Where candidates live matters to voters, and they show it in their voting

That voters prefer to elect local candidates is a long-held assumption of British politics. Professor Jocelyn Evans’ research has sought to test that assumption. He found that the geographical distance between candidates’ homes and the constituency had a measurable impact on voting behaviour. In this post he shares his findings and argues that voters should have access to more information about the ‘localness’ of those seeking to represent them.

Election candidates like to stress their local roots. Credit: Irish Labour Party (CC BY-ND 2.0)

Voters prefer their MPs to be local. What may seem to many a statement of the obvious has only recently begun to be confirmed empirically by research into the origins of MPs and the evaluations of voters. That this is received wisdom amongst MPs themselves has been shown in recent research by Richard Berry on Democratic Audit – those that can (and many can’t) go out of their way to emphasise their local origins. Voters too have indicated that the attribute mattering most in how they assess candidates’ aptitude for the job is where they live, as demonstrated by Rosie Campbell and Philip Cowley. Academic research from the UK and abroad is replete with reasons for why this should be the case – candidate commitment to the area, efficiency of constituency service, knowledge of the area, being ‘one of us’, even individual benefits from proximity to a winner.

Voters and MPs say it matters. Political science thinks it should matter. The research carried out by Kai Arzheimer and myself on the 2010 General Election looked to test, for the first time in the UK, the idea that local did matter to voters under election conditions. Even if MPs emphasise their roots, and voters claim localness is more important than, say, the gender or the occupation of the candidate, does this manifest itself to any extent in their actual behaviour in an election? Measure the distance by car from a voter’s home to each of the constituency candidates’ homes, and other things being equal, does the next-door neighbour beat the out-of-towner?

Many datasets for few variables

Before turning to our findings, it is worth considering the set-up for this analysis. Testing this in an election rather than an opinion survey or experiment needs a lot of distillation of dirty data. We had to retrieve the relevant information about locations, voting and constituency context by bringing together a disparate array of sources – an object lesson in Big Data. In many ways, 2010 was the first General Election to make an exploratory study such as this feasible. Full candidate data from all 650 constituencies were available, if long-winded to obtain, by download from local councils. A representative sample of the electorate with a relatively precise estimate of their location, together with necessary variables on pre-election voting intentions, as well as actual vote on the day, could be obtained as usual from the British Election Study. Finally, Ordnance Survey’s Code-Point® data now identifies the latitude and longitude of each of the circa 1.7 million postcodes in the country, and allowed us to match voters to constituencies and to candidates, and use GoogleMaps to calculate relevant distance. Our next constraint then was one of statistical necessity – we could only look at a set of voters who had all been presented with the same choice of relevant parties, so we had to restrict our analysis to England.

2010 was also a bad year to start. For the first time in 140 years, candidates did not have to provide their full home address on the ballot paper. For reasons of candidate privacy and security, a new clause added to the Political Parties and Elections Bill in 2009 allowed candidates to request that only their constituency of residence rather than full address be made public. In the context of this research, this is a significant change. Does a voter need to know exactly where their candidates live, rather than simply whether they live in the constituency or not? Such a question, which is surely for the voter themselves to answer, is now settled by the candidate. Finding out which constituencies we would be able to test, given missing addresses, itself threw up some interesting results. In fact, for 2010, well over 70% of candidates still provided their full address. Almost half of all candidates for the three main parties lived in the constituency they were contesting and gave their address. Comparing candidates living in their constituency with those living outside, a much higher proportion of the latter withheld their addresses – almost twice as many. The Liberal Democrats, closely followed by Plaid Cymru and the SNP, were the most likely to give their address – around 85% for all three. Finally, of all parties, the BNP candidates were most likely to withhold their addresses, as well as having the most candidates living outside the constituency.

The effect of distance

The bar we then set ourselves to confirm that what, for want of a better word, we termed ‘localness’ mattered, was high. Controlling for the standard explanations of vote and candidate performance – general party loyalty, socio-economic context, incumbency – did the relative distances that the voter lived from each of the candidates have any independent effect on which they chose? We would not naively claim that voters calculate the distances to their respective Conservative, Liberal Democrat and Labour candidates in order to work out who to vote for. But if we could show a significant relationship between something as direct as the distance between a voter’s home and the candidates’, it would be difficult to refute that current localness was playing some role in voters’ evaluations of candidates.

For voters in England, we found that this distance did matter. It was not a game-changer – the 2010 General Election could not have been won by Labour by some judicious house-moves. The standard explanations of vote still mattered more (and we would have been suspicious had they not). Nevertheless, the location of the candidate relative to the voter influenced the latter’s likelihood of voting for them. As a simulated example of how this might matter: in a notional constituency where the candidates from the three main parties all lived 26km from a voter (the average distance which candidates lived from the voters in our data), moving any one of those candidates to a location 120km away would lose him / her 16 percentage points, if it was the Conservative, 10 points if it was the LibDem, and 9 points if the Labour. In an election where a hypothetical loss of eight percentage points from the winning party, with half of that going to the second placed party, would have resulted in a different outcome in one-third of all constituencies, distance mattered.

Of course, distance taken from where a candidate lives is not the only, or necessarily the most important, aspect of ‘localness’. Birthplace and length of domicile all matter. Whilst geographical distance matters for some, more subjective district or regional locations may also colour voters’ perceptions and consequently alter behaviour. Lastly, there are obvious cases where no localness effect will overturn party loyalties. Shaun Woodward is not penalised in any meaningful sense amongst the voters of St Helen’s South and Whiston for his Cotswolds domicile (address withheld, incidentally). But in a marginal constituency, the evidence is equally clear: location will count.

Voters say that they want local MPs. Candidates respond. Our research shows that this is not lip service on the part of the electorate, and that the importance manifests itself in quite a direct manner – the closer you live, the more likely I am to vote for you. Returning to the 2009 amendment to the rules on statements of home address, it seems perverse to deny voters the opportunity to know where the candidates they have to choose from live, given the importance tests of this idea confirm. In many ways, we idealise the informed rational voter. Given an involved concept such as relative distance made a difference to how English voters behaved in 2010, shouldn’t all voters have that information to hand in 2015?

Note: This post represents the views of the author, not those of Democratic Audit or the London School of Economics.

Jocelyn Evans is Professor of Politics at the University of Leeds. Professor Evans’ research interests cover European politics and electoral behaviour, with a particular focus on France. He also works on aspects of electoral forecasting, especially predicting third-party support. He is editor of Palgrave Macmillan’s “French Politics, Society and Culture” monograph series. He tweets at @JocelynAJEvans.