By Joshua Townsley
At 10pm on the 7th May
2015, as polling stations were closing across the UK, the GfK, NOP and Ipsos MORI exit poll was released
that predicted the Liberal Democrats would be reduced
to just 10 seats in parliament. Former leader Paddy Ashdown, who had been at the centre of the Lib Dem general
election campaign and was live on air when
the poll was released, was so shocked that he promised to publicly eat his hat
if the exit poll proved to be correct. By
the end of the night, the Liberal Democrats
were left with only 8 seats. In the early hours of the
8th May, Nick Clegg, Lib Dem leader and re-elected MP for Sheffield Hallam,
stepped up to the podium at his seat’s count
and admitted that the night had been a “cruel and punishing
night for the Liberal Democrats”. Less than 12 hours later, Clegg announced his resignation as leader,
describing the election as “immeasurably more crushing and unkind
than (he) could ever have feared”.
The Lib Dem mantra ‘Where
We Work We Win’ captured the belief that where the party’s MPs were ‘dug in’
locally, they would be very hard to beat. Two months before the 2015 general
election, Clegg declared: “let me tell you this: we will do so much better than
anyone thinks. In those seats where we are out in force, making our case loudly
and proudly, we are the ones making the weather”. Political scientists and
commentators expected that the party would hold around half of its 57 seats,
based largely on local factors such as incumbency. However, the results of the
general election were devastating for the Lib Dems, reducing the party
overnight from 57 MPs to just 8. MPs perceived to be unbeatable in their
constituencies were swept away, and the party went from serving in government
to struggling for its very survival. The near wipe out of the Lib Dems raises
questions about the existing body of research into the party’s local
electioneering. The aim of this article is to assess whether one of the Liberal
Democrats’ top campaigning strengths – incumbency - made any difference to the
party’s electoral fate in 2015.
Incumbency
An incumbent MP has the
benefit of having won the constituency once, or several, times before and can
use this knowledge and the local networks built up to secure re-election. A
‘personal vote’ is a level of support for a candidate that is particular to
that candidate, rather than the party. MPs with a high profile in their
constituency can increase this personal support over time. For instance, an
incumbent MP usually benefits from greater name recognition than their
opponents, which can be vital in securing the support of voters based on a
‘personal vote’ for the candidate rather than the national party.
Lib Dems tend to be
particularly adept at this. According to British Election Study data prior to
the general election, around 70% of respondents in Labour or Conservative seats
could name their MPs, but among voters in Liberal Democrat seats, name
recognition was 82%. These figures were relatively consistent over time.
Crucially, given the Liberal Democrats’ reliance on building a ‘personal vote’
and on securing tactical votes, Lib Dem MPs tend to be more popular among
supporters of opposition parties. This makes it easier for Liberal Democrat MPs
to distance themselves from the unpopularity of the party nationally, and rely
on personal trust and name recognition in their constituencies.
(See Nottingham’s Phil
Cowley for more on these figures, here:
http://nottspolitics.org/2014/12/09/not-love-actually).
For the Lib Dems,
incumbency advantages such as the building of a ‘personal vote’ has been a
result of necessity. Traditionally, both main parties have long relied on
class-based support as working class voters support Labour and middle class
voters support the Conservatives. The Lib Dems on the other hand, lack this
robust social ‘base’ of support, so tend to rely on building local popularity
based on the personal qualities of hardworking councillors or MPs. The ability
of Lib Dem MPs to develop a ‘personal vote’ has been essential in building
electoral strongholds.
Overall therefore, there
was significant evidence that Liberal Democrat MPs enjoyed a larger incumbency
advantage than Labour and Conservative MPs that would minimise the decline in
vote share in their constituencies. This narrative was reinforced by
constituency polling carried out by Lord Ashcroft that revealed in some seats,
such as Eastleigh, Sutton and Cheam, and Carshalton and Wallington, the party’s
vote was holding up largely due to the ‘boost’ the party received in the
polling when respondents were prompted to consider their own constituencies and
the candidates standing. This heightened the belief that popular incumbent MPs
would be well placed to survive the storm that the party faced nationally.
Analysis
This article aims to
analyse whether incumbency had any effect on Liberal Democrat party performance
at the 2015 general election. We can do this by comparing the mean decline in
vote share between 2010 and 2015 in seats with and without an incumbent MP. In May,
the Lib Dems lost a larger proportion of their vote share where they were
previously strongest. Stephen Fisher (Oxford University) had predicted this
prior to the general election using data from the British Election Study.
However, this does not mean incumbency had no effect whatsoever. If incumbency
was irrelevant, and the party’s 2015 performance was based just on its 2010
vote share in each constituency (i.e. the party's vote share collapsed most
where it was previously highest), seats with a similar vote share would
experience broadly similar decline in party vote share regardless of whether
the seat had an incumbent MP or not. In 2010, the Lib Dems polled 23% of the
vote, but only 8% in 2015. To take into account the fact that the party,
mathematically, had to lose more votes where they were previously strongest, it
is necessary to compare the mean vote share decline in seats with and without
an incumbent across seats with similar 2010 vote shares. In this way, any
incumbency effect can be unpicked from the wider trend that the party suffered
larger swings where they had higher previous vote shares.
The table below compares
the mean vote share decline for the Liberal Democrats in seats with similar
2010 vote shares depending on whether the party had an incumbent MP
re-contesting the seat in 2015 or not. The fourth column shows the ‘difference’
in the mean decline in vote share between seats with an incumbent MP standing
again, and all other seats.
The figures show that the
party did indeed experience a larger decline in vote share in seats where they
were previously stronger. But the analysis also reveals evidence of an
incumbency effect for Lib Dem MPs in 2015. Among seats in which the Lib Dem
vote share was broadly similar in 2010, there was a significant difference in
the vote share decline depending on the incumbency status of the seat. For
example, in seats such as Argyll and Bute and Bradford East where the party
polled between 31% and 35% of the vote in 2010 AND had an incumbent MP
defending the seat in 2015, the Lib Dems only experienced a mean decline of
-4%. Meanwhile, in seats such as Oxford East and Manchester Gorton, where the
Lib Dems also polled between 31% and 35% of the vote in 2010, BUT did not have
an incumbent MP defending the seat in 2015, the party suffered a huge -22.6%
change in vote share in 2015. This trend is clear across all seats
according to their 2010 vote share, including the 57 seats the party was
defending in 2015.
At the last general
election, 17.86% of sitting Lib Dem MPs had decided not to recontest their
seats. This meant that out of 57 seats the Lib Dems were defending in 2015, the
party had new candidates defending 10 of them. As a proportion of the
parliamentary party, the amount of new candidates defending held seats was
higher for the Lib Dems than both Labour and the Conservatives. This hurt the
Lib Dems at the polls. The figure below compares the mean decline in Lib Dem
vote share in their 57 held seats based on whether the seat was being defending
by a new candidate replacing the existing MP (left hand column), or by the
existing MP (right hand column).
The figure shows that even
in seats in which the party polled over 51% in 2010 and won with a huge
majority, there was an 11 percentage point difference in the mean vote share
decline depending on whether the candidate in 2015 was an incumbent MP or not.
For example, in Bath the Lib Dems polled 56.6% of the vote in 2010, and
suffered a -27% swing in 2015 with a new candidate. Meanwhile in Westmorland
and Lonsdale, where the party polled a similarly high share of the vote in 2010
(60%), the incumbent MP only experienced a -9% decline in 2015. The anomaly in
the figure - the new candidate who managed to keep the party's vote share
decline to less than -4% - was Christine Jardine, who was defending the Gordon
seat previously held by Sir Malcolm Bruce against the SNP's Alex Salmond.
As shown in the figure
above, the decline in Lib Dem vote share tended to be much higher in seats
where the incumbent MP was standing down and a new candidate contested the
seat. The mean decline in Lib Dem vote share across all held seats was -15.7%.
Among the 45 seats in which the incumbent MP was restanding, the mean vote
share decline was slightly lower at -14.1%. But in the seats in which the
incumbent MP retired and was replaced by a new candidate, the mean vote share
decline was -21.3%. For example, in Bath the Lib Dems polled 56.6% of the vote
in 2010, and suffered a -27% vote share decline in 2015 with a new candidate.
Meanwhile in Westmorland and Lonsdale, where the party polled a similarly high
share of the vote in 2010 (60%), the incumbent MP only experienced a -9% vote
share decline in 2015. In held seats, the difference between having an
incumbent MP defending the seat and a new candidate was 7.2%, based on the mean
decline in Lib Dem vote share in both cases. While this ‘personal vote’ of
incumbent Lib Dem MPs was not enough to save the majority of seats, it was
strong enough to reduce the decline in Lib Dem vote share compared to held
seats where the incumbent MP stood down. This confirms arguments made prior to
the election that the party’s incumbency advantage is meaningless when the
incumbent MP stands down, regardless of whether the seat is held or not.
The results also seem to
confirm previous research on a phenomenon often referred to as the ‘sophomore
surge’. This describes the trend that first term incumbents tend to enjoy a
larger incumbency boost, and that this advantage does not tend to increase the
longer the MP remains an MP. There is clear evidence of a ‘sophomore surge’ for
first time Lib Dem incumbent MPs in 2015. While the mean decline in Lib Dem
vote share across all of its held seats was -15.7%, and the mean vote share
decline against all incumbent MPs was -14.1%, the mean vote share decline among
first time incumbent MPs was just -9.8%.
Therefore, contrary to
belief prior to the election, the party did in fact fare far worse in previous
‘strongholds’ in 2015. But there is evidence showing that even when previous
strength is taken into account by comparing seats with similar 2010 vote
shares, having an incumbent MP defending the seat in 2015 made a significant
difference in minimising the decline in Lib Dem vote share. The difference is
significant enough that it is likely to have saved the party from being wiped
out altogether at the 2015 general election. It is no coincidence that the 8
seats the party successfully defended all had incumbent MPs re-contesting their
seats.
Having an incumbent MP was
important, and may have saved the party from losing all of its MPs, but its
effect was essentially limited to reducing the swing – no incumbent was able to
buck the national trend and enjoy an increased vote share in 2015. Care must be
taken to avoid overstating the importance of incumbency to the Liberal
Democrats in light of this. 46 incumbent MPs re-contesting their seats were
still defeated in 2015. While the 9 first time incumbents experienced a much
smaller decline in their vote share, it was not enough to save any of them from
losing their seats. This was largely due to the small majorities each of the
first time incumbents was defending, meaning that even a modest swing away from
the party in these seats would not have saved these MPs from defeat. The party
does still enjoy an incumbency effect, which most likely prevented an entire
parliamentary wipe out, but it was nevertheless unable to prevent a ruinous
election night for the Liberal Democrats.
Conclusion
Altogether, the incumbency
effect remained strong for the Liberal Democrats, with evidence showing that in
held seats the party experienced a lower swing where they had an incumbent
re-standing, particularly a first time incumbent. Furthermore, having an
incumbent MP recontesting the seat made a difference in minimising the swing
regardless of how strong the party was previously, contrary to the theory that
the party was simply losing most where it was strongest. This seems to
contradict Stephen Fisher’s prediction, based on BES data, that the party would
do about “10 points worse in their own seats
compared with where they were previously
second”. This almost implied that the Lib Dems would enjoy no incumbency
boost whatsoever, when the results seem to indicate a ‘boost’ of between 3.7%
and 18.6%, depending on the party’s 2010 vote share in the constituency. But
this incumbency effect was not enough to stave off disaster in May.
Notes
Joshua Townsley is a PhD
researcher at the University of Kent, specialising in political campaigning and
electoral politics. For more information, see here: https://www.kent.ac.uk/politics/staff/assistant-lecturers/townsley.html
For more analysis of effect
of incumbency, see: Smith, T. (2012). Are You Sitting Comfortably?
Estimating Incumbency Advantage in the UK: 1983-2010. Electoral Studies, 32(1),
pp. 167-173.
