Representative Audit of Britain

Representative Audit of Britain June 2018 ES/L016508/1

Acknowledgements We are grateful for the generous support of the Economic & Social Research Council (ESRC) for funding the Representative Audit of Britain (ES/L016508/1). This report was produced by the RAB team led by Professor Rosie Campbell (Principal Investigator, Birkbeck), Dr Jennifer Hudson (co-investigator, UCL), and Dr Wolfgang Rüdig (co-investigator, Strathclyde). The project includes eight researchers from across UK universities: Dr Peter Allen (Bath), Professor Sarah Childs (Birkbeck), Dr Sofia Collignon Delmar (Royal Holloway), Dr Chrysa Lamprinakou (Birkbeck), Professor Joni Lovenduski (Birkbeck), Dr Caitlin Milazzo (Nottingham), Dr Javier Sajuria (Queen Mary) and Professor Maria Sobolewska (Manchester). We thank Agnes Magyar, Evangelina Moisi, Linda Hein, Naveen Saini, Nikki Soo and Lotte Hargrave for their research assistance on the project. We also thank Ruth Fox, Joel Blackwell and Luke Boga Mitchell of the Hansard Society for their support in the project launch, and the RAB advisory board for their expert guidance during the project: Dr Stephen Fisher, Professor Jane Green, Professor Anthony Heath, Professor David Sanders and Professor Hermann Schmitt. Parliamentary Candidates UK 2018 School of Public Policy 29/31 Tavistock Square London WC1H 9QU Email: candidates@ucl.ac.uk Web: parliamentarycandidates.org Twitter: @candidatesuk

Table of contents I. Introduction: The Representative Audit of Britain survey of parliamentary candidates 2 1. Background 2. Response rates 3. Interviews with candidates II: The importance of context: Comparing the 2015 and 2017 general elections..4 1. Selection processes: 2015 and 2017 in comparison..4 A. Selection and costs of standing B. Social background of parliamentary candidates C. Ideology: Candidates, voters, parties D. Representation 2. Campaigning for office..12 3. Harassment and intimidation: Experiences of 2017 parliamentary candidates...15 III. Representative Audit of Britain research..18 1. Do candidates and MPs differ from party officers and members in their political views? 2. Do candidates and voters have similar policy views? Does it matter? 3. Explaining motivation to represent: How does descriptive representation lead to the substantive representation of racial and ethnic minorities? 4. Is the local candidate advantage a myth? Analyzing the effects of localism in the 2015 UK general election 5. Descriptive or symbolic representation: Evidence from an elite/mass analysis IV. RAB research papers and reports.24 1

I. The Representative Audit of Britain: Survey of parliamentary candidates 1. Background The Representative Audit of Britain (RAB) is a comprehensive survey of parliamentary candidates who stood in the 2015 and 2017 British general elections. The RAB examines candidates attitudes, backgrounds and experiences of standing for Parliament. The research is funded by the Economic and Social Research Council (ESRC, ES/L016508/1). The data in this report draws on surveys of parliamentary candidates in the 2015 and 2017 general elections. We present analysis based on the responses of only those candidates standing for the Conservative, Labour, Liberal Democrat, UK Indepence (UKIP), Plaid Cymru, Scottish National or Green parties. 1 The 2015 survey was fielded in two waves. Wave one was first in the field in March 2015, wave two in July 2015, and all data collection was completed by June 2016. Due to the snap election, the 2017 survey was fielded after the election on 8 June, with major fieldwork concluding in February 2018. We are still collating some follow-up surveys and therefore the 2017 figures reported in this report are preliminary. The 2017 survey differs from the 2015 survey by including new items covering several important issues that rose in salience after 2015, in particular, new questions on Brexit and the security and harassment of candidates. We also compare candidates experiences of selection and election under normal circumstances in 2015 and in a snap election in 2017. The RAB continues a times-series of data collected on parliamentary candidates in Britain since 1992, as well as linking to cross-national data by including items from the comparative candidate study (CCS). 2 The 1992 British Candidate Survey was undertaken by Professors Pippa Norris and Joni Lovenduski and continued by the same team as the British Representation Study (BRS) in 1997 and 2001. In 2005 the BRS was run by Professors Joni Lovenduski, Sarah Childs and Rosie Campbell. The 2010 British CCS was conducted by Dr Wolfgang Rüdig. The 2015 and 2017 RAB is led by Professor Rosie Campbell, Dr Jennifer Hudson and Dr Wolfgang Rüdig. 2. Reponse rates In order to maximise response rates, and ensure that the data was of the best quality, we sought the support of the major mainland parties. Most of the parties agreed to provide a cover letter supporting the project and requesting candidates participation in the enclosed survey. We would particularly like to thank Gareth Fox (Head of Candidates) and Jane Maclaren (Head of HR) at the Conservative Party; John Stolliday (Director of Governance and Legal) at the Labour Party; Arfan Bhatti (Head of Diversity, Candidates and Talent Support) for the Liberal Democrats; Shirley-Anne Somerville (Deputy Chief Executive) at the SNP; and Peter Barnett, member of the Green Party Executive (GPEx). To achieve the highest possible response rates, we sent out both paper copies of questionnaires and e-mails offering candidates the option to complete the survey online. Approximately 50% of candidates who completed the survey opted to do so using the paper version, showing the importance of a multi-mode approach in elite surveys. Table 1 shows historical response rates for British candidate studies since 1992. The response rate for 2015 was 57%. The response rate for 2017 is 51%. 3 2

Table 1. UK candidate study response rates, 1992-2017 1992 1997 2001 2005 2010 2015 2017 Study BCS BRS BRS BRS CCS RAB RAB N 1290 999 1051 796 1507 1897 1446 Response rate % 67 61 52 41 52 57 51 Incumbent MPs 234 277 185 81 139 174 169 Conservative 353 295 349 242 234 228 144 Labour 399 333 289 182 321 357 378 Liberal Democrat 319 285 329 319 393 448 408 SNP 58 41 52 27 35 42 21 UKIP 0 0 0 0 265 271 143 Green 144 23 0 0 235 421 323 Plaid Cymru 17 22 32 26 24 31 29 A total of 3,304 candidates stood in the 2017 general election, down from 3,971 in 2015. 4 The Conservatives contested 638 seats (including seven in Northern Ireland), Labour 631 (all seats in Great Britain, save the Speaker s seat in Buckingham), and the Liberal Democrats 629. The SNP contested all 59 seats in Scotland and Plaid Cymru all 40 in Wales. UKIP contested 378 seats, considerably down from the 624 they fought in 2015 or the 558 fought in 2010. The Green Party also contested significantly fewer seats in 2017, with 467 candidates standing, down from 573 in 2015 but still higher than the 335 who stood in 2010. We sent surveys to these candidates through the political parties and to addresses candidates provided on the Statements of Persons Nominated posted on local authorities websites. 3. Interviews with candidates To add to the quantitative data, in 2015 we conducted 44 interviews with candidates (Figure 1). The interviews with candidates proved invaluable for providing more depth to our understanding of the experience of standing for election, and have helped inform our analysis of the quantitative data. For example, a number of respondents to the 2015 survey reported over 50,000 in selection expenses which seemed high, but in our interviews with candidates it became clear that this was an entirely reasonable figure once forgone earnings, travel costs and rentals were included. In total 12 of the interviewees were women, two were from BME communities and 14 were elected to parliament. 5 The interviews have been transcribed and anonymised transcripts will be submitted to the data archive with the dataset. Figure 1. Interviews conducted of 2015 candidates by party 3

II: The importance of context: Comparing the 2015 and 2017 general elections 1. Candidate selection processes: 2015 and 2017 in comparison In terms of candidate selection, the 2015 and 2017 elections were very different beasts. The 2015 election fell as was expected per the Fixed-Term Parliaments Act 2011 and the parties had time to plan their selection processes and follow their usual procedures. The 2017 general election was announced by the Prime Minister Theresa May on the 18 April, leaving just over seven weeks until polling day on 8 June. Hence parties were given precious little time to select candidates and campaign. The two largest parties, the Conservatives and Labour, adapted their rules considerably to meet the challenge. The Conservative Party set special rules agreed by the party board shortcutting the usual selection procedure. New candidates were screened through emergency parliamentary assessment boards (PABs) all conducted within a five-day period; they were comprised of a 45-minute interview with a senior party officer, with a pass or fail decision made on the day. Incumbent Conservative MPs who chose to stand again were required to secure a majority of their local association s executive council or, if not, a majority of the local association s members. A decision was taken not to allow candidates to apply directly to winnable seats (that is, retirement seats where the incumbent was standing down and target seats, which the party hoped to take). Instead, associations were usually offered a choice of three candidates (decided by the central party candidates team in consultation with the officers of the local association) at a general meeting of the local party. In seats that were neither retirements nor targets, a candidate was selected by the chairman of the party and the chairman of the National Conservative Convention, after consultation with officers from the local party. The party adopted a policy of including at least one woman on all retirement and target shortlists, and a BME candidate on shortlists in safe seats. Innovations employed in recent elections such as using open or closed primaries (constituency level elections of all voters or Conservative supporters) and open meetings were not used at all in 2017. The Scottish Conservative Party set their own selection rules and utilised a centralised process in most seats. Officially the Conservatives in Wales were due to follow the same rules as applied to the English seats, but the shortlists in target seats sometimes contained fewer than three names. A meeting of the Labour Party s National Executive Committee (NEC) officers agreed selection procedures on the day the election was announced (18 April). The meeting decided that any sitting MP would automatically be endorsed by the NEC and other selections would be made using the exceptional selections procedure. Sitting MPs were required to confirm whether they wished to stand again by 6pm on 20 April. Applications to stand in the 13 seats where Labour MPs stood down were opened the following day and closed on 23 April at noon. Longlisting was then conducted by an NEC panel made up of the most senior NEC officers. The panel sifted through CVs and listed candidates for interview. The interview panels consisted of three NEC members and a member of staff acting as a secretary to consider thirteen seats or between 78 and 130 candidates. For the late retirement of Steve Rotheram the MP for Liverpool Walton an even more expedited process was undertaken. For all other seats in England not held by Labour, applications opened on 21 April and were publicly advertised; they closed on 23 April at noon. Candidates were required to write to declare the seats they were applying to, with no limit on the number. Interviews were not conducted and no local selection hustings were held. For each English European Parliament region (Wales and Scotland ran their own processes) the party formed a panel consisting of two NEC members and a member of the regional board. A meeting or conference call was conducted for each regional panel. Candidates who stood in 2015 were considered first and the process was the same for all non-held seats. 4

In 2016, the Liberal Democrats contacted all candidates who stood the previous year saying that they would be assumed to be standing again, unless they explicitly stated otherwise. They then undertook an accelerated selection process for other seats based on their by-election procedure. More than 400 candidates were in place by the end of 2016, with the majority selected by September. Hustings were held in only the top sixty seats, and only in contested selections (no other candidates applied to stand against Vince Cable in Twickenham or Ed Davey in Kingston and Surbiton, for example). In the vast majority of other seats, candidates were appointed through an agreement between the regional party and local party chair. Selection was far easier for the SNP; whose 54 sitting MPs were all re-selected. This left just five candidates to choose through the SNP s normal general election selection process. Plaid Cymru stuck to their normal selection process with local party members selecting candidates in all seats across Wales that they did not hold. The Green Party used its usual local selection procedures based on a vote of local groups of party members. In 2017 a small number of parties opted to participate in an alliance to avoid splitting the vote for left-of-centre parties. The Progressive Alliance website listed 41 seats were a candidate stood down to boost the chances of another candidate of the left s election. In practice there were few reciprocal agreements, all between the Liberal Democrats and the Greens or the Greens and the Women s Equality Party. In Brighton Pavilion, the Liberal Democrats did not challenge the Green MP Caroline Lucas, and in turn, the Greens did not contest Brighton Kemptown. They came to a similar arrangement in Skipton and Ripon, which the Liberal Democrats did not contest, and in Harrogate and Knaresborough, where the Green candidate stood down. Within UKIP there was a heated debate about whether to select candidates to stand against pro- Brexit MPs, in the end the party stood in many fewer seats in 2017 than in 2015. 6 In the next four sub-sections, we compare candidates experiences of selection, their social background, and their views on the costs of standing, ideology, and representation for both the 2015 and 2017 elections. A. Selection and costs of standing Table 2 shows the percentage of respondents who indicated their selection was contested. Significantly fewer Labour and Conservative candidates experienced contested selection in 2017 compared to 2015 owing to the snap election The same is true for Liberal Democrats, UKIP and Plaid Cymru; only the Greens had more contested selections in 2017 than 2015 Table 2. Whether candidates selections were contested (%) Con Labour Lib Dem SNP UKIP Green PC 2015 74 86 25 100 46 23 33 2017 37 48 11 100 16 26 23 Question: Turning to your successful selection for the 2015/2017 general election, was your selection contested? N = 1486 (2015) and 1101 (2017); Note: figures based on non-incumbent candidates only Evidence from our 2015 interviews showed that selection procedures varied widely and that running for office can be time consuming, and sometimes, frustrating. My personal experience was that it was very, very slow and very drawn out. I think it should be done earlier because you have got to give the candidate the most opportunity to engage with voters and to campaign. So the earliest, earlier the better in terms of selecting a candidate. It was really frustrating for me to be waiting around. But the flipside of that is and I didn t really appreciate that at the time that the cost and the money sorry, the cost and the time. And the impact that it has on your personal life being a candidate, once you are selected. Actually having to do that over 5

a much longer period would be very difficult. I mean doing it for only five months was a bit of a blessing in some ways. [On decision to run] In 2015, it was that on a practical level, the party need more candidates, and the party didn t have a candidate, and the party really needed a candidate, and could run a campaign but didn t have a candidate. We put out a regionwide call and nobody came forward, and there was a gap, and I decided to go for it. Table 3. Candidate selection procedures in 2015 and 2017 (%) Con Labour Lib SNP UKIP Green PC Dem Postal/online 2015 1 63 45 97 6 88 29 vote of mems 2017 4 2 10 100 9 59 18 Meeting of 2015 96 96 73 54 95 76 92 members 2017 54 14 43-86 92 95 Question: In order to select the candidates to represent the party in the 2015/2017 general election, did your constituency carry out a postal or online ballot of members/hold a meeting of party members to vote on nominations? N = 1356 (2015) and 1035 (2017) In addition to the significant time costs, candidates also acknowledge significant personal financial costs in standing for office. As shown in Table 4, in 2015, 83% of Conservative and 81% of Labour and SNP candidates incurred expenses during selection. The figures for other party candidates (Liberal Democrat 61%, UKIP 74%, and Plaid 68%) are only slightly lower suggesting that even at the earliest stages of the electoral process, candidates are spending personal money to support their selection. Table 4. Selection expenses in 2015 and 2017 (%) Con Labour Lib Dem SNP UKIP Green PC 2015 83 81 61 81 74 41 68 2017 57 18 21-35 19 19 Question: Did you incur any expenses seeking selection as a candidate in the 2015/2017 general election? N = 1434 (2015) and 1104 (2017) Drawing on data for the 2015 general election, Table 5 shows that the costs of standing are not insignificant with an average spend across all candidates who had selection expenses of 1,966. Average costs for candidates who had selection expenses range from 970 for Plaid Cymru candidates, to 3,903 for Conservative candidates, but there is significant variation in selection expenses for candidates in the same parties. Table 5. Selection expenses of candidates in 2015 general election (excluding candidates who reported no selection expenses) Con Labour Lib Dem SNP UKIP Green PC Mean ( ) 3093 2179 1136 1208 2223 1090 970 Standard dev. 7067 7397 4655 2651 4030 3581 1038 Question: How much have you spent (approximately) on selection expenses for the 2015 general election? N = 772 (2015) Figure 2 shows the four most often cited selection expenses incurred in the 2017 election. Of these, travel and accommodation were the most frequently cited selection expenses candidates incurred, followed by the costs of leaflets, and deposits. A smaller number of candidates (26) did say they lost income in taking time off work to attend selection and other campaigning events. 6

Figure 2. Most significant selection expenses incurred in 2017 general election Salary loss Deposit Leafleting Travel/accommodation 0 20 40 60 80 100 120 More generally, candidates think the costs of entering politics is too high. As shown in Table 6, with the exception of UKIP in both 2017 and 2015, a majority of candidates felt the costs of politics were too high compared to about right or too low. While candidates did acknowledge help from the local and, in some cases, national party many commented on the personal financial costs. We received donations from people in the party, and I think the local party asked for funds from members for the local campaign, and they got some, but it wasn t a vast amount, and I think we had one person who made quite a large donation, but we didn t do a lot of fundraising. We were mainly trying to be visible, and we knew we had enough money to cover what we wanted to do, so we didn t have to spend time fundraising. I would just say it is a really difficult process. I mean it has I think one of the biggest problems is the impact that it has on you personally so your time, your money, your personal life. I mean I did calculate that I probably lost about 9,000 of my own money during that campaign of travelling up there every week which is 100 return on a train. I rented out a flat there to stay in which was like 400 a month. Table 6. (Monetary) costs of entering politics in Britain (%) Con Labour Lib Dem SNP UKIP Green PC Too high 2015 50 67 65 81 39 68 59 2017 50 54 58 50 31 67 57 About right 2015 47 30 28 14 54 25 33 2017 47 44 42 50 67 33 43 Too low 2015 2 1 1 0 3 0 0 2017 1 1 0 0 2 0 0 Question: Do you think the (monetary) cost of entering politics is too high, about right, or too low? N = 1554 (2015) and 1148 (2017) B. Social background of parliamentary candidates Table 7 shows the socio-demographic profile of candidates standing in the 2015 and 2017 general elections. Given the that the 2017 election followed so closely on the 2015 election, there are few significant differences between the two there was substantial overlap in terms of the candidates who stood, but also there was little time between elections such we would expect significant changes in composition of the candidate pool. Men make up more than 70% of candidates standing for office Fewer than 1 in 10 candidates in 2015 and 2017 are non-white 7

More than 8 in 10 candidates are university educated Approximately one-quarter of candidates went to a fee-paying school Table 7. Socio-demographic profiles of candidates Sex, Race, Education, Occupation & Age All Con Labour Lib Dem SNP UKIP Other Female 2015 27 26 34 26 36 13 36 2017 29 29 41 29 34 13 22 Male 2015 73 74 66 74 64 87 64 2017 71 72 59 71 66 87 78 Non-white 2015 7 10 8 8 2 7 4 2017 8 7 10 7 2 6 17 White 2015 93 90 92 92 98 93 96 2017 92 93 90 94 98 6 83 2015 74 57 87 73 94 77 79 Non-fee paying 2017 76 60 89 75 94 84 81 Fee-paying 2015 26 43 13 27 6 23 21 2017 24 40 11 25 6 16 19 No degree 2015 16 15 15 12 23 35 11 2017 15 13 15 10 24 44 12 University 2015 84 85 85 88 77 65 89 2017 85 87 85 90 76 56 88 Oxbridge 2015 19 29 18 17 0 6 11 2017 18 28 17 15 0 7 10 Brokerage 2015 15 13 19 16 10 12-2017 16 18 19 18 19 9 - Business 2015 25 40 13 22 16 40-2017 18 34 11 22 15 18 - Manual 2015 1 1 2 0 2 2-2017 1 1 1 1 2 4 - Instrumental 2015 28 28 38 28 38 19-2017 21 29 35 22 41 11 - Other 2015 31 19 28 35 35 27-2017 23 14 28 30 24 19-18-39 2015 27 24 27 28 26 21 36 2017 27 42 28 27-4 - 40-59 2015 54 63 50 49 61 50 47 2017 45 47 45 41 0 39-60+ 2015 19 12 23 23 12 29 17 2017 29 12 27 32 0 57 - N = 1800, 1255, 1800, 1572, 1801, 1442 and 1561 (2015) and 3194, 3304, 1249, 2059, 1905, 3298 and 1000 (2017) 7 C. Ideology: Candidates, voters and parties To better understand the distribution of liberal - conservative ideology, we asked candidates to place themselves on the left-right scale, where 0= the most left-wing and 10= the most right wing. Table 8a shows how candidates placed themselves on the left-right scale and largely conforms to expectations. For Conservative candidates, only 2% identified as to the left of centre ( 5 ), with 88% placing themselves between 6-8 on the scale. For Labour candidates, again only 2% placed themselves to the right of centre, with 85% placing themselves between 8

2-4 on the scale. Just 13% of Liberal Democrat candidates and no SNP candidates placed themselves to the right of centre. Table 8a. Left-right placement of yourself (%) Left-right placement 0 1 2 3 4 5 6 7 8 9 10 Conservative 2015 0 0 0 1 1 6 24 38 26 3 1 2017 0 0 0 0 2 9 32 34 19 3 2 Labour 2015 4 4 25 39 21 4 2 0 0 0 0 2017 9 10 30 34 13 3 0 0 0 0 0 Lib Dem 2015 0 2 9 23 30 23 10 3 0 0 0 2017 1 2 8 27 33 21 6 3 0 0 0 SNP 2015 0 8 25 36 19 11 0 0 0 0 0 2017 0 0 21 21 36 21 0 0 0 0 0 UKIP 2015 0 0 1 2 5 23 21 27 15 4 3 2017 0 0 1 3 2 25 21 21 20 5 4 Question: Using a scale from 0 to 10, where 0 means the most left-wing and 10 the most right-wing, where would you place your own views? N = 1548 (2015) and 1131 (2017) We then asked candidates to place the voters in their constituency using the same scale. Generally speaking, the data in Table 8b suggests that candidates think voters in the constituency are far more distributed along the ideological continuum. This makes sense given that candidates are asked to place all voters in the constituency, not just voters from their own party. For example, in 2015, 92% of Conservative candidates placed themselves to the right of centre, but placed just 43% of voters to the right of centre. For Labour, 94% of candidates placed themselves to the left of centre, but place just 33% of voters in their constituency left of centre. Table 8b. Left-right placement of voters in your constituency (%) Left-right placement 0 1 2 3 4 5 6 7 8 9 10 Conservative 2015 2 1 10 14 26 7 14 15 9 1 2 2017 1 3 5 15 27 17 10 14 7 0 1 Labour 2015 1 1 4 12 15 9 14 23 15 6 1 2017 1 0 3 10 18 20 14 18 11 3 2 Lib Dem 2015 0 1 12 12 10 14 15 21 11 3 1 2017 2 3 7 11 10 13 17 20 13 4 1 SNP 2015 0 3 3 38 22 16 16 0 3 0 0 2017 0 0 7 21 7 43 21 0 0 0 0 UKIP 2015 2 0 5 11 15 14 16 17 14 3 1 2017 2 1 4 15 13 21 21 19 4 0 2 Question: Using a scale from 0 to 10, where 0 means the most left-wing and 10 the most right-wing, where would you place voters in your constituency? N = 1429 (2015) and 1131 (2017) Finally, we asked candidates to place your party s views using the same 0-10 scale. Generally speaking, Conservative and Labour candidates thought their party s views occupied the centre/centre-right and centre/centre-left respectively. Only UKIP had significantly different alignment: whereas 71% of candidates placed themselves to the right of centre (i.e. 6-10 on the scale), 97% placed their party s view to the right of centre. 9

Table 8c. Left-right placement of your party (%) Left-right placement 0 1 2 3 4 5 6 7 8 9 10 Conservative 2015 0 0 0 1 1 3 21 44 29 1 0 2017 0 0 0 0 0 0 4 23 43 28 2 Labour 2015 0 0 8 41 35 12 2 1 1 0 0 2017 0 9 21 39 24 1 2 0 0 0 0 Lib Dem 2015 0 1 2 10 37 38 11 2 0 0 0 2017 0 0 2 7 22 39 28 2 0 0 0 SNP 2015 0 0 6 56 31 6 0 0 0 0 0 2017 0 0 0 21 50 29 0 0 0 0 0 UKIP 2015 0 0 0 1 2 15 33 29 19 0 2 2017 0 0 0 0 1 2 21 20 25 16 15 Question: Using a scale from 0 to 10, where 0 means the most left-wing and 10 the most right-wing, where would you place your party s views? N = 1883 (2015) and 1131 (2017) D. Representation In this section we examine candidates views on their representative role how an MP should vote, who an MP should represent, and traits that candidates think voters most prefer. Table 9 shows candidates views of how an MP should vote. There is strong evidence where a candidate s opinion differs from her party or constituency, she should vote based with her own opinion or views. On average, candidates think they should vote with their constituency over their party, with the exception of Labour and the SNP. Table 9. How should an MP vote, 2015 responses (%) Vote with: All Con Labour Lib Dem SNP UKIP Other Own opinion 83 70 70 86 82 64 36 Party 17 30 30 14 18 36 64 Party 41 46 58 42 56 21 38 Constituency 59 54 42 58 44 79 62 Own opinion 79 70 70 86 36 82 82 Constituency 21 30 30 14 64 18 18 Question: How should an MP vote if his/her own opinion on an issue does not correspond with his/her party s position? N= 905, 883 and 909 (2015) Table 10 shows candidates responses to who they should primarily represent. We asked, How important is it for MPs to represent the following groups of people in the House of Commons. On average, 99% of candidates said it was important for an MP to represent all citizens in the constituency, compared to 80% who said it was important to represent the voters who supported the MP in the constituency. 10

Table 10. Who should an MP represent, 2015 responses (%) Lib All Con Labour Dem SNP UKIP Other Voters who supported the MP in the constituency 80 92 85 79 87 83 71 All citizens in the constituency 99 97 98 100 100 100 100 The party's supporters in the UK as a whole 69 72 78 65 36 75 63 All citizens in the UK 72 74 75 68 33 78 72 Question: There are different opinions about whom an English/Scottish/Welsh MP should primarily represent. In your view, how important is it for MPs to represent the following groups of people in the House of Commons? N = 896, 930, 903 and 888 (2015) Finally, we asked candidates to think about the traits that voters may find important in deciding who to vote for. Table 11 shows the first most important and the second most important traits for voters in a candidate s constituency. Candidates thought the most important trait for voters was living in the constituency (44%), followed by being an excellent public speaker (20%), being well-versed in the machinery of politics (12%) and growing up in the area (10%). Table 11. Importance of different candidate traits, 2015 responses (%) All Con Labour Lib Dem SNP UKIP Other Candidate traits - first most important Excellent public speaker 20 15 18 18 12 22 24 Lived in the constituency for 44 43 43 48 47 47 40 Well versed in the machinery 12 10 13 9 18 11 15 Good personal appearance 5 6 10 6 0 1 4 Grew up in the area 10 15 8 10 12 15 8 Highly influential in his/her 8 10 9 8 12 5 9 Candidate traits - second most important Excellent public speaker 21 15 21 18 33 22 23 Lived in the constituency for 20 22 16 19 33 19 21 Well versed in the machinery 22 19 21 25 27 17 23 Good personal appearance 12 12 17 9 7 11 11 Grew up in the area 15 20 15 16 0 18 11 Highly influential in his/her 11 12 10 12 0 13 10 Question: Voters may find many different candidate traits important when deciding whom to vote for. Of the six traits listed, select the most important and second most important for voters in your constituency? N= 892 and 878 (2015) Table 12 shows the percentage of candidates and MPs who live in the constituency for which they stood in 2015 and 2017. More than 8 in 10 MPs lived in their constituency compared to 61% for non-incumbents in 2017 and 56% in 2015. 11

Table 12. Whether MPs and candidates lived in constituency for which they stood in 2015 and 2017 (%) Incumbent MPs Not incumbent MP Live in constituency Do not live in constituency Live in constituency Do not live in constituency 2015 86 14 56 44 2017 80 20 61 39 Question: Did you live in the constituency in which you stood in the 2015/2017 general election? N = 1567 (2015) and 1223 (2017) Candidates views on the importance of living in the constituency and local connections also came out in our interviews. I think, increasingly, people are more and more local, and I think people are expecting that. Therefore, a record of what you ve done in your local community as much as what you re doing in Parliament is important to the voters. [Being local] I think it's very important. I think it certainly helped in my case. It can work both ways because sometimes you get an outsider with a fresh perspective, but from an individual applying to a certain seat, I think it is very important, mainly because the people know you and they trust you and they've seen you work there. They feel, okay, you can sort of hit the ground running. There's less of a learning curve Yes, I think so. I think a lot of my constituents want to know that you know the area, you care about it, and local party members felt that as well. In the past, we ve had candidates from other parts of the country, and people would ask, Well, who are they? How do they know...? because I think my constituency, whilst it is part of the City of X, it is made up of individual towns and villages that have their own identity. So therefore, it s quite close-knit. I think those that did live in the constituency, or work in the constituency, made something of that. I made something of the fact that I had been a councillor in the area, and done work in the area as part of that. 2. Campaigning for office In this section we present candidates responses to time spent on various campaigning activities. Table 13a. Time (in hours) spent door-to-door canvassing in 2015 and 2017 (%) Con Labour Lib Dem SNP UKIP Green PC 0 2015 4 5 38 0 27 51 0 2017 5 12 55 0 79 84 14 0-5 2015 7 8 19 0 18 27 12 2017 3 1 0 8 0 0 0 6-10 2015 16 16 15 14 15 12 29 2017 1 0 0 8 0 0 0 11-20 2015 19 25 15 41 8 6 18 2017 54 44 20 23 14 6 43 20+ 2015 55 46 14 46 20 5 41 2017 38 43 24 62 7 11 43 Question: During the final month of the campaign, how many hours per week did you personally spend on the door-to-door canvassing? N = 1102 (2015) and 1249 (2017) 12

Table 13b. Time (in hours) spent leafleting in 2015 and 2017 (%) Con Labour Lib Dem SNP UKIP Green PC 0 2015 2 14 29 5 8 22 0 2017 6 25 40 8 51 68 9 0-5 2015 12 32 29 50 17 33 41 2017 6 12 2 17 0 0 0 6-10 2015 20 23 19 18 19 23 18 2017 2 5 1 17 0 0 0 11-20 2015 25 18 12 14 19 12 24 2017 48 33 23 25 33 12 27 20+ 2015 42 13 12 14 37 10 18 2017 39 24 34 33 17 20 64 Question: During the final month of the campaign, how many hours per week did you personally spend on leafleting? N = 1095 (2015) and 1249 (2017) Examination of the time candidates spent campaigning on social media sites shows a step change between 2015 and 2017, with a significantly larger proportion of candidates from all parties spending twenty hours or more campaigning via social media in 2017 than in 2015. However, in 2017 candidates of all parties- with the exception of UKIP- also spent more time in the more traditional activity of delivering leaflets than in 2017. In 2017 51% of UKIP candidates reported spending no time leafleting compared to 8% in 2015. With the exception of the SNP, more candidates of all parties reported spending no time calling voters in 2017 than 2015. Table 13c. Time (in hours) spent campaigning on social media in 2015 and 2017 (%) Con Labour Lib Dem SNP UKIP Green PC 0 2015 20 23 45 0 50 19 12 2017 35 21 66 0 73 67 17 0-5 2015 59 57 43 73 26 49 59 2017 26 29 3 58 0 0 33 6-10 2015 14 14 7 14 11 18 24 2017 11 8 0 17 0 1 0 11-20 2015 4 4 3 5 5 10 0 2017 12 18 13 8 15 14 17 20+ 2015 3 2 2 9 8 5 6 2017 16 24 18 17 13 18 33 Question: During the final month of the campaign, how many hours per week did you personally spend on social media campaigning (e.g. Facebook)? N = 1200 (2015) and 1249 (2017) Table 13d. Time (in hours) spent calling voters in 2015 and 2017 (%) Con Labour Lib Dem SNP UKIP Green PC 0 2015 27 32 58 48 83 94 56 2017 58 68 90 0 78 89 64 0-5 2015 44 40 31 38 13 4 25 2017 18 11 2 58 0 0 0 6-10 2015 15 14 7 10 2 2 13 2017 7 7 1 8 0 1 9 11-20 2015 8 8 3 0 2 0 6 2017 10 5 3 25 10 3 18 20+ 2015 6 6 1 5 1 0 0 2017 7 9 5 8 12 8 9 Question: During the final month of the campaign, how many hours per week did you personally spend phoning voters? N = 1097 (2015) and 1249 (2017) 13

Table 14. Average number of hours devoted to campaigning per week (%) Constituency Con Labour Lib SNP UKIP Green PC marginality* Dem Safe held 2015 55 50 63 1 - - 41 seat 2017 69 66 63 66 - - 55 Marginal 2015 37 54 39 - - 31 46 seat 2017 69 61 58 30 50-42 Safe opposition 2015 36 31 15 32 25 16 25 held seat 2017 51 34 25-17 16 26 Question: About how much time did you devote to campaigning per week in the last month? N = 1110 (2015) and 856 (2017); * Safe held = majority of >10%; marginal = <10%; safe opposition = opposition majority >10% In our interviews, candidates frequently commented that campaigning was extensive and arduous. That means after 6pm Monday to Friday that was my campaign time. Monday to Friday I travel from X to X, on the train going to work and back home I was able to communicate via email, go to hustings events after 6pm, make phone calls to local residents and answer their questions. Then when I got back home I was able to update my Facebook, Twitter, Weibo and WeChat. There are a few different online channels that I use to communicate with voters. Four to six weeks? Well, it was literally every single day and night, literally non-stop. It s great for losing weight. Probably for the first year I was 50% of my time on the campaign and for most of the next year I was probably 2/3 of my time on the campaign and then for the last 3 months I was full time on the campaign. [Estimate of time spent campaigning] Every day Mainly out and about canvassing. Moreover, for candidates with caring responsibilities, campaigning was even more challenging. [On balancing campaigning with childcare, work, family etc.] That was a big challenge. I ve got two small children I would just bring her along with me, which I ve always done in lieu of any other child carer; brought them along into council meetings and things like that as well, when I was a councillor Mum was very helpful and my husband was very helpful. We worked around it. It was difficult. [On managing commitments] The family commitments take a huge hit, because that s almost sometimes the easiest thing to go. That takes a massive hit. The work, I was very, very lucky. I took a lot of holiday and where I work you can buy work holiday, you can also carry your holiday from one year to the other. I did all the tricks I could, to get as much holiday as possible. I took about 25 days holiday. Finally, we probed candidates views on support from their central party. Here we found a great deal of variation in experiences, largely contingent on whether candidates were running in a safe or marginal seat. Yes, I had regular contact about various things. Also, some requests to go and help colleagues, and do some speaking, and do some hustings in other places as well. I had plenty of contact with the party machine, but I didn t get visits or anything like that, nor did I expect any. [unwinnable seat] 14

No, not really no because they never saw this as a winnable seat. I mean, there was a lot of support in the sense in terms of all the literature and all that was all available. There were daily media briefings. You got, sort of, a daily briefing pack where you could speak to- so you knew about what the party line was, what the party policy was on issues. There was a lot of background support but no direct support. Well, originally, essentially by me because, as I say, we weren't thought to win. We didn't have funding. Towards the end of the campaign, so sort of November time, the party had polling which suggested we could be in play. And then we at that stage were provided with a campaign manager by the party centrally, and funding for literature. Well, having been selected, I did need to know the members of the local association I worked at who were interested in campaigning, who were willing to give up time to help. And then when I got allocated a campaign manager by the party headquarters, there was a set structure then. What the party was trying to do was they were doing the national thing and they wanted you to make sure you were coming across a sort of good egg that would be a good local MP. And they'd take care of national things, and that worked quite well. [On contact with party HQ] Yes. We got regular briefings on things, literature was sent to us for putting out. Lots of crazy things like foam fingers that we ve never had before the referendum. Things to put on street stalls to give away, those kind of things. [On encouragement to stand] Women were very much encouraged to stand. We ve been doing women s academy programmes within the party to encourage women who... There were a number of folk that were elected who were councillors already. 3. Harassment and intimidation: Experiences of 2017 parliamentary candidates Over the past month I ve had swastikas carved into posters, social media posts like burn the witch and someone even urinated on my office door. Hardly kinder, gentler politics. Sheryll Murray, Conservative MP for South East Cornwall 8 The harassment and intimidation of candidates during electoral campaigns is not a new phenomenon, but has certainly become more relevant and worrisome since the death of Jo Cox, Labour MP for Batley and Spen, on 16 June 2016. A wealth of anecdotal evidence indicated that the 2017 campaign was particularly nasty, with candidates raising concerns throughout the campaign. For example, Conservative MP Simon Hart, said that he considered elections to be a few weeks of robust banter followed by a shake of the hand and a pint in the pub when first elected in 2010, but the most recent election was characterised by swastikas on election boards, offensive slogans and language on posters (The Guardian, 12 July 2017). His, and similar other statements, suggest that electoral campaigns have indeed become more aggressive in the last decade. Drawing on RAB data 9 from the 2017 general election, we asked: how widespread are harassment and intimidation during electoral campaigns; how does harassment affect candidates emotionally; and what are the most frequent forms of harassment and intimidation? 10 32% of candidates who have participated in the survey said they experienced some form of inappropriate behaviour during the 2017 general election campaign Women candidates were more likely to have experienced abusive behaviour (38%) compared to male candidates (30%) 15

Abuse of women candidates is not directed at any specific age group; women of all ages receive abuse Women and Conservative candidates are more likely to suffer harassment and intimidation Figure 3. Percentage of candidates experiencing harassment/intimidation by party 100 90 80 70 60 50 40 30 20 10 0 Conservative Green Labour LibDem Plaid Cymru SNP UKIP No Yes We asked candidates the extent to which they felt annoyed, concerned or fearful in response to any inappropriate behaviour they encountered. 95% of candidates find abuse annoying, a majority (83%) are concerned, and 60% say they are fearful Sending inappropriate emails and abuse on social media are the most common types of inappropriate behaviour; physical attacks, thankfully, are rare 64% of candidates report inappropriate behaviour by supporters of opposition parties/candidates Figure 4 shows that while sexual harassment and physical abuse are relatively rare, online abuse in social media or by email is more frequent. This is followed by threats made over the phone or by letter. The statistical analysis shows that female candidates are more likely than male candidates to be threatened and sexually harassed. We also find significant party differences in candidates receiving unwanted approaches, interference with their property, abuse expressed in social media, email, and by letters. 16

Figure 4. Types of harassment experienced by candidates Sexually Phone Letter Social media Email Interfered with property Loithered Followed you Approached you Threatened Physically attacked 0 10 20 30 40 50 60 70 80 90 100 Three times or more Once or twice No Finally, in an open-ended question we asked: What measures, if any, should be taken to increase the security of election candidates? Candidates offered ideas of possible lines of action that can be implemented to tackle harassment and intimidation during campaigns. Among the most frequent suggestions are: to keep the option of anonymity of candidates home addresses, provide candidates with security advice, force social media platforms to take action against identified users who threaten or harass candidates and to ensure that candidates are accompanied in all moments while campaigning. Maintain the option of anonymity of candidates home addresses Serious and quick response from police; several candidates indicated that police are unsupportive, e.g. "don't make a fuss about this" Security advice from parties, police and other organizations Social media platforms (Twitter, Facebook etc.) should be forced to take action against trolls A special watchdog to investigate this type of abuse Ensure that candidates are accompanied in all moments while campaigning Stronger actions taken against offenders (e.g. including jail and fines) Parties should tell their members that this behaviour is unacceptable. Finally, authorities, party leaders and the general public should be aware of the consequences that harassment has on the quality of elections. If it is becoming more frequent and distressful, and politicians testimony would suggest it is, then intimidation during campaigns may well put off potential candidates who would have been of service to the public and lower the quality and diversity of political representation. 17

III: Representative Audit of Britain research In this section, we report on the key findings and research outputs from RAB data. A full list of RAB research papers and reports can be found in Section IV. 1. Do candidates and MPs differ from party officers and members in their political views? Peter Allen, Tim Bale, Rosie Campbell, Jennifer Hudson, Monica Poletti & Paul Webb The question of whether the views of a party's MPs and candidates reflect the views of the membership is an important one. Unity levels affect the nature of policy proposals put forward by the party, the party s ability to present a united front to the electorate, and the likelihood that members will remain loyal. To examine how the views of party elites (MPs and candidates) compare to those of party members, we combined data from the Representative Audit of Britain with data from the ESRC-funded Party Members Project. 11 We compared the groups across two longstanding measures of political ideology a left-right self-placement and an established battery of questions measuring how liberal or authoritarian the respondent's views are. Within the members surveyed, we distinguish between active members who contributed more than 20 hours to campaigning activities in the five weeks preceding the 2015 general election, and passive members who contributed less than this. We also remove from the members sample any individuals who were candidates for, or holders of, elected public office. We then do the same for the 2017 general election. Here, we focus on the left-right distinction (Table 15). We asked respondents where they would place themselves on a 0-10 scale. Among Conservatives and UKIP, MPs and candidates are slightly more left-wing than both passive and active members while the opposite is true for Labour, the Liberal Democrats, and the Greens. As such, the ideological self-placement of party elites across the political spectrum relative to members is centripetal. Table 15. 2015 Left-right self-placement MPs and Party Active Members Passive Members Candidates Conservatives 7.79 7.77 6.98 Labour 2.11 2.45 3.06 Liberal Democrats 3.84 4.21 4.24 Greens 1.92 1.90 2.38 UKIP 7.12 7.44 6.60 0=extremely left-wing, 10=extremely right-wing Turning to 2017 (Table 16), some notable shifts are identifiable. First, Labour s MPs and candidates appear to have moved slightly further to the right, something perhaps contrary to the narrative surrounding the party, whilst members, especially passive members, have moved to the left. Conservative MPs and candidates have similarly shifted right, while the membership has remained fairly ideologically stable. Notwithstanding the smaller numbers of UKIP members and candidates in 2017 compared to 2015, there seems to have been some divergence between active members, who have moved slightly left, and MPs and candidates who have done the opposite. The cases of the Liberal Democrats and Greens also indicate greater divergence between active members and MPs and candidates, suggesting something 18

of a theme within British parties right now. What this means for policy development and party harmony in the next few years remains to be seen. Table 16. 2017 Left-right self-placement MPs and Party Active Members Passive Members Candidates Conservatives 7.70 7.53 7.74 Labour 2.09 2.10 3.47 Liberal Democrats 3.83 3.94 4.87 Greens 1.69 1.92 2.86 UKIP 6.97 7.56 7.50 0=extremely left-wing, 10=extremely right-wing 2. Do candidates and voters have similar policy views? Does it matter? Caitlin Milazzo & Jennifer Hudson Using data from the 2015 Representative Audit of Britain and the British Election Study 12, we address two questions: (1) Under what conditions do British candidates express issue position that are in line with public opinion in their constituency? (2) Is there a vote benefit to candidates who do so? When we compare the candidates positions on key policy issues with those of their constituents, we find that, regardless of party, candidates competing in marginal seats and sitting incumbents are more likely to be in tune with the views of their constituents. Incumbents also better approximate public opinion on specific issues, including income redistribution and the immigration. In addition, we find that issue convergence is not a pointless exercise. Candidates who better align their views to those of their constituents tend to win more votes. Figure 5 illustrates how a candidate s vote share is expected to decline as she moves further away from public opinion in her constituency. Our estimates indicate that the vote share of the average candidate is 1.4 points lower than what she would receive if her views were perfectively aligned with the preferences of her voters. And, when we limit the analysis to candidates from the Labour and Conservative parties only, we find that the effect of policy convergence is even more substantial; the vote share of the average candidate from major parties is nearly three points lower than she would have received had she been aligned with her voters. And, while the effect of convergence is modest, it is not significantly less than the effect of being local or having previous experience as an elected representative at other levels of government. 19

Figure 5. Policy differences between candidates and constituency opinion 0 Low High Difference in vote share -2-4 -6-8 All candidates Lab/Con only Finally, we also find differences in candidate positioning across parties may help to explain the Conservative victory in 2015. On average, the Conservative Party tended to select candidates whose views better aligned with voters preferences. While Labour candidates were, on average, closer to the median voter on income redistribution, Conservative candidates tended to better approximate their constituents views overall, as well as on the issues of European integration and immigration. As of March 2015, immigration was rated as the most important issue facing the country. While Europe was less salient, it nevertheless featured heavily in the campaign. That Labour s candidates tended to be more out-of-step with their constituents, generally and on these key issues, could help to explain the party s loss to the Conservatives, particularly in key marginal seats where small shifts in votes decide the outcome. 3. Explaining motivation to represent: how does descriptive representation lead to substantive representation of racial and ethnic minorities? Maria Sobolewska, Rebecca McKee & Rosie Campbell Research has shown a link between the representation of the interests of racial and ethnic minorities and their physical presence in legislatures. But scholars have moved beyond the simplistic claim that any woman, black or Asian will do and thus there is a need to understand under what conditions MPs from minority groups attempt to act in these groups interests. It might be that BAME voters and MPs have shared experiences or a sense of responsibility that lead BAME MPs to prioritise the representation of ethnic minority voters. Alternatively, rather than being normatively driven to represent BAME voters BAME MPs might secure an electoral advantage by advocating for BAME communities in their constituencies. The majority of the research on the political representation of BAME voters has been conducted in the USA. But in the USA it is almost impossible to differentiate between normative and electoral motivations for black legislators using observational data because the majority of black legislators represent minority-majority districts, something which is no longer the case in the UK. We find clear evidence that prospective minority MPs are influenced by both normative and electoral motivations. We find that minority candidates share a sense of common minority experiences, and feel a responsibility to represent minority voters, although this is moderated by political 20