Keeping an Eye on the People: Who Has Access to MPs on Twitter?

Abstract

Twitter is credited for allowing ordinary citizens to communicate with politicians directly. Yet few studies show who has access to politicians and whom politicians engage with, particularly outside campaign times. Here, we analyze the connection between the public and members of parliament (MPs) on Twitter in the Netherlands in-between elections in 2016. We examine over 60,000 accounts that MPs themselves befriended or that @-mentioned MPs. This shows that many lay citizens contact MPs via Twitter, yet MPs respond more to elite accounts (media, other politicians, organized interests,…), populist MPs are @-mentioned most but seem least interested in connecting and engaging with “the” people, and top MPs draw more attention but hardly engage—backbenchers are less contacted but engage more.

Keywords

Twitter politics political communication members of parliament social media populism

Introduction

Barack Obama is often called “the social media president” (Katz, Barris, & Jain, 2013). This is not only because of his successful use of social media in his election campaigns (Bimber, 2014, p. 131). Also after his election, he wanted “to draw on the power of digital engagement to tap into the wishes and talents of the American people” (Katz et al., 2013, p. 3). In his own words, “I felt that some of the things that we were doing to help us get elected could also be used once we were elected” (Obama, 2013). He is not alone in his ideas. As Jungherr (2016, p. 79) notes, “politicians see public exchanges referring to politics on Twitter as indicators of public opinion.”

Yet beyond the Obama case, empirical research on the matter is scarce. Indeed, there are at least two gaps in the literature on social media, public opinion, and politics. The first lacuna is related to the Burkean, top-down perspective that guides most studies in the field (Jackson & Lilleker, 2009, p. 246). Typically, these studies focus on political actors and their messages (Jungherr, 2016, p. 76). Yet as Vaccari and Valeriani (2015, p. 1028) put it succinctly, “the political Twitter-verse is not just about talking (writing) but also about listening (reading).” Social media are credited for allowing ordinary citizens and “outsiders” to approach politicians directly, but few studies have focused on whom politicians listen to on social media. A Dahlian perspective, analyzing who has access to politicians (e.g. who politicians follow)¹ and whom politicians listen to (e.g. who they reply to and who they retweet), is therefore badly needed.

The second lacuna is that most studies focus on campaign times (e.g. Hosch-Dayican, Amrit, Aarts, & Dassen, 2014; Jungherr, Schoen & Jürgens, 2015; Vergeer & Hermans, 2013), while only few study politicians’ everyday online practices (Larsson, 2015). However, during campaigns, politicians adopt a “broadcasting” style of communication (Jungherr, 2016, p. 76), while social media have an “always-on” logic and also offer politicians many opportunities outside of the campaign (Larsson, 2015, p. 150). In-between elections, politicians can use social media to forge long-lasting relationship by engaging with the public. Similarly, social media provide “the general public” an easy means to express its “opinion” to their representatives directly (Graham, Broersma, Hazelhoff & van’t Haar, 2013, p. 710). In other words, the Burkean, top-down perspective is useful in campaign times, but a Dahlian perspective may be more useful when politicians are not campaigning.

In this study, we analyze to what extent there is a vertical connection between the public and politicians on Twitter (Kruikemeier, 2014). Indeed, while a horizontal connection is a connection between political elites (such as members of parliament [MPs]) and other elite accounts, vertical connections between political elites and lay citizens are very important as well. Such vertical connections determine what politicians “see” of public opinion without having to actively search for it. “Seeing” public opinion on Twitter does, however, not necessarily equal noticing it. Therefore, we also examine the engagement of politicians with the accounts they are connected with. Specifically, we examine the over 60,000 accounts that were connected with all Dutch MPs on Twitter (befriended or @-mentioning) and of those the ones with which the MPs engaged (retweeted friends and @-mentioners @-mentioned by MPs). We also examine which types of MPs are more likely to vertically connect or engage with lay citizens. All data come from in-between elections (May–August 2016; cf. Mejova & Wever, 2014).

The Dutch case is a particularly interesting as the country is a digital front-runner (see Vergeer & Hermans, 2013). It has a very high Internet penetration rate (94% in 2013) and a relatively high Twitter penetration rate. It has 3.3 million users on a population of close to 17 million (Oosterveer, 2013) and a reported Twitter penetration rate among the population of around 20% in 2016 [Statista, 2017; Telecompaper, 2017]). Moreover, over 95% of the MPs have adopted Twitter and they themselves say Twitter is an important way to connect with the people (Hosch-Dayican et al., 2014; Jacobs & Spierings, 2016; Weber Shandwick, 2014). Hence, if we do not find a vertical connection and engagement with public opinion here, it is unlikely to exist in comparable countries.

Our addition to the literature is as follows: Currently, researchers only focus on which users politicians recognize by tweeting messages to them or whom they interact with (cf. Tromble, 2016a). They do not focus on who is tweeting to politicians and who politicians follow. Our study does both and thereby allows us to identify discrepancies between the two. Our main findings are 3-fold. First, while a substantial number of lay citizens try to draw the attention of MPs, the average MP is not likely to connect with such accounts. Instead, MPs focus mostly on elite accounts (e.g., media, other politicians, and organized interests accounts). If Twitter does not live up to its equalizing potential, this is not because no lay citizens want to. Second, MPs of populist parties seem least interested in connecting and engaging with “the” people. They rather create echo chambers with “their” people. They befriend few accounts of lay citizens, but those befriended are retweeted relatively often. At the same time, populist MPs are @-mentioned by the most accounts, but these accounts are often ignored. Third, MPs’ parliamentary position matters. Top MPs draw a lot of attention from citizens (@-mentions) but hardly engage with them and rarely befriend them. Conversely, backbenchers are less connected, but they engage more with their connections.

Theoretical Framework

Using Twitter as an indicator of public opinion

During election campaigns, politicians try to attract votes. In-between elections politicians have other motivations for communication too. They typically wish to implement specific policies and represent their voters (cf. Strom, 1990).² Therefore, they also inform themselves about what citizens think and test certain ideas they have. Under such circumstances, politicians are likely to keep an eye on public opinion (i.e., what citizens want, think, and feel). Politicians have several tools and channels at their disposal to do so such as opinion polls, internal polling by the party, communication with citizens directly, and traditional media (e.g. Van Aelst & Vliegenthart, 2014). More recently, Twitter has been added to these tools, and it has several features that give it a distinct added value (Hosch-Dayican et al., 2014).

As Jacobs and Spierings (2016, p. 20) stress, Twitter has “unique characteristics” that make it different from, for instance, opinion polls and media coverage. First, Twitter is inexpensive and does not require specialist technical knowledge: Anyone can open an account and send and read tweets (Jacobs & Spierings, 2016, p. 22).³ Content generation “by lay persons” and “amateur activity, by those who may have authentic knowledge and information access” is at the heart of social media such as Twitter (Klinger & Svensson, 2014, p. 6). Especially for politicians who specialize in the less salient topics, such information is useful. The topics on social media are vastly more diverse than on traditional media and thereby paint a richer picture of public opinion, one that taps into its “sentiment” more directly (cf. Chadwick, 2013).⁴ Social media such as Twitter allow “hearing the voice of the people” (Katz et al., 2013, p. 13).⁵

Second, Twitter is unmediated (Klinger & Svensson, 2014, p. 10). Journalists filter and edit content in traditional media and pollsters decide the focus of public opinion polls. Twitter on the other hand offers raw information on any topic. The public can directly communicate with politicians, and politicians can directly view what “the public” wants, thinks, or feels (e.g., Jacobs & Spierings, 2016, p. 64).⁶

Third, Twitter is characterized by speed (Klinger & Svensson, 2014, p. 8). Opinion polls do not allow politicians to (re)act swiftly and similarly traditional media have a relatively slow news cycle compared to social media. Twitter is fast and messages can go viral within hours (Klinger & Svensson, 2014), thereby potentially triggering spillover effects to traditional media (Chadwick, 2013). As such, it can act as a “weather forecast” (Bleijerveld, personal interview, August 21, 2013).

By no means is it our claim that Twitter is the only tool used by politicians, but it is one of the tools that they can use, that they say they use (Weber Shandwick, 2014), and, most interestingly, one that allows traditionally excluded people to have quick and direct access to politicians.

Vertical connections and engaging with the public

Some politicians might use advanced software to keep track of what is going on Twitter. The information collected this way typically refers to general trending topics on twitter, but it tells us less about who has access to politicians. Our focus is on the accounts MPs vertically connect with, particularly who it is that politicians look at on Twitter for information (who is followed by MPs) and who tries to draw the attention of politicians (who @-mentions MPs).

While politicians—and journalists alike—are aware that “users tend not to be representative of a population as a whole” (Jungherr, 2016, p. 77), they still seem to treat Twitter as a proxy for public opinion (Jacobs & Spierings, 2016, p. 96). A recent study by Freelon and Karpf (2015, p. 398) showed that politicians, journalists, and celebrities are the most active in salient political discussions. So even if politicians form vertical connections with the Twitter public, this means that they get a distorted view of what public opinion is. Therefore, it is important to know with whom politicians form connections on Twitter and to examine to what extent lay citizens are part of these connections.

These issues echo a broader discussion in the field of the effects of technological innovation: Whether they open up politics and allow new voices to be heard (the equalization thesis) or whether new technologies simply replicate existing inequalities (the normalization thesis; Gibson & McAllister, 2015; Vergeer & Hermans, 2013). Specifically, one can expect that if the equalization perspective holds true, politicians will mainly have vertical connections with public opinion—lay citizens—on Twitter (not e.g., organized interests, other politicians, and journalists). Conversely, from a normalization perspective, one would expect that politicians mainly connect (horizontally) with, for example, organized interests, other politicians, or journalists (not lay citizens). In the first part of our empirical analysis, we will analyze which of the two perspectives is most aligned with the empirics.

Research Question 1a: To what extent do politicians connect with (lay citizens among) the public on Twitter?

It is one thing to connect with lay citizens on Twitter but that does not necessarily mean that politicians notice and engage with them, for example, by replying to them or retweeting them. After all, the first is merely unilateral and says very little about whether politicians noticed what citizens said (or tried to bring to the politician’s attention). Engagement, as a form of reciprocal communication, tells us far more in this respect. Indeed, as Klinger and Svensson (2014, p. 10) note, online information spreads so fast that individual tweets get snowed under quite quickly. It is therefore important to check what type of accounts politicians engage with. If they engage with an account, we can be certain it was genuinely noticed (however short that attention to it may have been). From the campaign literature, we know that politicians mainly focus on “elite accounts” and journalists in particular (Jungherr, 2016, p. 76; Kreiss, 2016, p. 1475). During campaign times, it therefore seems that politicians mainly behave in line with the normalization perspective. However, as mentioned above, during such times, politicians are in a broadcasting mode. It could be that they behave differently in-between elections. In the first part of our analysis, we therefore also examine the following research question:

Research Question 1b: To what extent do politicians engage with (lay citizens in) their network on Twitter?

Differences between MPs

So far, we have considered politicians in general, but it is likely that some politicians connect and engage more than others. Even in the campaign literature, little is known about who is more likely to connect and engage with the public via social media (Jungherr, 2016, p. 77; Vergeer & Hermans, 2013).

Regarding connecting with the public on Twitter, it seems reasonable to expect that the top politicians or politicians from large parties will be less reliant on Twitter as they have the resources to issue polls themselves and generally deal with topics that are well covered in the media and in general polls. We can therefore expect them to be less likely to follow accounts of lay citizens. Conversely, parties and politicians that focus on traditionally excluded voices or topics, ones that typically are less visible in traditional media or in opinion polls, are more likely to take to Twitter to feel the pulse of (certain segments of) public opinion. As such, we can expect backbenchers (i.e., lower ranked, less prominent politicians), MPs from smaller parties, female MPs, and MPs with an ethnic-minority background to connect more with lay citizens (cf. Bailey, Steeves, Burkell, & Regan, 2013; Lassen & Brown, 2010). Lastly, one can expect that MPs from populist parties connect with citizens “as a means of demonstrating that the politician and party are ‘of the people’ and not the political establishment” (Tromble, 2016b, p. 9).

At the same time, one can expect that the “public” mainly wants to draw the attention of the most influential and well-known politicians and parties (via @-mentions) instead of backbenchers and MPs from smaller parties. However, building on the aforementioned reasoning, one can also expect that female MPs, MPs from an ethnic-minority background, and MPs from populist parties are more likely to be contacted by lay citizens. Indeed, the former two can function as a contact person for these underrepresented groups. And MPs from populist parties can be contacted by citizens who feel excluded by “the mainstream establishment.”

Hence, in the second part of our analysis, we wish to answer the following question:

Research Question 2a: What type of MP is vertically connected to lay citizens on Twitter (by following them or being @-mentioned by them)?

Regarding engagement,⁷ we can fall back on some studies from the campaigning literature (though it is obviously not certain that their results also hold in-between elections). Gibson and McAllister (2014, p. 531) suggest that politicians from postmaterialist/green parties engage more with the public because interactivity would fit their party identity.⁸ Similarly, Jacobs and Spierings (2016, pp. 114–120) find that while big parties and top politicians seem more focused on journalists, at least some backbenchers are more likely to engage with lay citizens.⁹ Tromble (2016a, p. 13), who analyzed politicians’ interaction on Twitter in-between elections, found that female MPs are more likely to hold conversations with their followers. If female MPs indeed function as a contact person and also themselves identify as such, one would expect more engagement from them as well. While Tromble did not include ethnicity in her analysis, a similar pattern for ethnic-minority MPs can be expected. Tromble also found populists are less likely to hold conversations, which is not entirely what one would expect based on the reasoning outlined earlier. Indeed, if populists want to demonstrate that they are a party of the people, one would expect them to interact with ordinary citizens.¹⁰ It is therefore worthwhile to see whether our analysis replicates this finding.

The second part of our analysis will address this:

Research Question 2b: What type of MP engages with lay citizens on Twitter (by retweeting or actively @-mentioning those with whom they connected)?

Method and Data¹¹

Operationalizing the Concepts

As discussed, we operationalize connections by (1) being followed by an MP and (2) an MP being @-mentioned. Both make accounts “visible” to that MP. By following an account, an MP actively links to that person or organization and the tweets of that account will appear on the time line of the politician.¹² The MP grants these accounts—“friends” in Twitter jargon—direct access to the information flow toward the MP. Contrarily, from an account holder’s perspective, the most direct way to try to get an MP’s attention is @-mentioning that MP. The MP gets a notification and is drawn to the message. These incoming @-mentions (from the MP perspective) thus indicate which people or organizations try to push their views to the MP.

Befriending and incoming @-mentions measure reach out or potential visibility. Therefore, we also look at further engagement by looking at (3) retweets of accounts the MP follows and (4) outgoing @-mentions by the MP of the accounts that @-mentioned them. While many users explicitly claim that a retweet is not an endorsement, a retweeted message is at least noticed and deemed interesting enough to be retweeted (Metaxas et al., 2015; see also boyd, Golder, & Lotan, 2010). The MP in a way raises the status of the tweet to something that is worth of being shared in a larger public debate (Klinger & Svenson, 2014).¹³ Similarly, outgoing @-mentions of MPs serve as an indication that the account @-mentioning an MP is getting the MP’s attention.

Twitter Data

We collected data on all 150 Dutch MPs of the June 16, 2016 parliament.¹⁴ No less than 146 (96%) had a twitter account, 140 of the accounts tweeted at least once during our period of study (May–August 2016), and 142 had friends. Of the 146 MPs, we collected and verified their twitter account handles and uploaded these to our script.

Our Twitter data collecting relies on the Representational State Transfer (REST) and Search Application programming interface (APIs).¹⁵ Via the REST API, the friends of an account are accessible as are the statuses on the user’s time line (including retweets). We collected and stored their openly accessible user information including names and short biographical descriptions.¹⁶ Regarding the retweets, the accessible backlog only goes back 3,200 statuses per account.¹⁷ As the final data collection started early August 2016, the 3,200 statuses trail allows us to include all retweets of MPs from May to August 2016. Of all retweeted accounts, we selected those of which the retweeted account was also friend of the retweeting MP (58% of the retweet-MP dyads).

Information on @-mentioning an MP cannot be accessed from the MP’s account.¹⁸ The @-mentioning data are therefore collected through the Search API, searching for the text “@[MP’s screen_name]” in Tweets of the last 7 days.¹⁹ Once found, the account name of the @-mentioning account was collected. Running daily searches, we could collect all MPs incoming @-mentions for August.²⁰ Next, from the MPs’ statuses, we obtained the @-mentions by the MPs (outgoing @-mentions) for the same month and selected those which involved @-mentioning an account that @-mentioned the MP.

The number of friend and @-mention dyads and unique accounts is given in Table 1 in the Findings section.

Table 1.

Summary Statistics of MPs’ Friends, Retweeted Friends, Incoming @-Mentions, and Reciprocal @-Mentions.

Variable	Mean^a	Max.	#MPs With 0	Median	Total # of Dyads	# Unique Accounts
Friends of MP	783	5,206	8	547	117,438	50,065
Friends retweeted by MP May–August	32	147	19	27	4,807	2,604
Accounts @-mentioning MP August	163	6,056	9	43	24,487	13,800
@-mentioning account being @mentioned by MP	13	119	35	5	1,964	1,546

^aThe denominator is 150 as all MPs can open a Twitter account.

Politicians, Parties, Pundits, and People

The information collected above only provides a general overview of the connection and engagement of MPs with other accounts. We are also interested in who or what kind of account has access to politicians. We therefore developed a semiautomated procedure to classify as many accounts as possible. We build on the seminal classifications by Vaccari and Valeriani (2016) and Graham, Broersma, Hazelhoff, and van’t Haar (2013), distinguishing:

other MPs;

non-MP political accounts including party officials, local council members, MEPs, party staff, elder(wo)men, mayors, provincial governors, party accounts, former MPs, local party accounts, and foreign politicians;²¹

government accounts including accounts of ministers and government organizations;

news accounts including journalists, media organizations, commentators, and news accounts but not including sports and showbiz news;

civil society and business interest including nongovernmental organization, their leaders, lobbyists, and other accounts clearly focused on public interests, group representation, and collective issues;

science and research including academic institutes and statistical bureaus, professors, and other scientific university employees;

entertainment and commerce including sports, showbiz, humoristic accounts, and commercial organizations (but not organized business interest).

Accounts can fall in multiple categories. The list above is used hierarchically.²² To classify, we applied four steps.²³ (1) A first round of automated coding based on exhaustive (hand-checked) lists of account names of all MPs, Members of European Parliament (MEPs), and parliamentary candidates in the 2012 elections, government officials, and parties as collected for prior studies (e.g., Jacobs & Spierings, 2016; Spierings & Jacobs, 2014). Moreover, we took relevant accounts from the main Dutch top100 twitter lists on, for example, political accounts, media accounts, NGO’s, and television (Berger, 2016). The accounts on these lists are ranked by the number of followers. To be on the list, they needed (a) to be based in the Netherlands or (b) to tweet in Dutch. The composers of the lists classified accounts as a certain group (e.g., politics) based on information in the self-reported Twitter description and local knowledge (Berger, 2016). They were hand-checked for this study. As the project is running for several years and is regularly updated, the lists are a reliable source to assure that prominent accounts in the groups at the core of this study are not overlooked. (2) For all friends, we searched their Twitter description for (combinations of) words identifying the account. For each category, (long) lists or identifiers were developed including alternative spellings and abbreviations.²⁴ For each identifier term, we looked at the Twitter description of a random sample of accounts it returned to check whether the search terms were too generic. If so, search terms were dropped or refined. For instance, the term “reporter” returns journalists as well as sports reporters, so the script was expanded to identify people with “reporter” and “sport” in their description and those were then recoded from “news account” to “entertainment and commerce” (see above). We also searched the account names for mentioning parties.²⁵ (3) All remaining accounts were ranked by number of MP followers. Those with more than 10 MP followers were hand-coded. They were looked up on Twitter and the Internet at large. Consequentially, many more, political, media, and civil society and business accounts were identified. This step is particularly important as these accounts weigh relatively strongly on the overall numbers. Overlooking them would have led to overestimating the number connections between MPs and ordinary citizens. Our first three steps identified over 60% of the friends and 80% of the retweeted friends. (4) From the remaining unclassified accounts for each indicator (friends, retweeted friends, incoming @-mentions, and outgoing @-mentions), we drew random samples of 100 accounts. These accounts were examined and hand-coded (see above).²⁶ The numbers for these samples were extrapolated to all unclassified accounts. The unadjusted numbers are provided in Appendix Table A1.²⁷

Differences Between MPs

For the explanatory analyses (Research Questions 2a and 2b), the dependent variables are the number of accounts befriended, retweeted, and @-mentioned by the MPs. We linked our MP-level aggregated Twitter data to information about MPs’ position in the party (political leader, top5 in parliament, list position—the higher the number, particularly of large parties, the more likely a politician is a backbencher), sex, ethnicity, age (control variable [e.g., Lassen & Brown, 2010]), and party affiliation (D66 and GL are postmaterialist parties (Jacobs & Spierings, 2016); PVV and SP are considered populist parties (Schumacher & Rooduijn, 2013); PvdA and VVD are the two “bigger” parties and in the government coalition). To analyze these data, we use Ordinary Least Sqaures (OLS) regression models (IBM SPSS Statistics 23), and to obtain more normally distributed dependent variables, the natural logarithms of these are used.

Findings

The Extent to Which MPs Connect and Engage

Table 1 shows that a direct connection with the broader public is restricted to a minor part of “the public.” The median Dutch MP follows 547 other Twitter accounts. Collectively, the 150 MPs follow only 50,065 unique accounts (compared to about 3.3 million registered Dutch Twitter accounts). Moreover, merely 5% of these unique friends (2,604 accounts) were retweeted by an MP from May to August. The median MP retweeted posts of 26 accounts she follows. These numbers suggest that MPs strongly select which accounts are worth following and an even more strongly which are retweet-worthy.

The bottom half of Table 1 shows that in the month of August, 13,800 unique accounts contacted at least one MP via Twitter. A whopping 89.4% of them were not befriended by that MP. The median MP was @-mentioned by 43 accounts.²⁸ Moreover, MPs rarely engage with these accounts: Only 8.0% of the accounts contacting an MP (1,964/24,487) were also @-mentioned by the MP, and 26 MPs were @-mentioned at least once but did not @-mention anyone back. So if Twitter is facilitating a direct connection between politicians and the general public, the practice does not live up to this potential and seems one-sided. Lay citizens are taking the initiative to provide MPs with their opinion, but the typical MP does not seem to reciprocate.

Type of Account MPs Connect and Engage with

Table 2 shows that of the 117,438 MP–friend connections, 63% was with other political or with media accounts. Another 16% represent civil society and business interest. Only 13% MP–friend connections involved ordinary citizens. This weak direct linkage of MPs with the general public is even more pronounced for the retweets: Only 4% of MP-retweeted friend dyads involved ordinary citizens being retweeted. In contrast, the proportion of other MPs increased from 7% (friends relations) to 16% (the retweeted friend dyads). Clearly, MPs do not focus much on ordinary citizens directly. If they obtain information about public opinion on Twitter, it is through media and organized interests’ accounts.

Table 2.

What Kind of Accounts Are Followed by, Retweeted by, @-Mention and Are @-Mentioned by MPs.

Account Type	Friends of MP	Retweeted Friends	MP Being @-Mentioned	MP @-Mentioning Back
%MPs	7	16	0.5	4
%Non-MP political	31	27	14	26
%Government	3	4	0	1
%News	22	26	5	15
%Civil society and business interests	16	19	22	24
%Science and research	3	2	0.5	3
%Entertainment and commerce	5	3	7	5
%Lay citizens	13	4	50	23
Total N	117,438 (=100%)	4,807 (=100%)	24,487 (=100%)	1,964 (=100%)

Note. Classifications without sample-based estimation are given in Appendix Table A1; percentages are calculated with the number of dyads as denominator (see Table 1).

Accounts that @-mention MPs include a large share of lay citizen accounts: 50%, or over 12,000 @-mention relations. Yet few also receive @-mentions from these MPs. Of the 1,964 returned @-mentions, only 23% are to lay citizens. This means that of the lay citizens, @-mentioning an MP only a mere 4% is @-mentioned by that MP (0.23 × 1,964)/(0.50 × 24,487). This is in stark contrast with lay citizens actually asking for the attention of politicians on Twitter. Relatively speaking, MPs have far more reciprocal @-mention relations with their colleagues, other political accounts, media, and organized interests.

Differences between MPs

In this section, we explore how MPs differ regarding their connection and engagement behavior (Research Questions 2a and 2b). Table 3 presents the summary statistics disaggregated by party, while Tables 4 and 5 show the explanatory models. As the engagement models in those tables (Models 4.3, 4.4, 5.3, and 5.4) include the respective connection measurement as a control, the other coefficients should be read as relative engagement.²⁹

Table 3.

Summary Statistics of MPs’ Friends, Retweeted Friends, Being @-Mentioned, and Reciprocal @-Mentions per Party.

Party	Friends of MP	Retweeted Friends	MP Being @-Mentioned	MP @-Mentioning Back
VVD (41)—bigger party	904	34	34	7
PvdA (36)—bigger party	854	28	56	9
D66 (12)—postmaterialist	928	46	262	25
GL (4)—postmaterialist	896	34	339	30
PVV (12)—populist	212	23	710	11
SP (15)—populist	504	34	160	22
CDA (13)	1,015	36	169	15
CU (5)	378	31	215	31
Other^a (12)	812	24	198	10
All 150 MPs	783	32	163	13

Note. The denominator is all MPs of the party. ^aAll delegations with three or fewer seats: SGP (3), PvdD (1), 50Plus (2), VNL (2), DENK (2), and De Klein and Houwers (1 + 1).

Table 4.

Explanatory OLS Regression Models of Number of Friends and Retweets.

Explanatory Variable	Number of Friends (Log)		Number of Lay Citizen Friends (Log)		Number Retweeted Friends (Log)		Number Retweeted Lay Citizens Friends (Log)
Explanatory Variable	B	SE	B	SE	B	SE	B	SE
Parties
VVD—coalition	Ref.		Ref.		Ref.		Ref.
PvdA—coalition	0.026	.220	0.096	.301	−0.240	.275	−.091	.214
D66—postmaterialist	0.172	.346	0.359	.473	1.077*	.433	.270	.337
GL—postmaterialist	0.377	.521	0.512	.712	0.217	.653	−.101	.507
PVV—populist	−1.410***	.341	−1.382***	.466	1.105*	.453	.897*	.342
SP—populist	−0.297	.321	−0.324	.439	0.803*	.403	.331	.313
CDA	0.396	.329	0.591	.450	0.453	.414	.032	.322
CU	−0.178	.480	0.052	.656	0.866	.600	.117	.466
Other party^a	0.208	.371	0.554	.508	0.793†	.465	.305	.363
Individual characteristics
Fraction leader	−0.024	.296	−0.141	.405	−0.590	.370	−.227	.288
Top 5	−0.471*	.237	−0.506	.324	0.283	.301	.279	.233
List position	−0.005	.008	0.000	.011	0.017†	.010	.006	.008
Sex (1 = female)	0.203	.164	0.294	.224	−0.083	.206	.053	.160
Ethnicity (1 = ethnic minority)	0.041	.251	0.027	.343	−0.108	.314	−.172	.244
Controls
Age	−0.003	.010	−0.001	.013	0.006	.012	.005	.009
Number of (lay citizen) friends (log)					0.964***	.108	.553***	.061
Constant	2.651***	.579	3.080*	.791	−0.272	.779	−0.787	.594
N	150		150		150		150
R ²	.237		.179		.438		.400

*p < .05. **p < .01. ***p < .001. †p < .10.

Table 5.

Explanatory OLS Regression Models of @-Mentions.

	Number of Incoming @-Mentioning Accounts		Number of Incoming @-Mentioning Lay Citizens		Number of @-Mentioning Accounts @-Mentioned		Number of @-Mentioning Accounts of Lay Citizens @-Mentioned
Variable	B	SE	B	SE	B	SE	B	SE
Parties
VVD—coalition	Ref.		Ref.		Ref.		Ref.
PvdA—coalition	0.463	.353	0.460	.347	−0.188	.203	−0.042	0.215
D66—postmaterialist	1.114*	.548	1.104*	.539	0.295	.316	0.045	0.335
GL—postmaterialist	1.192	.826	1.209	.812	−0.133	.472	−.135	.499
PVV—populist	1.501**	.570	1.707**	.561	−0.853*	.349	−.951*	.374
SP—populist	0.741	.509	0.760	.501	0.086	.291	−.049	.309
CDA	0.478	.522	0.436	.513	0.226	.297	−.061	.314
CU	0.598	.761	0.729	.748	0.723†	.433	.729	.459
Other party	0.157	.589	0.241	.580	0.200	.334	−.008	.354
Individual characteristics
Fraction leader	1.585**	.471	1.654**	.463	−1.455***	.278	−1.510***	.295
Top 5	0.075	.384	0.062	.378	−0.046	.219	−0.084	.231
List position	−0.028*	.013	−0.033*	.013	0.016*	.008	0.004	.008
Sex (1 = female)	−0.514†	.262	−0.539*	.257	0.274†	.151	0.166	.161
Ethnicity (1 = ethnic minority)	0.252	.398	0.351	.392	−0.496*	.229	−0.506*	.243
Controls
Age	−0.021	.015	−0.019	.015	0.013	.009	0.014	.009
Number of incoming @mentions (from lay citizens; log)					0.802***	.055	0.652***	.058
Constant	4.669***	.927	4.315***	.911	−1.925	.591	−1.425*	.619
N	146		146		141		150
R ²	.362		.410		.669		.569

*p < .05. **p < .01. ***p < .001. †p < .10.

Turning to the differences between parties, Table 3 shows that MPs from both government parties (PvdA and VVD) have a number of friends and retweet behavior similar to other parties; however, they are @-mentioned less. How to explain this? One might suggest this is related to them being government coalition parties. However, as Table 5 shows, the lower averages are in fact an artifact of these two parties being biggest and having a high number of less known MPs (i.e., these parties have more MPs and therefore more MPs with higher list positions; the party differences in the regression models are controlled for this).³⁰

Overall, in Table 3, the MPs of the populist radical right PVV stand out: They befriend least people but get @-mentioned most. Also, the two postmaterialist parties, D66 and GroenLinks, seem somewhat more connected and engaging on Twitter than the two biggest parties, but it remains to be seen if these fairly limited differences hold in regression analyses.

Tables 4 and 5 zoom in on connections and engagement with lay citizens and include both party and individual characteristics as possible explanations. Regarding party differences, they broadly confirm the patterns discussed above. After controlling for list position (and therewith that bigger parties have more MPs in parliament and more backbenchers who have the highest list position scores), the biggest parties do not significantly differ from the smaller parties, except the MPs from populist and the some extent the postmaterialist parties. The populist MPs follow fewer accounts and (consequently) fewer accounts of lay citizens. However, the ones that they do follow are more likely to be retweeted.³¹ Further corroboration for this, particularly for the PVV, comes from Table 5, showing that populist MPs get @-mentioned more often by lay citizens (and other accounts) but reply to fewer of these @-mentions.

The differences between the two postmaterialist parties and the biggest parties are a bit more nuanced in the multivariate analysis. Most importantly, D66 MPs retweet more accounts (Model 4.3) but not of lay citizens (Model 4.4). Table 5 adds that postmaterialist MPs get @-mentioned by more accounts and by lay citizens than the MPs of other parties (except PVV; Models 5.1–5.2) and consequently @-mention back more, though not disproportionally so (Models 5.3–5.4).³² This may reflect that both postmaterialist parties in the Netherlands still have more Twitter-savvy electorates than the biggest parties, while the biggest parties nowadays are on par with them regarding engagement ratios (Gibson & McAllister, 2014; Jacobs & Spierings, 2016).

Regarding the individual-level characteristics, MPs’ position within the party hierarchy seems most important. As expected, MPs who are leaders of their parliamentary faction are @-mentioned by lay citizens most (Model 5.2), but relatively speaking, they are less likely to return @-mentions. They also seem to retweet friends relatively less, but the rather substantial coefficient is only just below conventional levels of significance. MPs with higher list positions show the reverse pattern: They are @-mentioned by fewer account and lay citizens (Models 5.1–5.2), but those that do @-mention them are more likely to be @-mentioned by the MPs too. A similar (though weaker) pattern was found for retweeting friends (Model 4.3). However, these relative engaging responses were not found when the @-mention or friend involved a lay citizen (Models 4.4 and 5.4).

Finally, regarding underrepresented groups, we find no clear-cut effects regarding friends and retweets, but it seems that, ceteris paribus, female MPs get @-mentioned by fewer lay citizens. Ethnic-minority MPs get @-mentioned by more accounts,³³ but they significantly engage less with those @-mentioning accounts. Some literature suggests they prefer face-to-face contact and the more personal Facebook to communicate with their electorate (Jacobs & Spierings, 2016, p. 121). Additionally, the @-mentions here might include substantial numbers of inflammatory tweets, given the recent heated debates on ethnicity in the Netherlands.

Conclusion

Politicians use social media to inform public opinion (Jungherr, 2016), but public opinion on social media also informs politicians. Social mendia are a means to “tap” information flows and “track” and “monitor” the public’s sentiment (Chadwick, 2013, p. 157).

In this study, we examined to what extent MPs established connections and engage on Twitter, particularly with lay citizens. We adopted a Dahlian perspective and examined who has access to politicians. Moreover, we focus on an in-between elections period, when politicians might be less in a broadcasting-style use and be more receptive toward lay citizens. This was hardly the case. MPs mainly connect with elite accounts (other politicians, media, and established organizations), even though citizens do try to draw the attention by @-mentioning MPs (cf. Research Question 1a). These attempts mainly fall on deaf ears: Lay citizens followed by MPs get retweeted less than elite accounts and their @-mentions are less reciprocated too (cf. Research Question 1b). Thus, while Twitter potentially gives access to new voices (“equalization”), the average MPs’ behavior is focused on the usual suspects (“normalization”). If Twitter is an indicator of public opinion to politicians, it is still heavily skewed toward what elite accounts say is public opinion—not “the public” itself.

We also looked at which politicians were more likely to connect and engage with lay citizens (cf. Research Questions 2a and 2b). Mainly, populists stood out in a negative way. They connected and engaged less with lay citizens, though they did retweet more of their limited number of friends. They appear to contribute to creating echo chambers rather than being interesting in broader public opinion and the public at large. Also, MPs of postmaterialist parties appear to connect and engage somewhat more than other parties. Within parties, backbenchers stood out: The MPs with higher list positions engaged with relatively more of the accounts that they connected with but not with those of lay citizens.

How do these results relate to existing studies in the field? As mentioned in the introduction, most studies focus on interaction, so we focus on that part of our study here. Like in our study, they find that MPs mainly, though not exclusively, use Twitter as a “broadcasting tool” (Ahmed, Jaidka, & Cho, 2016, p. 1079; Arnaboldi, Passarella, Conti, & Dunbar, 2017, p. 243; Frame & Brachotte, 2015, p. 282; see Jungherr, 2016 for an overview). Furthermore, these studies have also found that there is still some interaction taking place (Frame & Brachotte, 2015; Tromble, 2016a), especially in the early days of Twitter (Graham, Jackson, & Broersma, 2016, p. 774) and especially when the incoming @-mentions are positive and when the MP’s party is not populist (Tromble, 2016a). Regarding the latter, we also show that the lower degree of interaction by MPs of populist parties still holds now Twitter is far more widespread then several elections ago (cf. Tromble, 2016a). Like in our study, these researchers also note that most interaction takes place between MPs and accounts that do not belong to lay citizens (Ahmed et al., 2016; Frame & Brachotte, 2015; Tromble, 2016b; but see: Graham, Jackson & Broersma, 2016, p. 775).³⁴ Our study thus confirms and updates existing insights and additionally highlights that the position an MP has within a party seems to matter as well: The more an MP is a backbencher, the more she is likely to interaction with accounts of lay citizens. Moreover, we showed that MPs of postmaterialist parties are contacted more, leading to more interaction. In short, beyond our results on connection and engagement with different groups of users, our results regarding interaction are in line with and complementary to those found in other studies.

While this lends credence to our overall findings, these are certainly not definitive. At least three venues may be promising for future research. First, other approaches might further deepen our findings. Given our Dahlian perspective, we focused on who was connected to and engaged with. Future content analyses might help to show what types of messages are retweeted by MPs and which ones they respond to. Similarly, in-depth interviews with MPs could yield useful insights in how they see and deal with Twitter in-between elections. Second, future studies should check to what extent our findings hold in other political systems. In the Dutch proportional electoral system, MPs do not perform constituency services. In majoritarian political systems, direct contact is especially important (Kreppel, 2011, p. 125). Consequently, MPs might connect and engage with their district’s lay citizens more (cf. Lassen & Brown, 2010; Tromble, 2016a). Third, the patterns we revealed for populist MPs deserve more attention, particularly in times of a contagious populism zeitgeist (Rooduijn, De Lange, & Van Der Brug, 2014). Why do they not connect to “the” people but to “their” people? And what are the differences between left and radical right populist MPs? Each of these venues for future research might help us understand why politicians let pass the opportunities Twitter offers to connect and engage with the direct source of public opinion: the public.

Footnotes

Appendix

Table A2.

Type of Accounts Befriended per Party.

Friends of MP		VVD	PvdA	PVV	SP	CDA	D66	CU	GL	Other
%MPs	7	7	7	4	8	5	5	10	5	6
%Non-MP political	23	22	21	14	23	28	22	24	33	17
%Government	2	3	3	1	2	2	2	1	2	2
%News	18	16	18	18	26	16	21	21	16	18
%Civil society and business interests	11	11	12	8	12	13	13	12	12	9
%Science and research	2	1	2	2	2	1	2	2	1	2
%Entertainment and commerce	0.5	0.5	0.5	0.5	0.5	0.5	1	0	0.5	0.5
%Unclassified accounts	37	39	37	52	27	34	33	30	31	46
Retweeted friends		VVD	PvdA	PVV	SP	CDA	D66	CU	GL	Other
%MPs	16	21	12	12	18	14	19	13	8	10
%Non-MP political	24	24	21	25	25	34	28	34	27	23
%Government	3	4	3	0	2	6	2	1	3	1
%News	23	20	26	29	23	21	24	19	17	31
%Civil society and business interests	12	9	15	7	17	14	12	18	21	9
%Science and research	1	1	2	1	3	1	2	2	2	2
%Entertainment and commerce	0.5	0.5	0.5	0.5	0	0	0.5	0	1	1
%Unclassified accounts	19	21	27	27	13	17	13	13	22	23
@-mentioned		VVD	PvdA	PVV	SP	CDA	D66	CU	GL	Other
%MPs	0.5	3	1	0	0.5	1	1	0.5	1	0.5
%Non-MP political	8	16	12	3	11	9	11	9	11	7
%Government	0	0.5	0.5	0	0	0.5	0	0.5	0	0
%News	3	6	5	2	4	3	4	3	3	4
%Civil society and business interests	3	7	8	1	5	3	4	3	3	3
%Science and research	0.5	0.5	1	0	1	0.5	0.5	0.5	0.5	0.5
%Entertainment and commerce	0	0	0	0	0	0	0	0	0	0
%Unclassified accounts	84	68	72	94	79	83	80	84	81	85
Reciprocal @-mention		VVD	PvdA	PVV	SP	CDA	D66	CU	GL	Other
%MPs	4	8	3	2	3	6	5	4	6	1
%Non-MP political	20	19	20	21	20	17	24	13	22	18
%Government	0.5	1	0.5	0	0.5	0.5	0	0	1	1
%News	10	13	7	14	8	15	10	12	3	15
%Civil society and business interests	10	11	13	4	12	8	12	6	10	11
%Science and research	2	0.5	2	2	3	2	1	1	2	2
%Entertainment and commerce	0	0	0	0	0	0	0	0	0	0
%Unclassified accounts	54	49	55	57	55	53	49	65	58	53

Authors’ Note

As described in more detail in Notes 11, 23, and 24, the Twitter data, additional information on MPs, hand-coded samples, and the account lists per category as well as the code used for automatically identifying other accounts into these categories can be obtained from the authors.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Ahmed

Jaidka

Cho

(2016). The 2014 Indian elections on Twitter: A comparison of campaign strategies of political parties. Telematics and Informatics, 33, 1071–1087.

Arnaboldi

Passarella

Conti

Dunbar

(2017). Structure of ego-alter relationships of politicians in Twitter. Journal of Computer-Mediated Communication, 22, 231–247.

Bailey

Steeves

Burkell

Regan

(2013). Negotiating with gender stereotypes on social networking sites: From “bicycle face” to Facebook. Journal of Communication Inquiry, 37, 91–112.

Berger

(2016). Twittergids. Retrieved September 5, 2016, from http://twittergids.nl

Bimber

(2014). Digital media in the Obama campaigns of 2008 and 2012: Adaptation to the personalized political communication environment. Journal of Information Technology & Politics, 11, 130–150.

Bleijerveld

(2013). Personal interview.

boyd

Golder

Lotan

(2010, 1). Tweet, tweet, retweet: Conversational aspects of retweeting on twitter. 43rd Hawaii International Conference on System Sciences (HICSS), IEEE, Kauai, HI.

Chadwick

(2013). The hybrid media system: Politics and power. Oxford University Press.

Dahl

(1961). Who govern. New Haven: Yale.

10.

Dijkstra

(2013). Telephone interview.

11.

Frame

Brachotte

(2015). Le tweet stratégique: Use of Twitter as a PR tool by French politicians. Public Relations Review, 41, 278–287.

12.

Freelon

Karpf

(2015). Of big birds and bayonets: hybrid Twitter interactivity in the 2012 Presidential debates. Information, Communication & Society, 18, 390–406.

13.

Gibson

McAllister

(2015). Normalising or equalising party competition? Assessing the impact of the web on election campaigning. Political Studies, 63, 529–547.

14.

Graham

Broersma

Hazelhoff

van’t Haar

(2013). Between broadcasting political messages and interacting with voters: The use of Twitter during the 2010 UK general election campaign. Information, Communication & Society, 16, 692–716.

15.

Graham

Jackson

Broersma

(2016). New platform, old habits? Candidates’ use of Twitter during the 2010 British and Dutch general election campaigns. New Media & Society, 18, 765–783.

16.

Hosch-Dayican

Amrit

Aarts

Dassen

(2014). How Do Online Citizens Persuade Fellow Voter? Using Twitter during the 2012 Dutch parliamentary election campaign. Social Science Computer Review, 34, 135–152.

17.

Jackson

Lilleker

(2009). Building an architecture of participation? Political parties and Web 2.0 in Britain. Journal of Information Technology & Politics, 6, 232–250.

18.

Jacobs

Spierings

(2016). Social Media, Parties, and Political Inequalities. New York: Palgrave MacMillan.

19.

Jungherr

(2016). Twitter use in election campaigns: A systematic literature review. Journal of Information Technology & Politics, 13, 72–91.

20.

Jungherr

Schoen

Jürgens

(2016). The mediation of politics through Twitter: An analysis of messages posted during the campaign for the German federal election 2013. Journal of Computer? Mediated Communication, 21, 50–68.

21.

Jürgens

Jungherr

(2016). A Tutorial for Using Twitter Data in the Social Sciences: Data Collection, Preparation, and Analysis. Retrieved from September 15, 2016, https://www.researchgate.net/profile/Andreas_Jungherr/publication/289524980_A_Tutorial_for_Using_Twitter_Data_in_the_Social_Sciences_Data_Collection_Preparation_and_Analysis/links/568e6bbd08aeaa1481b018cb.pdf

22.

Katz

Barris

Jain

(2013). The social media president: Barack Obama and the politics of digital engagement. New York: Palgrave MacMillan.

23.

Klinger

Svensson

(2014). The emergence of network media logic in political communication: A theoretical approach. New Media & Society, 1461444814522952.

24.

Kreiss

(2016). Seizing the moment: The presidential campaigns’ use of Twitter during the 2012 electoral cycle. New Media & Society, 18, 1473–1490.

25.

Kreppel

(2011). Legislatures. In Caramani

(Ed.), Comparative politics (2nd ed.). Oxford, England: Oxford University Press.

26.

Kruikemeier

(2014). Getting connected: The effects of online political communication on citizens’ political involvement (Unpublished dissertation). Retrieved September 25, 2016, from http://dare.uva.nl/record/477409

27.

Larsson

(2015). The EU Parliament on Twitter—Assessing the permanent online practices of parliamentarians. Journal of Information Technology & Politics, 12, 149–166.

28.

Lassen

Brown

(2010). Twitter. The electoral connection? Social Science Computer Review, 29, 419–436.

29.

Mejova

Weber

(2014). Quantifying politics using online data: Introduction. Social Science Computer Review, 32, 131.

30.

Metaxas

Mustafaraj

Wong

Zeng

O’Keefe

Finn

(2015, 4). What do retweets indicate? Results from user survey and meta-review of research. Proceedings of the Ninth International AAAI Conference on Web and Social Media (pp.658–661).

31.

Obama

(2013). Remarks by the president presenting new management agenda. Retrieved September 29, 2016, from https://www.whitehouse.gov/the-press-office/2013/07/08/remarks-president-presenting-new-management-agenda

32.

Oosterveer

(2013). Social media in Nederland 2013: Groei van gebruik Twitter en Facebook afgevlakt. Retrieved September 15, 2016, from http://www.marketingfacts.nl/berichten/social-media-in-nederland-twitter-en-facebook-het-meest-actief-gebruikt

33.

Rooduijn

De Lange

Van Der Brug

(2014). A populist Zeitgeist? Programmatic contagion by populist parties in Western Europe. Party Politics, 20, 563–575.

34.

Schumacher

Rooduijn

(2013). Sympathy for the ‘devil’? Voting for populists in the 2006 and 2010 Dutch general elections. Electoral Studies, 32, 124–133.

35.

Spierings

Jacobs

(2014). Getting personal? The impact of social media on preferential voting. Political Behavior, 36, 215–234.

36.

Statista. (2017). Share of individual using Twitter in the Netherlands in 2016, by province. Retrieved February 6, 2018, from https://www.statista.com/statistics/579080/twitter-penetration-in-the-netherlands-by-province/

37.

Strom

(1990). A behavioral theory of competitive political parties. American Journal of Political Science, 34, 565–598.

38.

Telecompaper. (2017). Jongeren verwijderen Twitter steeds vaker van smartphone. Retrieved February 6, 2018, from https://www.telecompaper.com/nieuws/jongeren-verwijderen-twitter-steeds-vaker-van-smartphone-2--1170420

39.

Tromble

(2016a). Thanks for (actually) responding! How citizen demand shapes politicians’ interactive practices on Twitter. New Media & Society. doi:10.1177/1461444816669158

40.

Tromble

(2016b). The great leveler? Comparing politician-citizen Twitter engagement in three Western democracies. European Political Science. doi.org/10.1057/s41304-016-0022-6.

41.

Urbinati

Warren

(2008). The concept of representation in contemporary democratic theory. Annual Review of Political Science, 11, 387–412.

42.

Vaccari

Valeriani

(2015). Follow the leader! Direct and indirect flows of political communication during the 2013 general election campaign. New Media & Society, 17, 1025–1042.

43.

Van Aelst

Vliegenthart

(2014). Studying the tango: An analysis of parliamentary questions and press coverage in the Netherlands. Journalism Studies, 15, 392–410.

44.

Vergeer

Hermans

(2013). Campaigning on Twitter: Microblogging and online social networking as campaign tools in the 2010 general elections in the Netherlands. Journal of Computer-Mediated Communication, 18, 399–419.

45.

Weber Shandwick. (2014). Twitter en de Tweede Kamer. Den Haag, the Netherlands: Weber Shandwick NL. Retrieved September 15, 2016, from http://webershandwick.nl/wp-content/uploads/2015/01/WS-TWeTWK-def-2-2.pdf

Keeping an Eye on the People: Who Has Access to MPs on Twitter?

Abstract

Keywords

Introduction

Theoretical Framework

Using Twitter as an indicator of public opinion

Vertical connections and engaging with the public

Differences between MPs

Method and Data 11

Operationalizing the Concepts

Twitter Data

Politicians, Parties, Pundits, and People

Differences Between MPs

Findings

The Extent to Which MPs Connect and Engage

Type of Account MPs Connect and Engage with

Differences between MPs

Conclusion

Footnotes

Appendix

Authors’ Note

Declaration of Conflicting Interests

Funding

Notes

References

Method and Data¹¹