How to Convey Trustworthiness via Social Media: Content Analysis and Citizen Testing of British MPs’ Facebook Posts

Abstract

The low and declining level of political trust in the UK is well documented but less is known about how citizens form trust judgements and how politicians can influence perceptions of trustworthiness through social media. This article addresses these important questions using a novel combination of content analysis and survey research, operationalising academic theory on political trust within the context of Facebook communication through a new lens. The findings offer a unique insight into how trustworthiness is conveyed on social media and the types of content that display competence, integrity and authenticity. UK citizens’ perceptions of MP communication are tested in an original survey using embedded Facebook content, allowing key features of high- and low-trust posts to be identified.

Declining political trust and growing disillusionment with politics are common trends across many parts of the world including the USA, Canada, Japan and much of Europe (Pharr and Putnam, 2000). In the UK, distrust is becoming ever more intense (Clarke et al., 2018). Two-thirds of the British public believe MPs and ministers are too easily influenced by the rich and powerful. Around half (49%) disagree that MPs tell the truth and 43% believe that standards of conduct are worse today than 5–10 years ago (The Committee on Standards in Public Life, 2021).

Trust matters. In the context of the global pandemic and climate crisis, governments have asked citizens to make significant changes to their daily lives for the public good (Weinberg, 2022). But without moral authority, it is harder for politicians to persuade people to take part in socially beneficial activities (Allen and Birch, 2015) and comply with policy measures (Levi and Stoker, 2000; Devine et al., 2021; Weinberg, 2022). While a certain level of distrust may be helpful to keep citizens vigilant (Whiteley et al., 2016), in the absence of trust, governments may be constrained in tackling long-term issues (Flinders, 2012; Stoker and Evans, 2019) and citizens may turn to populism (Clarke et al., 2018). It is a gloomy picture.

While extensive research has documented the rise of distrust (Hay, 2007), less is known about the way in which individuals form trust judgements in relation to politics and what might be done to improve the situation. Although there may be valid reasons for scepticism among the public, not least recent scandals such as ‘Partygate’, this does not account for the long-term nature of the trend. Very few citizens ever directly experience political misconduct (Allen and Birch, 2015) and most MPs are never implicated in wrongdoing, yet most people still regard politicians as unethical and dishonest—how are these impressions formed?

Changing modes of political interaction make it increasingly unlikely that citizens will update judgements through meeting MPs face to face. In contrast to the ‘spirit of the hustings’ (Clarke et al., 2018) of previous centuries, the internet now plays a critical role in politics (Page and Duffy, 2018). While some blame online channels for reinforcing partisan views through the ‘echo chamber’ effect (Flinders, 2012) others suggest that social media could have a positive impact on trust (Painter, 2015).

Citizens increasingly use social media to find political information or interact with politicians. 50% of social media users in the US have used social networking sites for civic or political activities and in Europe, 37% of users follow politicians or parties on social media (Starke et al., 2020). Social media enables politicians to communicate directly with voters without the ‘filter’ of party or traditional media, giving them more control over how they present themselves (Giger et al., 2021). Politicians’ social media interactions may be perceived as more ‘authentic’ than those on TV and radio, even when they contain the same content (Lee et al., 2020). The potential for two-way communication on social media can contribute to a sense of ‘imagined intimacy’ with politicians (Lee, 2013 in Starke et al., 2020) a key element in judgements on authenticity (Luebke, 2021).

This article explores how Facebook communication contributes to citizen judgements of MP (un)trustworthiness. Drawing on original research combining qualitative and quantitative methods, it proceeds in four stages. First, I review the literature and identify three different dimensions of trustworthiness on which the subsequent analysis is based: competence, integrity and authenticity. Second, I analyse the content of 817 posts published on Facebook by British MPs during February and March 2022 to assess the extent to which these attributes are present. I find that more than 60% display at least one trust attribute, with competence appearing the most frequently. Drawing on this analysis, typical features of high- and low-trust content are identified.

Third, I test 10 exemplar Facebook posts on a sample of 458 members of the public and demonstrate that—despite a low-trust political environment—MP Facebook posts can still elicit positive evaluations of trustworthiness, especially when competence and integrity are displayed. Finally, I conclude with recommendations for how MPs might use Facebook as part of efforts to rebuild political trust, as well as suggesting avenues for further study.

1. Background

1.1 Conceptualising political trust

At the general level, political trust could be described as a global judgement (Hooghe, 2011) about performance and the extent to which ‘government is producing outcomes consistent with [people’s] expectations … a pragmatic running tally of how people think the government is doing’ (Hetherington, 2005). The key underlying factors which inform this judgement are still the subject of debate.

Interpersonal trust within society (also called ‘social capital’) has been identified as an important antecedent to political trust but not sufficient alone (Newton, 2001; Warren, 2018). The marked divergence between citizens’ improving quality of life ratings and diminishing levels of trust (Kaase, 1999) suggests only a weak link between objective policy performance and trust. Similarly, de Blok et al. (2022) demonstrate that the effects of policy area satisfaction on trust are likely overexaggerated and are contingent on citizen understanding of areas of government responsibility in a complex multi-level system. While the media’s role in (negative) reporting on government performance has been highlighted as a key factor by some authors (e.g. Flinders, 2012), Newton (2017) argues the influence of the mass media has also been overemphasised and it more likely that citizens’ media choices serve to reinforce existing beliefs.

So what are such beliefs based on? While partisanship may have previously provided a ‘warrant’ for trust in the form of shared norms (Warren, 2018), in the context of the declining role of political parties, citizens may use the conduct of individual politicians—especially leaders—as a heuristic for judging the overall honesty of the government and its institutions in general (Whiteley et al., 2016; Starke et al., 2020; Valgarðsson et al., 2021). Judging individual politicians (rather than abstract institutions) may be easier for citizens as they are similar to the informal evaluations of interpersonal trustworthiness performed in everyday life (Starke et al., 2020).

There seems to be broad agreement (Levi and Stoker, 2000; Hay, 2007; Whiteley et al., 2016) that global judgements of trustworthiness consist of at least two underlying dimensions: competence (or ability/performance) and honesty (or integrity/promise-keeping). These two different elements parallel Hibbing and Theiss-Morse’s (2002) distinction between policy and process space (Valgarðsson et al., 2021).

Competence is about the politician’s ability to do the job and is related to experience, training and being well-informed (Druckman et al., 2004) as well as persistence, hard work, qualifications and achievement (Cwalina and Falkowski, 2016). Perceptions of competence can be enhanced by the politician’s visual setting (e.g. the White House) or by highlighting issues about which they have personal expertise (Druckman et al., 2004). Text-only posts may also convey competence as words may signal greater power than images (Amit et al., 2022).

Whereas competence may be contextual, integrity is a more fundamental moral trait that would be expected to remain constant across different situations. Its defining features include ‘frankness, helpfulness, reliability or honesty’ (Cwalina and Falkowski, 2016). Several studies have shown that honesty and related traits (‘trustworthy’; ‘means what they say’; ‘genuine’) are the most important criteria among voters for judging politicians, even more so than competence (Allen and Birch, 2015; Clarke et al., 2018).

Recent research suggests that a third dimension, authenticity—related to being ‘normal’ and ‘in touch’ with the realities of ordinary people’s lives (Clarke et al., 2018)—may also be important (Valgarðsson et al., 2021). A common criticism of MPs is that they live in ‘a bubble’ and are unable to understand the needs of their constituents (Coleman and Moss, 2022). Authenticity is closely linked to benevolence, which the psychological literature identifies as the third attribute of trustworthiness (Mayer et al., 1995), in that being in touch with ‘real’ people is required in order to act in their best interests. However, it differs in that it also relates to being true to oneself (Enli and Rosenberg, 2018). Enli (2017) identifies authenticity markers as elements of communication which display a kind of ‘backstage or passionate side of the candidate’. Use of social media is often associated with authenticity (Luebke, 2021).

This study seeks to understand how British politicians are using social media to communicate with citizens; how the qualities of competence, integrity and authenticity are displayed (or not) in such communications; and how this, in turn, influences trust judgements. The effects of other factors that previous studies have found to be related to trust, including political interest, partisanship and demographic variables, are also considered (Johns and Shephard, 2007; Cwalina and Falkowski, 2016; Jennings et al., 2016; Whiteley et al., 2016; Collignon and Sajuria, 2018).

1.2 Research design

This research focuses on Facebook as it is the world’s most popular social media platform, with 51.15 million users in the UK (July 2021). Facebook is a key channel for political advertising: 3.31 million euros were spent on targeted political adverts in the UK in 2019 (Statista, 2019). Almost all British MPs have a verified profile on Facebook, with 27,000 followers on average. In total, UK politicians publish around 200 Facebook posts every day, generating an average of 58,000 online interactions daily.

Based on the literature, we would expect that variation in the content and style of the posts would lead to differing evaluations of politicians’ competence, integrity and authenticity (Druckman et al., 2004; Enli, 2017; Page and Duffy, 2018) and it is hypothesised that this, in turn, will lead to differing judgements of trustworthiness. Previous research suggests that MPs may particularly use social media to convey the attribute of authenticity (Luebke, 2021).

It is expected that overall levels of trust in government and politicians will be low, particularly among supporters of the opposition (Whiteley et al., 2016) but that individual MPs who display the trust attributes will be perceived as more trustworthy than the government in general (Starke et al., 2020).

There are two elements to the research. In the first study, content analysis is used to categorise 817 Facebook posts published by British MPs during the first half of 2022. This reveals the most common styles of posts, and predictions are made about the way that different content may relate to trust judgements. These predictions are tested in a second study, in which 10 exemplar posts are presented to a sample of 458 British adults within the format of an online questionnaire fielded via Prolific. Online surveys have been successfully used in other studies related to social media and political trust (Enli and Rosenberg, 2018; Lee et al., 2020; Giger et al., 2021).

It is important to note that the study does not seek to make judgements on the extent to which individual politicians are actually trustworthy or not. The research is concerned with citizen perceptions of trustworthiness based on exposure to individual Facebook posts which may not necessarily be representative of that MP’s communication in general. It is also recognised that trust is a cumulative effect and exposure to an individual message is one of many inputs into a running tally of trust evaluations.

2. Study 1: How do British MPs use Facebook to communicate with citizens?

In Study 1, content analysis is carried out on Facebook posts from the cohort of MPs elected in 2017. This cohort is chosen as it provides a manageable-sized group of MPs who are operating under similar circumstances (Allen, 2013) and have some years’ experience but are not yet in high-profile positions.¹ The fact the MPs are not well-known outside politics is an advantage for the second study, as it means participants are less likely to have pre-conceived ideas about them. The cohort is made up of 60 MPs of which 24 are Conservative, 28 Labour, 5 Liberal Democrat and 3 others.

2.1 Method

Of the 60 MPs in the cohort, one did not have a Facebook page and four had set up private—rather than public—profiles. These were excluded from the data collection. All posts published by the remaining 55 MPs between 14 and 20 February 2022 (when Parliament was in recess) and 21 and 27 March 2022 (when Parliament was in session) were downloaded.

This produced a sample of 817 posts in total. Content was downloaded through CrowdTangle to ensure that only publicly available content was accessed, in line with Facebook privacy policies.

2.1.1 Coding scheme

Each post was coded four times: firstly, by category of content and then for the different dimensions of trust. The content categories were developed inductively in a previous pilot study, and continued to be refined throughout the process of coding to ensure they were ‘exhaustive, mutually exclusive and relevant’ (Benoit, 2010). This resulted in 18 different categories. A total of 95% of posts were coded to a descriptive category with 5% coded as ‘other’ (see Table 1). Each post was also coded for whether it was positive, neutral or negative in relation to the three trust attributes of competence, integrity and authenticity. All coding was carried out by the author.

Table 1.

Coding scheme and percentage of posts by category of content

Category	Description	Example	N	%
Political position	Stating a political position	Making a comment on the Spring Statement and why it is welcome/unfair	212	26%
Useful info	Providing useful information	Sharing emergency contact details for people affected by Storm Eunice	110	13%
Local	Visiting a local place, business or organisation	Visiting local attractions during #EnglishTourismWeek; joining a police walkaround	80	10%
Champion	Championing a local issue	Campaigning against closure of a local A&E	63	8%
Supporting campaign	Supporting a national campaign	Attending an event for #downsyndromeday	60	7%
Parliament	Speaking in Parliament (including committees)	YouTube video of the MP speaking in a Parliamentary debate about dental services	59	7%
Other	None of the above	Does not fit any other category	39	5%
Local news	Sharing local news	Sharing a news report that a local school has been rated ‘Good’ by Ofsted	35	4%
Party activity	Taking part in local party activities	Photo of MP going door knocking with local activists	32	4%
Seeking views	Asking for views/feedback	Advertising details of a drop-in session or Facebook Live; launching a survey	23	3%
Other role	Activities relating to a non-constituency political role	Visiting the Polish border in the role of Labour spokesperson for international development	21	3%
Communicating	Giving an update on their work	This week’s email update has just gone out—read it here	20	2%
Deaths	Commemorating a death or tragic event	Marking the 5th anniversary of the Westminster Bridge attack	15	2%
Listening	Following up on constituents’ concerns	Meeting the local police commander to discuss concerns that residents have raised	14	2%
Personal	Personal information unrelated to professional role as an MP	Taking part in the Walsall Arboretum Parkrun	12	1%
Good cause	Supporting a charity or good cause	Sharing details of a local event raising funds for Ukraine	11	1%
Seasonal event	Marking a seasonal event	Happy Mothering Sunday to all the wonderful mums in Battersea	10	1%
Challenge	Challenging own party’s position	Disagreeing with proposals from own party about how to tackle drug addiction	1	<1%

2.1.2 Competence

Competence is defined as working hard and having the skills and abilities to do the job (Jarvenpaa & Leidner, 1998 in Porumbescu, 2017; Valgarðsson et al., 2021).

Posts which showed an MP speaking in Parliament, championing local issues or fulfilling a senior role (e.g. Shadow Minister) were perceived to communicate competence, as were those which demonstrated a detailed understanding of a policy issue or provided genuinely useful information.

Posts which criticised government policy without suggesting an alternative, or shared generic information which was not locally relevant, were coded as neutral in relation to competence.

2.1.3 Integrity

Integrity is understood as taking a moral position, acting in the interests of constituents (rather than for personal gain), delivering on commitments, acknowledging the contributions of others and telling the truth (Allen and Birch, 2015).

Posts coded as positive in relation to integrity showed MPs listening to constituents and acting on their behalf; supporting charities; fighting injustice and acknowledging challenges. Admitting mistakes would also have supported this attribute but no examples were found.

2.1.4 Authenticity

Unlike integrity, which relates to upholding socially agreed moral standards, authenticity is about self-consistency—revealing and staying true to one’s ‘real’ self (Enli and Rosenberg, 2018; Stiers et al., 2021 ). Thus a politician could be seen as authentic while making inaccurate or unethical statements (Luebke, 2021), as long as they are consistent (Trump may be seen as an example of this). Authenticity is also about appearing to be in touch with the needs of ‘real’ people (Valgarðsson et al., 2021) and ‘relatable’ to voters (Stiers et al., 2021).

Posts which showed MPs interacting with local people, described personal experiences or revealed personal information were coded as positive in relation to authenticity. The format and style of posts were important for authenticity (Enli, 2017): those with low-production quality gave the impression of greater authenticity, while those which appeared to have been produced by a third party and contained no personal comment were coded as negative.

2.2 Findings

2.2.1 Types of content

MPs most often used Facebook to make political statements: this accounted for a quarter of all posts. Common topics included the Spring Statement; the cost-of-living crisis; sacking of P&O workers; immigration and foreign policy. Other common categories were providing useful information; visiting a local place or organisation; championing a local issue; supporting a campaign and speaking in Parliament (see examples in Table 1). These six categories made up 70% of all content during the two weeks analysed and are consistent with findings that constituency service is a growing part of MPs’ roles (Campbell and Lovenduski, 2015).

Metadata provided by CrowdTangle showed that the most common format for posts was sharing a photo (56% of posts). Around a fifth of posts (22%) shared a link and 19% shared a video. Just 3% were text-only.

2.2.2 Communicating trust attributes

Most politicians came across as trustworthy in their Facebook communication. Almost two-thirds (63%) of posts supported at least one attribute of trust. Only 3% of posts were perceived to have a negative effect on trust. The posts displayed all dimensions of trustworthiness but there were significantly more posts conveying competence (45% of all posts) than authenticity (29%) or integrity (25%).

This is somewhat surprising since social media is often associated with performed authenticity (Luebke, 2021); indeed Starke, Marcinkowski and Wintterlin (2021) suggest that politicians ‘primarily use SNS [social networking sites] to communicate private information to appear more authentic, approachable and likeable’. The findings suggest the same is not true in the UK: within this sample, only 1% of the content consisted of personal information unrelated to the MP’s professional role. Almost all MPs published posts that supported all three attributes which may suggest a conscious strategy to balance different types of content.

2.2.3 Features of high- and low-trust content

Posts categorised as ‘Listening’ scored highly on all three trust attributes because they tended to show an MP taking action (competence), delivering on promises (integrity) and understanding the needs of constituents (authenticity). A small number acknowledged challenges or failures which further supported the trait of integrity (truth-telling).

Almost all posts categorised as ‘Parliament’ communicated competence as they demonstrated that the MP had the skills to do the job. Many also demonstrated integrity (45%) and authenticity (35%)—these were often speeches where the MP was following up on an issue that had been raised by constituents (integrity) and included personal stories (authenticity), for example, the consequences of fuel poverty.

While a third of posts (34%) were coded as neutral in relation to trust, only 3% were coded as detrimental to trust and this was related more to the format and style of communication rather than the type of content—in fact, similar content could suggest low or high trust, depending on its treatment.

Posts about Mother’s Day were a good example of this. Some posts used a personal photo and message that conveyed a strong sense of authenticity while others used a branded graphic and a generic message, perhaps suggesting the MP was paying lip service to the date for political reasons (low integrity).

2.3 Discussion

Overall, the findings from Study 1 suggest that the use of Facebook by British MPs is positive in relation to (re)building political trust. Almost all MPs who were frequent users of Facebook published a range of content that supported all three trust attributes.

Perhaps surprisingly, politicians’ posts are more often related to competence than to integrity or authenticity. It may be that integrity is a more difficult concept to communicate but a higher proportion of personal (authentic) content might have been expected.

To help rebuild trust, MPs might consider posting more content relating to integrity as previous research suggests it may be the most fundamental trust attribute (Allen and Birch, 2015; Clarke et al., 2018). Posts which demonstrate that politicians are listening to constituents and following through on commitments might be a good way to do this, as well as those showing active support for local charities and good causes (this will be tested in Study 2). MPs might also consider a more open style of communication, acknowledging challenges and even occasionally admitting mistakes.

While there was only a very small proportion of posts that prompted negative trust assessments, the format and style of posts were important in this respect; overuse of party branding and overproduction of video content may reduce perceived trustworthiness.

3. Study 2: How do citizens interpret politicians’ Facebook posts in relation to trust?

The second study sought to test whether the coding conducted in Study 1 reflects the way this content is perceived by citizens and to better understand the relationship between individual trust attributes and overall perceptions of trustworthiness. It also aimed to test the effect of other factors on perceptions of trustworthiness, including partisanship and political interest.

The survey was conducted during the period of Liz Truss’s failed economic reforms and the subsequent change of Prime Minister; given this level of turbulence, it was expected that trust in politicians would be close to zero or below.

3.1 Method

The study presented ten Facebook posts from different MPs to 458 participants, who were asked to rate each politician on different attributes of trust. The posts were chosen to represent an even number of Conservative/Labour and male/female MPs and to exemplify the common types of content found in Study 1 (see examples in Figure 1). The posts were embedded in an online survey using Qualtrics and were followed by questions about political trust, political interest and partisanship.

Figure 1.

Posts chosen to exemplify high- and low-trust content. Bim Afolami’s post (left) was rated the highest overall for trustworthiness (1.89) while Tony Lloyd’s was rated lowest (0.03), both on a scale of −5 to + 5

Source: https://www.facebook.com/Bim4HandH/posts/pfbid0BASPta6BGkm8duH1dRu1ev9diXN8xESbyfFBTiQJ6ZhLYZre4wDZx3dToVfYLq2Ll

https://www.facebook.com/Tony4Rochdale/posts/pfbid06vpoxmh3NNyNX8tZRPPfHMTb2rNL3Ko8jN3JEByQwmCMQE7XoHAHZ5NTY3KGGYMvl

3.2 Participants

Participants were recruited via online survey platform Prolific, which has been widely used for academic studies in the social sciences (Weinberg, 2022). They were paid a rate of £1.05 for the 7-minute survey which was completed on two dates in October (n = 148) and November (n = 310) 2022. The sample was balanced for gender (51% female) and included only Labour and Conservative supporters (42% vs 58%). All respondents were residents of Britain (excluding Northern Ireland) and ranged from 20 to 85 in age (M = 46). A total of 91% were White/Caucasian. Demographic information came from a pre-screening questionnaire conducted by Prolific.

3.3 Survey design

The Facebook posts were presented one at a time in a random order. Below each post, participants were asked ‘to what extent do the following words phrases describe MP [name]?’. The attributes ‘Competent’, ‘Acts with integrity’, ‘In touch’, ‘Authentic (real)’ and ‘Trustworthy’ were listed, next to a scale of −5 to + 5, with 0 as the default. A short explanation of each attribute was provided in the introduction to the survey. Participants were also asked if they had previously heard of the MP. In the light of Study 1, authenticity was separated into two dimensions: understanding the needs of ordinary people (‘in touch’) and revealing one’s true self (‘authenticity’). Using the same −5 to + 5 scale, participants were asked to rate their trust in the government and the Prime Minister. They were also asked about their level of interest in politics.

3.4 Expectations

Aside from the overall expectations that average trust ratings would be neutral or negative, it was expected that MPs perceived to be conveying the attributes of competence, integrity and authenticity would be rated as more trustworthy than those who were not (Valgarðsson et al., 2021) and that judgements would be mediated by partisanship (Jennings et al., 2016; Whiteley et al., 2016). It was also expected that participants’ evaluations would broadly reflect the coding carried out by the author in the first study: for example, that posts showing an MP doing their job would rate positively for competence and those showing MPs interacting with local people would score positively for authenticity.

3.4.1 Trust in government, leadership and individual MPs

As expected, general trust in government was low, as was trust in the political leadership. Among Conservative supporters, levels of trust in government improved from an average of zero under Liz Truss in October to + 1 in November under Rishi Sunak. However, there was little change among Labour supporters, whose average level of trust in government remained around −3. This partisan difference was a key feature of the data and is discussed further below.

In contrast, average ratings for trustworthiness of individual MPs, when based on these posts, were predominantly positive and significantly higher than trust in government. This is consistent with previous research which has shown that people tend to evaluate their own local MP, and those with whom they have had direct contact, more positively than MPs in general (Flinders, 2012; Allen and Birch, 2015) and may suggest that Facebook content from MPs has the potential to play a positive role in rebuilding trust. It may also reflect the nature of the questions which encouraged respondents to look for evidence of positive attributes in the MPs’ communication, rather than relying on previous impressions gained from a range of sources.

3.4.2 Relationship between the trust attributes and perceptions of trustworthiness

There was a strong correlation between the average attribute scores and overall trustworthiness (r (4578) = 0.86, p < 0.001), which may suggest that participants formed a global judgement for each MP which informed all scores. Using a linear regression model to test the effects of each attribute showed that all of them had a strong positive relationship with overall trustworthiness (Table 2). There was no relationship between how well-known MPs were and how trustworthy they were considered.

Table 2.

Linear regression model showing effect of attribute scores on ratings for trustworthiness

Variable	Competence		Integrity		‘In touch’		Authenticity
(Intercept)	−0.107		−0.277	***	−0.054		−0.053
x	0.809	***	0.835	***	0.740	***	0.757	***
Same party	0.119	***	0.093	***	0.137	***	0.065	*
Gender (male)	−0.191	***	−0.142	***	−0.170	***	−0.132	***
Age	0.005	***	0.004	***	0.001		0.003	***
University educated	−0.017		−0.033		−0.113	**	−0.017
R2	0.596		0.660		0.587		0.666
N	4,550		4,550		4,550		4,550

Significance levels are shown as * <0.05, ** <0.005, *** <0.001.

Ratings for the overall trustworthiness of each politician were significantly lower than each of the attribute scores which suggests that the positive evidence provided by (most of) the posts was tempered slightly by the participants’ existing distrust of government. Regressing government trust ratings on trust scores confirms that it has a small but significant effect (R² = 0.08, F(1, 4578) = 75.66, p < 0.001). Nonetheless, all the MPs received positive rather than negative trust scores on average.

There were significant differences between the highest- and lowest-rated posts. Although scores for each individual post were highly correlated, there were also significant differences between attributes in some cases: for example, Douglas Ross’s post about his family was rated 1.95 on average for authenticity and 0.73 for competence. This suggests that respondents were genuinely using the stimulus material and were able to distinguish between the different attributes in clear cases.

3.4.3 Effects of partisanship on trust

For some (but not all) posts, partisanship made a significant difference in attribute ratings and judgements of trustworthiness. Conservative MP Gillian Keegan was rated significantly more trustworthy by Conservative supporters than by Labour supporters (t(393.63) = 2.30, p = 0.02), whereas Labour MPs Marsha de Cordova (t(452.37) = −7.54, p ≤ 0.001), Tan Dhesi (t(434.06) = −3.24, p = 0.001) and Tony Lloyd (t(443.89) = −4.04, p < 0.001) received significantly higher ratings from Labour supporters. A linear regression (Table 2) shows that partisanship had a significant positive effect on all variables, particularly ‘in touch’.

In most cases where opinion was divided along partisan lines, posts used obvious party branding and/or put forward a clear political position: for example, Gillian Keegan gave thanks to social workers ‘on behalf of the Government’ while Tony Lloyd used a red Labour Party graphic.

Interestingly, Conservative MP Eddie Hughes received significantly higher scores for competence from Labour supporters than Conservative supporters, perhaps because he was posting about strengthening rights for renters, who are more likely to be Labour voters, and his profile photo, post images and text were otherwise politically neutral.

3.4.4 Types of content that communicated competence, integrity and authenticity

To test whether each post scored highly on a particular attribute a two-sided t-test was used to compare the post score to the mean score for that attribute. The difference between the attribute mean and the post mean, and the significance level, is shown in Table 3. The table shows that respondents’ evaluations were broadly consistent (65% agreement) with the coding carried out in Study 1.

Table 3.

How posts in the survey were judged in relation to expectations

Significance levels are shown as *<0.05/ **<0.005/ ***<0.001

For example, Bim Afolami’s post (see Figure 1) was chosen as an exemplar of a high-trust post: it showed him meeting constituents (in touch), listening to their views (integrity) and taking action on an important issue (competence), illustrated by clear but amateur-quality photos (authenticity). This was validated by participants, who rated him highly on all trust attributes and overall trustworthiness.

Tony Lloyd’s post, on the other hand, was chosen as an exemplar of a low-trust post. It used a graphic about Mother’s Day which was branded by the Labour Party (low integrity/authenticity), did not provide any evidence of doing the work of an MP or the skills needed to do so (low competence) and did not provide any personal comment (low authenticity). The survey confirmed these expectations, returning below average scores for Lloyd across all attributes—indeed this was the only post to receive one attribute score below zero. While this may seem unsurprising, it is worth noting that the graphic appears to have been produced by the Labour Party itself, so the findings could have implications for political communicators in the way they produce materials and advise politicians.

However, some of the posts did not perform exactly as expected. For example, Vicky Ford’s post which showed her visiting a local building site and speaking to local people rated highly for ‘in touch’ (as predicted) but unexpectedly also scored above average for competence and authenticity. Similarly, Tonia Antoniazzi’s post about supporting a Macmillan petition was chosen as an example of ‘integrity’ and ‘in touch’ but rated above average for all four attributes. This may suggest that the scores form part of a global judgement which is also informed by other factors.

3.5 Other factors affecting trust judgements

A linear regression model was created for each attribute with overall trustworthiness as the dependent variable (Table 2). This showed that as well as partisanship, gender and education had a significant effect on trust evaluations. Respondents identifying as female gave significantly higher average trust scores to MPs, government and political leaders than respondents identifying as male. This supports previous research which has found men tend to be more cynical and more critical of politicians’ technical competence (Jennings et al., 2016; Whiteley et al., 2016), although other studies have found no gender difference in levels of political trust (Clarke et al., 2018; Enli and Rosenberg, 2018; Rose and Wessels, 2019).

University education had a significant negative relationship with the ‘in touch’ variable. Respondents with a higher level of education (degree level or above) also reported significantly lower trust in government in general (M = −1.11) than those who had not been to university (M = −0.84 t(3861) = 2.92, p < 0.01). This is in line with much previous research (Hay, 2007; Jennings et al., 2016) except for Enli and Rosenberg’s study (2018) which found that more educated respondents were more likely to regard politicians as honest.

The data did not show any clear relationship between political trust and other variables such as political interest, descriptive similarity, employment status or income. Age had a negligible effect.

3.6 Discussion

As expected, the survey showed that trust in government and political leadership is currently very low, particularly among non-Conservative voters. However, participants reacted more positively than expected to the posts published by individual politicians, suggesting that MPs can still gain the trust of the electorate if they demonstrate competence, integrity and authenticity. When used effectively, Facebook can enable politicians to do this. While different content and styles of posts may emphasise different attributes (see Table 4) the most highly rated posts conveyed all trust attributes simultaneously.

Table 4.

How to use Facebook as part of efforts to rebuild trust

Attribute	May be communicated by:	Unlikely to be communicated by:
Competence	Meeting constituents to discuss an issue of concern and committing to work with local agencies to tackle it Introducing a White Paper in the House of Commons Working cross-party to tackle an issue of concern Actively supporting a charity campaign Visiting a local site to view safety improvements Stating a political position that the reader supports Using statistics (rather than emotion) to make an argument	Posts that are entirely personal Posts that do not show any evidence of the MP doing their job Reposting generic seasonal content Posts that criticise the party the reader supports (for example, ‘The Tories did nothing … Judge them by their inactions’)
Integrity	Standing up for the rights of vulnerable people Actively supporting a charity campaign Representing the views of constituents Listening and asking for feedback Acting for others	Posts that seem insincere (e.g. saying #ThankYou to social workers on behalf of the government without taking any action) Personal content Reposting generic seasonal content Visiting a local site but not taking action
Authenticity (in touch)	Photos that show MPs alongside ‘real’ people such as builders or community groups Campaigning on issues that are important to ordinary people such as health, housing and employment	Posts that seem insincere Reposting generic seasonal content
Authenticity (true self)	Amateur or ‘behind the scenes’ style of images Taking action on an important issue	Posts that are critical of the party the audience supports Reposting generic seasonal content Talking or learning about an issue without taking action Posts that are too personal

Note these are tentative recommendations based on testing ten posts among a relatively small sample of citizens.

The attribute of authenticity was of particular interest as it has been linked to social media communication and may be particularly important to distrusting citizens (Valgarðsson et al., 2021). Respondents were therefore asked to consider how the MPs displayed two different dimensions of authenticity—being ‘in touch’ and revealing one’s ‘true self’. It seemed that being ‘in touch’ was much easier to convey, with almost all MPs scoring significantly higher for being ‘in touch’ than all the other attributes.

Authenticity, in the sense of ‘true self’ seemed more difficult to communicate and, against expectations, entirely personal posts did not score the highest for authenticity; those which showed MPs taking action on an important issue were regarded as more authentic. Furthermore, authenticity without competence and integrity did not result in high trust ratings, suggesting that personality alone is not enough for voter trust. This is consistent with a recent study based on fictional Twitter posts (Giger et al., 2021) which found that participants were significantly more likely to say they would vote for politicians who communicated policy positions, compared with those who posted only personal content. It is also in line with previous research which has found integrity to be the most fundamental attribute in trust judgements (Allen and Birch, 2015; Cwalina and Falkowski, 2016). It may be that recent events, such as the ‘Partygate’ scandal and Liz Truss’s mishandling of the economy, have led voters to place even greater importance on both integrity and competence over authenticity.

While the survey results suggest that the original coding was reasonably accurate, they also reveal that trustworthiness is not an objective quality: perceptions of Facebook content are strongly mediated by partisanship, especially where there is a clear political position. This could be an example of motivated reasoning: social media consumers may look for information to support their existing beliefs, rather than critically evaluating the evidence (Taber and Lodge, 2006).

4. Conclusions

In this article, I have shown that citizens’ perceptions of competence, integrity and authenticity are closely associated with judgements of trustworthiness and that MPs may be perceived as more trustworthy if they convey these attributes via social media. The findings build on previous research (e.g. Valgarðsson et al., 2021) by operationalising theory on trust in the context of Facebook communication, and offer valuable insights for politicians and policymakers. A summary of practical recommendations for how Facebook could be used as part of efforts to rebuild trust is included in Table 4.

The most common type of Facebook posts published by MPs in the period sampled were those explicitly stating a political position, which is likely to be evaluated very differently depending on the recipient’s partisanship. MPs aiming to communicate primarily with their own local party members will likely find that this type of content is positively received, but it may not help to build trust amongst their wider constituencies. Posts which show MPs visiting a local place (10% of posts) or championing a local issue (8%) may be more likely to convey the trust attributes—in the survey, these posts received significantly higher ratings for trustworthiness than average. Those which showed a politician taking action were rated higher than those where the MP was simply observing (see Table 3).

Posts that demonstrated MPs were listening were uncommon (2%); a greater number of these could increase perceptions of integrity. Very few posts actively invited feedback, suggesting that many politicians are still using social media in broadcast mode—better exploiting its potential for two-way communication could be another way to display integrity. Analysis of Facebook comments and replies would be a worthwhile area for future study.

Personal posts were also uncommon (1%) and this is likely the right strategy by MPs: the survey showed that entirely personal content was not the most effective way to convey authenticity or improve perceptions of trustworthiness—perhaps the British public does not want this type of communication from their politicians.

Further research could examine the Facebook use of more high-profile politicians or test specific variables, such as the style of photography. Qualitative research to better understand the thought process of individuals when making a trust judgement would also be worthwhile. The format of the survey may have encouraged respondents to look for positive evidence of each attribute and it is possible this inflated the trust ratings; future research could test this by including negative as well as positive attributes.

Together the findings from both studies suggest that politicians and political parties seeking to rebuild trust with the electorate could effectively use Facebook as a tool. This is not to imply that MPs can expect to gain the public’s trust without upholding high standards of behaviour, but Facebook communication could help to reassure citizens of such efforts, especially if competence and integrity are conveyed.

Funding

This work was supported by the Economic and Social Research Council.

Conflict of interest

There are no conflicts of interest.

Footnotes

1

The list was cross referenced with the YouGov fame rankings for politicians. Only 12 of the sample were included in the rankings, meaning they were known by more than 12% of people. The most well-known (and only MP in the sample known by more than half of YouGov respondents) was Esther McVey, former Secretary of State for Work and Pensions, known by 54% of people.