Psephology: if Calder Valley had voted for an MP in 2014

Post length: 3,214 words, about 14 and a half minutes.

Despite a swing to the Labour Party, if the pattern of voting seen when comparing the 2010 local election results (a poll taken at the same time as the last UK general election) to the 2014 local election results is applied to the general election results, the bellweather seat of Calder Valley would remain a Conservative seat.

There are of course a number of caveats when making a comparison of this type, but it seems that had the general election been held on 22nd May 2014 then the Conservative candidate would have been returned to parliament, albeit with a substantially reduced majority. Additionally, it appears that all of the indicators agree with this projected outcome. What follows is my long-form statistical analysis of the situation. The calculations from which I draw my conclusions are available for download at the end. If you do not want to read the full analysis then I suggest you jump straight to the conclusion.

Before I get into the numbers, a little background on the UK Parliament constituency of Calder Valley. Created in 1983 the constituency has historically been a bellweather seat, albeit only having seen three MPs in this time. The constituency is made up of nine wards covering the south and west of the Metropolitan Borough of Calderdale, one of two Parliamentary seats in the district. While not strictly a key marginal seat, in the 2010 general election it was the closest three-way marginal in the north of England. More stats on the make up of the constituency can be seen on UK Polling Report. It therefore presents an interesting case for psephological study.

One final point before I begin: I must declare a couple of interests. I lived in the Ryburn ward of the constituency for 10 years, between the ages of 8 and 18. I was involved with electioneering during that time, including during the 1997 general election. I am also old friends with the Labour Party candidate for the 2015 general election, although no longer living in the area had nothing to do with his selection or ongoing campaign. I have attempted to preclude any bias from the following study, my methods are available for inspection at the end of the article, and I welcome any comment, correction and criticism in the comments section below.

This analysis uses data from the election of local councilors in 2010 (the same day as the general election), 2012 and 2014 (the same day as the European Parliament election) and attempts to apply the outcomes to predict a general election result, had the election taken place in May 2014, one year before it is scheduled.

2010: local government vs. national government results

The national results of the 2010 general election lead to a coalition government between the Conservative Party and the Liberal Democrats, the Conservatives being the much larger partner. This result was reflected to some extent at the local elections in the nine council wards in Calder Valley: the Conservatives won two seats, the Liberal Democrats won five seats and the remaining two seats went to independent candidates. It’s very important to note, however, that while the parties involved were the same as those who went on to form a government, the split of seats themselves was reversed from the national result — the Liberal Democrats gaining 2.5 times more actual seats than the Conservatives. Conversely, the constituency taken as a whole on local results would have returned the Liberal Democrats in a very close second place to the Conservative Party, with the Labour Party training badly in third place. Therefore, although the parties winning council seats reflected the general balance of power in Westminster after the days following the general election, neither the actual split of local seats, nor the split of real votes cast across all nine wards reflected the actual party split of seats in the House of Commons.

On the other hand the split of votes cast at the same time for the national representative is broadly in-line with the national result: Conservative 39.27% (36.1% nationally); Labour 26.89% (29% nationally); Lib Dem 25.10% (23% nationally). Turnout was marginally higher at 67.54% than the national 65.1% (and turnout in both national and local elections was roughly the same at 67.54% and 66.69% respectively). (All national stats from Wikipedia.)

This indicates that the electorate in Calder Valley don’t just vote along party lines — there is a clear difference in voting patterns in local and national elections. While this does lead to some muddying of the waters when applying a local result to a national one, given that both the total local vote and the actual national vote would have sent the same MP to Westminster (with a majority of just 368 rather than the actual majority of 6,431), the calculation is not completely without basis.

Interim local elections

For brevity I have only studied the 2012 local government elections, the mid point in the four years of my study.┬áIt’s worth noting that, due to the nature of local politics, the outcome of local elections relies not only on the party represented, but the candidate themselves. I have not taken into account standing councilors seeking re-election, nor attempted to gauge the influence local issues may have had from one election to another while comparing 2010 to 2012.

In these elections both the Conservatives and Liberal Democrats won three seats each, while the remaining three seats went to Labour candidates. Seat wise, a perfect 3-way tie. Two wards — Luddendenfoot and Todmorden — changed from electing Liberal Democrat councilors to electing Labour councilors. Labour won no seats from the Conservatives.

In the ward won by the Conservative candidate in both 2010 and 2012 there is a noticeable fall in majority. In Rastrick ward the majority was more than halved from 20.81% of turnout to just 10.07% of turnout (1,136 votes to 265 votes respectively). The Conservative party performed marginally better in Ryburn where the majority fell from 18.80% to 11.08%. It is worth noting that in Rastrick the Liberal Democrats, who came second in the 2010 elections in this ward, did not field a candidate in 2012.

The story is somewhat different for the Liberal Democrats. Not only did the Labour Party pick up two seats which the Liberal Democrats had won back in 2010, but two of the wards returning a Liberal Democrat councilor for a second time returned them with a much reduced majority. Both Calder and Elland elected Liberal Democrat councilors with single figure percentage majorities, the majority in Calder being reduced by 14.39% to just 3.54%, a majority of just 145 votes. Greetland and Stainland, on the other hand, elected a Liberal Democrat with a marginally increased majority of 13.68% up from 12.45%.

Taking all votes cast across all nine wards into account and applying them as you would in a general election (i.e. simulating the election of an MP using the votes cast for each party in the local government elections) gives a different result to applying the same process to the 2010 local election results. While the 2010 local election votes would have sent a Conservative representative to the UK Parliament, in 2012 a Labour MP would have been elected. Indeed using this simplistic calculation the majority the Labour MP would have taken as their mandate would have been 3.26% of the total turnout — 876 votes — in comparison to the very slim 0.71% — 368 votes — majority the Conservative MP would have taken in 2010.

Swing in 2012

Of course, as highlighted previously, applying such a simplistic calculation to convert local election votes to general election votes is significantly flawed, not least because it is widely recognised that the electorate uses interim elections as a means of expressing a protest to the incumbent national government. Therefore I have gone on to calculate swing using the two widely accepted methods from the 2010 – 2012 local elections to the general election result of 2010. While this still doesn’t account for local issues and personalities, it should go so some way to accounting for the political movements as a nation.

The first method of calculating swing is known as Butler (or conventional) swing. While it is used to express a shift in voting behavior between just two parties, it does take into account all other votes cast in the given elections. The second method is known as Steed (or two-party) swing and excludes all votes for parties not included in the comparison. They can produce very different results.

Of interest is the conventional swing from the Conservative Party to the Labour Party on a ward-by-ward basis, even though this is only of limited use in a 3-way marginal election. Only one ward, Hipperholme & Lightcliffe, had a positive swing to the Conservatives (1.09%), while all of the other wards have a swing towards Labour. Of specific interest is Calder where the support for the Conservative Party seeming collapsed — from 23.53% down to 11.65% — and support for the Labour Party climbed more than 12 and a half percentage points to 35.52% (the ward still elected a Liberal Democrat councilor).

When viewing the constituency as a whole both calculations of swing show a swing away from the Conservative Party and towards the Labour Party between 2010 and 2012. Steed swing shows the largest swing of -12.54% while conventional swing shows a -7.06% swing. Applying either of these measures of swing to the 2010 general election (assuming the turnout is the same as the 2010 turnout, as it would be unfair to take the turnout of a standalone local election and use that as the basis of a general election turnout) brings about the same overall result: the constituency would return a Labour MP. In the case of conventional swing the majority would be 900 votes (a mere 1.73%), while Steed swing puts the majority at 2,190, or 6.37%.

A majority of 1.73% is tiny and, given the flaws discussed above, I would normally discount it. To use (mostly) American electoral terminology it would be “too close to call.” However, when sat beside two other calculations which indicate that Calder Valley would have sent a Labour MP to Parliament, it does add weight to the overall impression that, had the election have been held in 2012, just two years after the formation of the Conservative / Liberal Democrat coallition, Calder Valley would have been represented by a Labour MP.

2014: local elections one year out

The local elections to Calderdale Metropolitan Borough Council took place on the same day at the 2014 European Parliament elections. Turnout was 3.16 percentage points higher in 2014 than in 2012 although this could have been, at least in part, down to the rise in profile of the UK Independence Party (who took a lead in the European elections nationwide), and European issues as a whole.

Given the four year term of office for local council seats, while it is somewhat useful to compare 2014 with the results from 2012, it is arguably more telling to compare the 2014 results with those of 2010. Of the nine seats contested, four contests included the incumbent councilor. In two cases (Hipperholm & Lightcliffe and Rastrick) the incumbent was returned to the seat; in another two (Brighouse and Elland) the incumbent was defeated. Of these Brighouse was the a conservative gain from an independent councillor, and Elland was the closest result; the Labour Party candidate defeating the incumbent Liberal Democat by only 48 votes (1.68% of turnout).

In total the Conservative Party won four seats, one up from the 2012 local election (gaining Luddendenfoot and Brighouse, but losing Hipperholme & Lightcliffe to an independent) and two up from the 2010 content, the Labour Party won three seats equaling their 2012 result (although with only one seat being in the same ward as in 2012), the Liberal Democrats won only one seat, down two from 2012 and down four from 2010 (and even then only winning the Greetland and Stainland ward by 60 votes). It’s notable that the collapse in support for the Liberal Democrats in Calder Valley seems to mirror the results across the majority of UK councils which held local elections at the same time in 2014 — the party lost 310 seats nationwide, losing control of two councils.

Comparing 2012 to 2014

As mentioned previously there is only limited value in comparing 2012 results to those of 2014 when an arguably more accurate comparison — using a number of the same candidates, and thus reducing (but not removing) the fluctuation based on local issues and personalities — exists. However this comparison is as valid as comparing 2010 to 2012, and thus I shall briefly do so.

As highlighted above, the Liberal Democrat support dropped off considerably between 2012 and 2014. The party won only one seat, and then only by 60 votes. In the constituency as a whole Liberal Democrat support dropped from 22.72% to 14.34%, a drop of 8.38 percentage points. This placed them 4th, training independents and others (excluding the Green Party) by more than 750 votes. In the single ward where the party won a seat the conventional swing away from the Liberal Democrats to the second placed Conservatives was 5.86% while in Calder, where Labour took the seat with a 730 majority, the swing away from the Liberal Democrats to the Labour Party was 13.02%.

Significantly the swing to Labour from the Conservatives seen between 2010 and 2012 was not repeated between 2012 and 2014. In fact, rather than one single ward seeing a swing towards the Conservative Party, just one single ward saw a swing away from the Conservative Party (perhaps coincidentally, due to the re-election of a sitting independent councilor, this was the same ward). Due to this the Conservative Party won an extra seat over their results in 2012.

Local results 2010 to 2014

The story is somewhat different when comparing the outcome of the 2014 local election to the 2010 results — our key indicator. Rather than the majority of the nine wards seeing a swing from the Labour to the Conservatives as is the case when comparing to 2012, the conventional swing from 2010 to 2014 in all but two wards still shows a swing in the opposite direction: from Labour to the Conservatives. The dilution of the swing is visible for example in Elland: one of the wards where the incumbent councilor was deposed in 2014, the swing from 2010 to 2012 was -7.51% (in favour of Lab) and between 2012 and 2014 was 3.56% (in favour of Con). This caused the swing from 2010 to 2014 to be -3.68% (in favour of Lab), almost exactly half the swing seen from 2010 to 2012.

Of the two seats gained by the Conservatives over the 2010 election one of them was previously held by the Liberal Democrats while the other was a gain from an independent candidate. This seems broadly in-line with the national result both for the local government election and the European election where support for the Liberal Democrats collapsed from a record high at the time of the 2010 general election. Likewise, but perhaps significantly, the seats won by the Labour Party in 2014 were all won by the Liberal Democrats in 2010; the Labour Party did not manage to win any seats previously won by the Conservative Party.

On a constituency level it’s noticeable the gains made by the independent / others between 2010 and 2014. Even though fewer independent councilors were elected in 2014 (one was re-elected, while one was deposed), the total share of the vote in this group jumped into double figures, increasing 7.64 percentage points to 16.96%. Support for the Green party also increased significantly from the 4.24% they achieved in 2010, the percentage of the electorate voting for the the Greens more than doubled to 9.28% in 2014. Support for the Liberal Democrats dropped, in line the the drop in elected councilors and the national result, 17.4 percentage points to 14.34%, while support for the Conservatives and Labour parties changed only marginally the former losing just 1.4 percentage points to 31.05% and the latter gaining only 5.91 percentage points to 27.50%.

Using the simplistic method of total local votes to determine the outcome of a general election should it have happened on 22nd May 2014, the result is the same as that of the 2010 local election figures — the constituency would have returned the Conservative MP to Westminster. Indeed the candidate would have been returned to the seat with a larger majority than had the election been decided on local government votes in 2010.

Constituency level swing

Both calculations of swing between 2010 and 2014 show the same dilution of the 2012 pro-Labour swing seen in all but one of the wards. Unlike the swing 2010 to 2012, the 2012 – 2014 swing goes towards the Conservative Party. While the Labour Party do manage to maintain the swing towards them over the full four years, due to the virtue of the 2012 – 2014 swing the margin is much smaller than that seen in 2012. Both measures of swing bear this out: the Butler swing giving a figure of -3.65% and the Steed swing giving a figure of -7.02%.

Applying these swing figures to the 2010 general election results gives the best indication of how the constituency might have voted if asked to select an MP in 2014. Assuming that the turnout of a general election is the same as in 2010, as with the projection made using the 2012 results, both projections would return a Conservative MP for Calder Valley by a slim majority. In an interesting, albeit insignificant switch, applying the 2014 swings to the 2010 general election result shows a larger projected majority using the conventional Butler swing than that of the Steed swing, unlike applying the 2012 swings to the same source numbers.

Conclusion

The constituency shows good adherence to the general national political mood, as expected. This is most obvious when looking at the Liberal Democrat vote — the large number of seats won in 2010, and the subsequent drop in their support is, broadly speaking, mirrored when looking at the election results across the country in the years following. While this was not the aim of my study, it does bear out the ‘bellweather’ label often applied to the constituency.

Having detailed three means of projecting the possible outcome at a general election (straight transposition of the local election results, application of Butler swing and application of Steed swing) and having considered the pitfalls associated with projecting from a local to a national level, I am confident to say that had a general election taken place in May of 2014 Calder Valley would have elected a Conservative MP to represent the constituency in the House of Commons.

While each of the projections does show a swing towards the Labour Party at the time of the 2014 election, and the majority held by the incumbent Conservative MP would have been significantly reduced, this swing would not have been significant enough for the Labour Party to win back control of the parliamentary seat. The Labour Party did, in fact, lose some ground to the Conservative Party between 2012 and 2014. Had the election been held in May of 2012 there is an indication that the seat would have been won by the Labour Party (although some methods suggest this would have been by a very narrow majority, easily within the margin for error).

Projecting forward another year from this point is, of course, impossible. The effect of a fourth party growing in popularity (as has been seen during the 2014 local and European election) has not been factored into this study. Additionally, ahead lies a year of campaigning and the effect of a national campaign cannot be discounted. Nor can the effect of local campaigning, issues and personalities, little of which have been accounted for in this study.

There is a real chance that the Calder Valley constituency will be a very close race during the 2015 general election and I believe that the vote-by-vote results will reflect very closely the makeup of the next parliament.

Statistical data can be downloaded here: Calder Valley Analysis (Open Document Format), Calder Valley Analysis (Excel format).

Posted on Tuesday 27th May, 2014 at 6:27 pm in Obiter dicta.
It was tagged with , , , , , , , .

Leave a comment