telephone, mail or face-to-face
canvassing. This was followed by contact by friends and family, with 16% of the electorate
reporting that someone tried to convince them to vote for one candidate or another (BES
data). According to the BMRB survey, a similar proportion of the British electorate (15%)
received messages or campaign-related content through the internet. In Britain however,
online contact by parties and political organisations was only marginal (around 2% of the
messages or content through the internet from a friend, a member of your family or someone at work? Yes; No; Don’t know or not sure”. 4 The fieldwork was conducted between the 7th of May and the 5th of September 2010 by TNS-BMRB. The sampling followed a multi-stage stratified random design. Weighting was then used to ensure survey respondents are representative of residents in private households in Great Britain aged 18 years or older. The overall N was 3,075. 5 These indicators are derived from the following battery: a) “Did any of the political parties contact you during the recent election campaign?”- b) “Which of the political parties contacted you during the recent election campaign?” - and c) “Please tell me all the ways that the party X contacted you during the recent election campaign: Telephone call; Leaflets or other material delivered to your home; Email or text message; Visit to your home; Contact in the street; Facebook, Twitter, YouTube”. 6 This question did not specify through which method the contact took place (offline or online). The exact wording was: “Did anyone, for example, a friend, a member of your family, someone at work, or some other person try to convince you which party to vote for in the recent general election?” 7 Data collection for the ANES 2012 Time Series Study started in early September and ended in January, 2013. Pre-election interviews were conducted with study respondents during the two months prior to the 2012 elections and were followed by post-election interviews beginning November 7, 2012. The overall N was 2,056. Note that the contacting variables are drawn from the data gathered from running the CSES module IV battery. 8 These indicators are derived from the following questions: during the campaign, did a party or candidate contact you in person or by any other means? (formal contact); during the campaign, did a friend, family member or other acquaintance try to persuade you to vote for a particular party or candidate? (informal contact); did they contact (try to persuade) you in person/face-to-face?; did they contact (try to persuade) you by mail?; did they contact (try to persuade) you by phone? (offline contact); Did they contact (try to persuade) you by text-message or SMS?; did they contact (try to persuade) you by email?; did they contact (try to persuade) you through a social network site or other web-based methods? (online).
electorate). This is particularly low if we compare it to the same type of contact in the U.S.
Over 17% of the American electorate reported having received an email, SMS, a message
through a social network, or contact by another web-based method from candidates or
campaigns. Albeit high when compared to the UK, this figure does not yet match the
numbers of other types of contact during the last presidential campaign in the U.S. Here, 36%
were reached offline by candidates and campaigns. A similar number of people (39%)
reported that a family member, friend, or acquaintance tried to persuade them to support a
specific candidate in the election. And indirect mobilizing efforts through emails, messages in
social networks or other web-based methods reached 23% of the electorate. To help put these
findings into context, Tables 3 and 4 show figures on offline contact from the BES and
ANES for up to three previous national elections.
Tables 3 and 4 about here
Here we can see that rates of offline contact differed by mode and by country with U.S.
campaigns favouring face to face very strongly while UK campaigners tend to rely more
heavily on leaflets. A common story to emerge, however, is the general decline in reported
contact in more labour intensive methods such as face to face and phone by the campaigners
and a simultaneous increase in contact occurring indirectly through friends and family. This
is particularly apparent in the UK. Taking all forms of contact into account 30% of the UK
sample and 61% of the U.S. reported encountering some effort to mobilize them to vote in
2010 and 2012 respectively. This compares with 35% and 71% in 2001 and 2000.9
To examine the mobilizing effect of these different forms of contact, we focused on
two types of electoral participation: voting and campaign participation. Analysing voting
behaviour using survey data is problematic due to over-reporting. Indeed, 75% of the BMRB
sample reported having voted in the general election (77% in the BES sample), when the
actual turnout rate was 65%10. In the United States, the ANES data suggests a turnout rate of
71% while the actual rate was 58%11. Keeping these limitations in mind, our first main
dependent variable in each country was a dichotomous measure of voting in the last election.
Binary logistic regression models were conducted. Our second dependent variable captured 9 Percentages include those contacted face-to-face, by phone or by mail for both informal and formal contact in 2000, and additionally includes those contacted by text-message or online for both informal and formal contact in 2012 for the US data, and include those contacted face-to-face or by phone for both informal and formal contact in 2001 and 2010 for the British data. 10 Source: The Electoral Commission: http://www.electoralcommission.org.uk/ 11 Source: United States Election Project: http://elections.gmu.edu/Turnout_2012G.html
participation during the electoral campaign. In the UK data we used a variable available in
the BES survey measuring the likelihood of working for a political party or candidate during
a campaign in the next few years (0 being very unlikely and 10 very likely). This variable
was highly positively skewed, with 61% of the sample reporting the lowest level of
likelihood of engagement. After running standard diagnostics, OLS estimation methods were
discarded and a negative binomial regression model was conducted instead. In the U.S.
sample, we employed a dichotomous variable measuring whether the respondent had worn a
campaign button or bumper sticker, attended a rally, given money to either party or
candidate, or done other works for candidates. 22% of the sample reported having done at
least one of these campaign activities, similar to levels found in other recent campaigns.
The four types of contact, direct online and offline, and indirect online and offline
(when available) constituted our main independent variables. Each of these was dichotomous,
coded 1 if the respondent had been contacted. In the model using BMRB data, since the
survey did not include an item for offline indirect contact, we included an indicator of
political discussion as a control. In all the models, the following additional controls were
included: gender, age, education, a measure of socioeconomic status (income in the US and
social class in the UK), interest in politics and party identification. In order to control for
differences between safe and contested seats/states which may affect the extent of parties’
campaign intensity and resource distribution, dummy variables were included that were
coded 1 if the respondent lived in a marginal constituency (UK) or battleground state (US). In
the US models and UK models using BMRB data, a variable measuring whether the
respondent had previously signed up online to receive information about the campaign was
included (not available in the BES). We discuss below the implications of adding this
variable to the models. The US analyses also included controls for race (Hispanic and black).
The exact codings of all the variables can be found in the appendix.
Findings
The results of the regressions for the two countries and the two dependent measures are found
in Tables 5 – 8. As noted above, while three models were estimated by simply logistic
regressions, the UK measure of campaign participation is different, so that a negative
binomial estimator was used instead. The results are an interesting mixture of outcomes.
Tables 5 to 8 about here
First, with respect to the control variables, note that they are both statistically
significant and substantively strong (and appropriately signed) in many cases with respect to
turnout. Fewer are statistically significant in modelling campaign participation. Political
interest and strength of party identification are significant with large effects, however, in all
four estimations.
In all four cases there is at least one contact variable that is statistically significant. In
three of the four cases it is the offline direct contacting, i.e. the face to face ‘gold standard’
method that has the strongest effect as measured by its significant and positive coefficient.
The exception is for the campaign participation variable in the UK which we return to discuss
below. With respect to turnout, offline direct contacting is actually the only measure of
contact that achieves statistical significance in both countries. More generally, there are
strong similarities in the results of this model across the UK and U.S. Notably the online,
direct contact variable has a very large coefficient in the UK, but does not attain significance
due to the very large standard error associated with it (due in turn to its very small N).
The results for campaign participation are interesting in that online contacting does
emerge as significant in predicting behaviour in both cases. In the U.S. offline direct contact
also proves to be influential although this is not the case in the UK where indirect contact
from friends and family is more important. While there are grounds for expecting these
differences that we discuss below, one important qualifier to make here is that the dependent
variable is measured differently in each case which may account for the disparities. Also in
the UK, because we switched to using BES data it was not possible to split indirect contact
into offline and online and the measure is an aggregate one. Of the control variables included
we can see that in both cases (not surprisingly) interest and partisanship are significant. Age
and gender are also significant in the UK although none of the demographics except for race
are significant in the U.S. The significance of race (in particular African-Americans but not
Hispanics), is interesting since it confirms results of other studies that ceteris paribus, Blacks
are actually more likely to turnout and more likely to participate actively in campaigns than
are whites. Certainly based on the official figures for 2012 turnout among Black turnout did
exceed that of Whites.
A second and even more major qualifier to our findings about campaign participation
is of course that we cannot rule out that the effect of online contact is actually capturing prior
behaviour of voters. Given the ‘pull’ nature of online contact whereby voters need to first
sign up to receive messages (as compared to the more random ‘push’ quality of offline
methods) it is clearly possible that those reporting having received messages from the
campaign are likely to have already signed up for such material and thus are already more
involved in the campaign. It was not possible to address this claim in the UK since the BES
did not include a variable as to whether a respondent had signed up to receive information
from the campaign. The 2012 ANES, however, does include a variable that asked if an
individual had signed up to receive information from the party during the campaign. Table 9
shows the results of the model with this new item.
Table 9 about here
The table shows clearly that the addition of this item removes the significance of the
online contact variable and so confirms that any effects of online contact reported earlier are
basically attributable to individuals having already engaged with the campaign in some way.
Of course it may be that the stimulus of online contact then provides an important basis for
maintaining and supporting this activity. However, overall these results clearly challenge the
idea that online contact generates campaign participation among the less active who do not
show political initiative by signing up online.
Discussion and Conclusions
This study has presented a comparative analysis of the frequency and impact of different
types of voter contact in general elections in the U.S. and UK. The main aim of the paper has
been to update the literature on contacting and to include contacts taking place via online
tools as well as more traditional offline methods. The central questions posed have been two-
fold: First, we have sought to establish how extensively the new forms of online contact are
in comparison to the more traditional modes that preceded it. Digital channels offer parties
and candidates a faster and potentially more convenient means of contacting citizens during
elections. In addition, through social media sites such as Facebook and Twitter, campaigns
have a whole new range of more indirect methods for reaching voters, with supporters
filtering messages out to their online networks in a new and highly extensive two-step flow
model. Are political actors availing themselves of these new channels? And if so, is this
increasing the overall amount of contact or are we seeing a displacement of older forms for
newer digital modes? The second major question posed has been about the effectiveness of
the newer modes of voter contact. If digital tools are now being used are they helping in the
effort to mobilize citizens and if direct and indirect forms are identified, to what extent is one
more useful or successful than another? Given that we have a comparative research design we
have also been able to look at whether the technology is used in a similar way in each country
and whether it has a stronger effect in certain institutional settings.
Overall it does appear that the intensity of offline campaigning in the U.S. has
decreased in presidential races over the past decade. Rates of informal contacting, however,
have risen slightly. In the UK, by contrast, although levels of official voter contact appear to
be quite high, this is due largely to leafleting. The use of more influential methods of direct
canvassing and doorstep mobilization has actually decreased over time. Interestingly, the
rates of informal persuasion have increased quite notably in the UK. In terms of the relative
rates of uptake of online campaigning the data has shown that U.S. campaigns have moved
quickly to update their efforts and now make significant use of online tools. In the UK,
however, uptake is much lower. Interestingly the levels of indirect online contact across the
two countries are more comparable showing that while the British parties and candidates may
not set much store in the web as a means of reaching their supporters, citizens themselves are
fairly active in using the internet to mobilize other voters. Overall then the results suggest that
while adoption rates vary, where the new tools are being incorporated into campaigns this
may be leading to a substitution of new methods for old. The most enthusiastic adopters of
online campaigning, however, appear to be citizens themselves who are using them to spread
political messages of support in their social networks. This holds for both countries.
In terms of our second question about the impact of online campaigning, the results
have revealed a qualified effect. Certainly in terms of voting offline personalized contact
remains the ‘gold standard’ and the only method that according to our data really works for
campaigns in mobilizing turnout. When it comes to wider campaign activities beyond voting,
however, initial results suggest a more positive story with efforts of parties to send messages
through email and the web being linked to a significant increase in commitment to help
parties and candidates campaign in both countries. This effect is however shown to be
essentially a proxy for prior activities by those voters in signing up to receive online updates
and information from the candidates. The effect of direct face to face or mail and phone
canvassing in the U.S. remains, however, as the trump card in the campaign contact hand. In
the UK it is the efforts of friends and family to mobilize that further increase interest in
helping a party or candidate rather than any additional official stimuli. These latter
differences are interesting in that they suggest that U.S. campaigns in general do a much
better of job in generating volunteers and assistance than their UK counterparts. The much
looser nature of party membership in America, the longer campaign period and lack of
official party members to draw on of course means that they need to be more persistent and
efficient in sourcing voluntary labour. Within the UK it seems that campaign activism is
much more related to the social circles in which an individual moves rather than whether a
party contacts them.
Despite not supporting our expectations, these results do offer a benchmark in the
study of online contacting. We have shown that its occurrence varies across national contexts
with U.S. campaigns being the most enthusiastic adopters. As well as being engaged in by
campaigns, our results have shown that online media also offers a way for ordinary voters to
spread the message and mobilize others to get involved in campaign. A fairly healthy
minority of the electorate now seem to be comfortable with doing this. While being on the
receiving end of this new mode of direct and two-step mobilization on voting seems to make
no difference to whether one votes, the more direct approach does appear to produce or at
least reinforce interest in working to help a party or candidate in their campaign. Further
research and more nuanced data on people’s prior campaign activity in particular is required
to unpack the nature of this relationship and assess how genuinely mobilizing such contact is.
Table 1: Modes of Voter Contact
MODE OFFLINE ONLINE
Source of Contact
Direct (campaign)
Q1 F2F Mail
Phone Leaflet
Q3 Email
Social network Twitter Website
Indirect (F&F)
Q2 F2F Mail
Phone Passing on leaflet
Leaflet
Q4 Email
Social network Twitter
Website Leaflet
Table 2 – Percentage of citizens contacted in the last elections in the UK and the U.S.: UK BMRB
2010 UK BES 2010 U.S. ANES
2012
Direct - Online 2.6 1.6 17.3
Direct - Offline 46.7 50.6 36.2
Indirect - Online 15.1 16.5
22.8
Indirect - Offline NA 38.9
Note: in the BES 2010, indirect contact does not specify mode.
Table 3 – % of citizens contacted offline in the UK, 2001-2010:
BES 2001 BES 2005 BES 2010
Direct - Telephone 7.9 7.5 4.3
Direct - Canvasser 23.8 21.5 13.9
Direct - Leaflet NA NA 45.4
Direct - Street NA NA 2.3
Indirect 11.5 15.7 16.5
Source BES post-election face to face surveys.
Table 4 – % of citizens contacted offline in the U.S., 2000-2008:
ANES 2000 ANES 2004 ANES 2008 ANES 2012
Direct - F-to-F; phone 35.6 43.5 43.2 31.9
Direct - Mail 61.1 NA NA 31.6
Indirect - F-to-F; phone
10.3 18.2 17.2 11.9
Indirect - Mail 23.7 NA NA 5.6
Source: ANES post-election face to face surveys.
Table 5. Logistic regression model of voting, UK 2010 Coef. Std.
Err. Gender (female) 0.05 0.14 Age 0.42** 0.05 Education 0.14* 0.06 Social class -0.27** 0.06 Interest in politics 0.47** 0.07 Party Id. 0.57** 0.08 Signed up online -0.10 0.41 Online indirect contact 0.21 0.21 Online direct contact 0.65 0.52 Offline direct contact 0.38** 0.14 Discuss 0.41** 0.15 Constant -2.09** 0.39 Pseudo-R2 0.24 N 1,545 Source: BMRB post-election face to face survey. **p<0.01, *p<.05
Table 6. Logistic regression model of voting, U.S. 2012 Coef. Std.
Err. Gender (female) -0.13 0.17 Age 0.27** 0.06 Education 0.42** 0.10 Income 0.25** 0.05 Race: white (ref.) - - Race: black 0.78** 0.24 Race: Hispanic -0.07 0.23 Interest in politics 0.32** 0.09 Party Id. 0.60** 0.08 Signed up online 1.59** 0.42 Battleground state 0.35 0.21 Online indirect contact -0.09 0.30 Offline indirect contact 0.30 0.24 Online direct contact 0.03 0.38 Offline direct contact 0.61* 0.24 Constant -4.99** 0.46 Pseudo-R2 0.26 N 1,754 Source: ANES post-election face to face survey. **p<0.01, *p<.05
Table 7. Negative Binomial Regression Model of Campaign Participation, UK 2010 Coef. Std.
Err. Gender (female) -0.17* 0.07 Age -0.12** 0.02 Education 0.04 0.03 Social class 0.04 0.04 Interest in politics 0.33** 0.04 Party ID. strength 0.33** 0.04 Marginal constituency -0.01 0.08 Indirect contact 0.26** 0.09 Online direct contact 0.56* 0.26 Offline direct contact -0.05 0.08 Constant -0.91** 0.15 Pseudo-R2 0.03 N 2,934 Source: BES post-election face to face survey. **p<0.01, *p<.05
Table 8. Logistic Regression Model of Campaign Participation, U.S. 2012 Coef. Std.
Err. Gender (female) 0.34 0.18 Age 0.08 0.07 Education 0.10 0.10 Income -0.06 0.06 Race: white (ref.) Race: black 0.60** 0.23 Race: Hispanic -0.51 0.27 Interest in politics 0.32** 0.09 Party Id. 0.51** 0.10 Battleground state 0.31 0.20 Online indirect contact 0.26 0.25 Offline indirect contact 0.15 0.21 Online direct contact 0.74** 0.24 Offline direct contact 0.71** 0.21 Constant -5.17** 0.53 Pseudo-R2 0.18 N 1,750 Source: ANES post-election face to face survey. **p<0.01, *p<.05
Table 9. Logistic Regression Model of Campaign Participation, U.S. 2012 (including signing up)
Coef. Std.
Err. Gender (female) 0.33 0.18 Age 0.08 0.07 Education 0.09 0.11 Income -0.04 0.06 Race: white (ref.) Race: black 0.57** 0.23 Race: Hispanic -0.51 0.27 Interest in politics 0.32** 0.09 Party Id. 0.51** 0.10 Signed up online 1.00** 0.27 Battleground state 0.28 0.20 Online indirect contact 0.15 0.25 Offline indirect contact 0.16 0.21 Online direct contact 0.47 0.26 Offline direct contact 0.75** 0.21 Constant -5.22** 0.53 Pseudo-R2 0.19 N 1,749 Source: ANES post-election face to face survey. **p<0.01, *p<.05
References Beck, Paul A. and Richard Gunther (2010), “Global Patterns of Political Intermediation”, unpublished manuscript. Beck, Paul A. et al. (2002), “The Social Calculus of Voting: Interpersonal, Media, and Organizational Influences on Presidential Choices”. American Political Science Review, 96: 57-73. Bergan, Daniel E., Alan S. Gerber, Donald P. Green & Costas Panagopoulos (2005) ‘Grassroots Mobilization and Voter Turnout in 2004.’ Public Opinion Quarterly, 69:760-777. Bond et al. 2012 ‘A 61-million-person experiment in social influence and political mobilization.’ Nature 89, 295–298 Blydenburgh, John. 1971. ‘A Controlled Experiment to Measure the Effects of Personal Contact Campaigning.’ Midwest Journal of Political Science 15(2): 365-381. Bochtel and Denver.1971. "The Impact of the Campaign on the Results of Local Government Elections," British Journal of Political Science 2:239-260. Cutright, Phillips. 1963. "Measuring the Impact of Local Party Activity on the General Election Vote." Public Opinion Quarterly. 27:372-386. D’Alessio, D W. 1997. ‘Use of the Web in the 1996 US Election.’ Electoral Studies 16: 489-501. Dale, A. and A. Strauss (2007a) Text Messaging as a Youth Mobilization Tool: An Experiment with a Post-Treatment Survey’ Paper presented at the annual meeting of the Midwest Political Science Association, Chicago http://www.iop.harvard.edu/var/ezp_site/storage/fckeditor/youth_text.pdf Dale A. and A. Strauss. (2007b) Fact Sheet on Youth Vote and Text Messaging http://www.studentpirgs.org/new-voters-project/research/youth-vote-and-text-messaging Davis, R., Owen, D., Taras, D. and S.J. Ward. (2008) (eds) Making a Difference? Internet Campaigning in Comparative Perspective, Lexington Books: Lanham MD. Foot, K. and S. Schneider. 2006. Web Campaigning Cambridge, MA: The MIT Press Gerber, Alan S. & Donald P. Green (2008) Get Out the Vote: How to Increase Voter Turnou. (2nd ed) Washington DC: Brookings Instittution Gerber, Alan S. & Donald P. Green (2000) The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment». American Polilitical Science Review, 94:653-663.
Gibson, R. K. 2013. ‘Social Media and the Rise of Citizen-Initiated Campaigning’ Party Politics (forthcoming) Gibson, R. K. and I. McAllister. (2011) Do Online Election Campaigns Win Votes?: The 2007 Australian YouTube Election’ with Ian McAllister Political Communication 28 (2) 227-244 Gibson, R. K. and Ward, S. (1998) ‘UK Political Parties and the Internet—“Politics as Usual” in the New Media?’ Harvard International Journal of Press/Politics 3 (3), 14–38. Gibson, R. K., Margolis, M. Resnick, D. and Ward, S. (2003a). ‘Election Campaigning on the WWW in the US and UK: A Comparative Analysis.’ Party Politics 9 (1), 47-76. Gosnell, Harold. 1927. Getting out the Vote: An Experiment in the Stimulation of Voting. Chicago: University of Chicago Press. Harfoush, R. 2009. Yes We Did: An Inside Look at How Social Media Built the Obama Brand. Berkeley, California: New Riders Huckfeldt, R. et al. (1995). “Political environments, cohesive social groups, and the communication of public opinion.” American Journal of Political Science 39(4): 1025–1054. Huckfeldt, R., and J. Sprague. (1995). Citizens, politics, and social communication: Information and influence in an election campaign. Cambridge University Press New York. Kluver, R. Jankowski, N., Foot, K. and S. Schneide . (eds), (2007) The Internet and National Elections: A Comparative Study of Web Campaigning. London Routledge. Karp, Jeffrey, Susan A. Banducci & Shaun Bowler (2007), «Getting Out the Vote: Party Mobilization in a Comparative Perspective». British Journal of Political Science, 38:91-112. Krueger, B. 2008. ‘Opt In or Tune Out: Email Mobilization and Political Participation’ International Journal of E-Politics Kramer, Gerald. 1971. "The Effects of Precinct Level Canvassing on Voting Behavior." Public Opinion Quarterly. 34:560-572. Leighley, J. E. 1990. “Social interaction and contextual influences on political participation.” American Politics Research 18(4): 459. Magalhães, Pedro C. (2010), “Mobilization, Informal Networks, and the Social Contexts of Turnout”, unpublished manuscript. Magalhães, Pedro C. (2007), “Voting and Intermediation: Informational Biases and Electoral Choices in Comparative Perspective”, in R. Gunther, H-J Phule, and J.R. Montero (eds.), Democracy, Intermediation, and Voting in Four Continents. Oxford: Oxford University Press. Margolis, M. and D. Resnick. 2000. Politics as Usual? The Cyberspace Revolution. London: Sage.
Nickerson, David. W. 2007. Quarterly Journal of Political Science 2: 369-379 Panagopoulos, Costas & Peter L. Francia (2009), «Grassroots Mobilization in the 2008 Presidential Election». Journal of Political Marketing, 8:315-333. Partheymüller, J. and Schmitt-Beck, R. 2012. “A ‘Social Logic’ of Demobilization: The Influence of Political Discussants on Electoral Participation at the 2009 German Federal Election.” Journal of Elections, Public Opinion & Parties, 22,4: 457-478. Popkin, S. L. 1991. The reasoning voter: Communication and persuasion in presidential campaigns. Univ of Chicago Pr. Rosenstone and Hansen. 1993. Mobilization, Participation, and Democracy in America. MacMillan:New York. Schmitt-Beck, R. 2003. “Mass Communication, Personal Communication and Vote Choice: The Filter Hypothesis of Media Influence in Comparative Perspective.” British Journal of Political Science 33(02): 233-259. Sniderman, P. M et al. 1991. Reasoning and choice. Cambridge University Press New York. Stollwerk, Alissa F. 2006. "Does E-mail Affect Voter Turnout? An Experimental Study of the New York City 2005 Election." Unpublished Manuscript. Institution for Social and Policy Studies, Yale University. Sudulich, M. L. and Wall, M. (2010) ‘Every Little Helps. Cyber Campaigning in the 2007 Irish General Election.’ Journal of Information Technology and Politics 7 (4), 340-355. Vergeer, M. (Ed.) (2013). The state of online campaigning in politics (special issue), New Media & Society. Wielhouwer, Peter W. & Brad Lockerbie (1994), «Party Contacting and Political Participation». American Journal of Political Science, 38:211-229. Williams, A. P. and K. D. Trammell. 2005. ‘Candidate Campaign E-Mail Messages in the Presidential Election 2004.’ The American Behavioral Scientist 49(4): 560-574. Zaller, J. 1992. The Nature and Origins of Mass Opinion. Cambridge, UK: Cambridge University Press.
Appendix
Variables in US data (ANES 2012):
Voting: 1=Voted in the 2012 Presidential election; 0=Didn’t vote.
Campaign participation: 1= wore campaign button or bumper sticker, attended rally, did other works for candidates, gave money to either party or candidate; 0= didn’t do any of these activities.
Online direct, offline direct, online indirect, offline indirect contact: 1= Contacted; 0= Not contacted.
Gender: 1=Female; 0=Male.
Age: 1= 17-24; 2= 25-34; 3= 35-44; 4= 45-54; 5= 55-64; 6= 65+.
Education: 1= No High School, 2= High School; 3= Some college; 4= 4 years of college/Bachelor/post-graduate.
Family income: 1=less than $10,000, 2=$10,000-14,999, 3=$15,000-$24,999, 4=$25,000-$49,999, 5= $50,000-$99,999, 6=$100,000-$149,999, 7=$150,000-$249,999, 8=over$250,000.
Race: Black (1=yes), Hispanic (1=yes), White (reference category).
Party Identification: 1=Pure independents; 2= independents who lean either Democratic or Republican; 3=weak Democratic and weak Republican; 4= strong Democrat and strong Republican.
Interest in politics: “How often do you pay attention to what is going on in politics?” 1=never; 2=some of the time; 3=half the time; 4=most of the time; 5=always.
Signing-up online: “Prior to, or during the campaign, did you use the internet or your mobile phone to sign up for information or alerts (e.g. e-newsletters, text messages, RSS or blog feed) from a party or candidate”. 1= Yes, 0= No.
Battleground State: 0= non battleground state; 1=battleground state (based on Real Clear Politics poll averages)
Variables in UK data (BMRB 2010):
Voting: 1=Voted in the 2012 Presidential election; 0=Didn’t vote.
Online direct, offline direct, online indirect contact: 1= Contacted; 0= Not contacted.
Gender: 1=Female; 0=Male.
Age: 1= 17-24; 2= 25-34; 3= 35-44; 4= 45-54; 5= 55-64; 6= 65+.
Education: 1=no Secondary; 2=Secondary; 3=College; 4=Further between college and University; 5= University degree or post-graduate.
Social class: 1= E Non Working; 2= D Working class; 3= C2 Skilled working class; 4= C1 Lower middle class; 5= AB Upper middle class or middle class.
Party Identification: 0=Not close to any party; 1= not very strong Party Id.; 2= Fairly strong party Id.; 3= Very strong party Id.
Interest in politics: 0=None at all; 1= Not very much; 2=Some; 3= Quite a lot; 4= A great deal.
Discuss politics: with family or friends: 1= Yes (in the past 12 months); 0= No.
Variables in UK data (BES 2010):
Campaign participation: How likely to work for a party or candidate in a campaign, 0= Very unlikely; 10=Very likely.
Online direct, offline direct, indirect contact: 1= Contacted; 0= Not contacted.
Gender: Female (1=Yes).
Age: 1= 17-24; 2= 25-34; 3= 35-44; 4= 45-54; 5= 55-64; 6= 65+.
Education: 1=no Secondary; 2=Secondary; 3=College; 4=Further between college and University; 5= University degree or post-graduate.
Social class: 1= Never worked or other; 2= Manual occupations, including sales and service; 3= Small business owner; 4= Non-manual, including clerical.
Party Identification: 0=Not close to any party; 1= not very strong Party Id.; 2= Fairly strong party Id.; 3= Very strong party Id.
Interest in politics: 0=None at all; 1= Not very much; 2=Some; 3= Quite a lot; 4= A great deal.
Marginal Constituency: 0 = non marginal; 1 = marginal (constituencies in which the difference in vote share between the first and second parties was under 10% in 2005).
