Thinking The 2023 UK Local Elections In Real Time


Are You Sitting Comfortably? Then I'll Begin

1:19am - 5th May 2023 

Local elections are always a fascinating part of the electoral cycle in the United Kingdom. Some may argue that local elections are just ‘not important’. Although local authorities do not control national policy, the day-to-day oversight of the public services that make our lives more liveable is within their purview. Whether we be in a Unitary Authority, Metropolitan Borough, District Council or County Council in England, Scotland, Northern Ireland or Wales, and whether our concern be local transport, social care, schooling, housing, planning, or waste collection - local government sits at the heart of it all. Therefore, just because the machinations of local government are closer to home and may feel increasingly mundane or common-place, this emphatically does not mean that the process of selecting representative decision-making agents on the local level holds any less intrinsic significance to one being able to live a liveable life than that for the national assembly. The salience of domestic policy on the national level may always and forever be wide reaching. However, the public odour of uncollected waste, the lack of available bus routes, ineffective social care, crumbling public services, absent childcare facilities, flooded roads, or a deficiency of available housing undermines our potential to simply live a liveable life both immediately and markedly. To have, decide upon and provide a liveable life for all – that is why local elections are significant.

Aside from the manner in which Local Elections effect ones life individually, Local Elections can give us an indicator of how the public feels about the state of British Politics as a whole. Nonetheless, it is important we are aware that Local Elections are not a good indicator for nation-wide or general elections. Although they can be useful for taking the pulse of the electorate, people vote differently at different levels of elections, wherein local issues do naturally play a part in local elections and national issues at national elections.[1]

This being said; however, the effect of local elections can be tectonic. For instance, in their ‘British General Elections Since 1964: Diversity, Dealignment and Disillusion’, David Denver and Mark Garnett highlight the significance of UKIP gains in the 2014 Local and European Parliamentary Elections to the decision made by David Cameron in promising a referendum on Britain’s membership in the European Union.[2] This is but just one example of the manner in which Local elections can affect the milieu of national policy and policy delivery. Equally, they can indicate the rise in popularity of the leading national opposition party prior to a victory on the parliamentary level; as happened for the Conservatives’ Vote Share in the change between 2008 (34%) and 2009 (44.4%) – the year before they came into power in coalition with the Liberal Democrats.[3] Therefore, it is by the manifestation of their effect that the outcome of Local Elections can have an influence on the national level.

Alongside this, amidst the evidence presented by Edward Fieldhouse (et.al.) in their ‘Electoral Shocks: The Volatile Voter in a Turbulent World’, we cannot ignore the fact that the British electorate is comprised of so-called ‘volatile voters’ on an aggregate level, irrespective of electoral level. Building on the work of Ascher and Tarrow, for Fieldhouse (et.al.) this should be understood as “the net change within the electoral party system resulting from vote transfers”.[4] This is significant as we can see that the twenty-first century has, so far at least, seen the British voter unanchored from factors influencing traditional voting alignment, observed in the so-called contemporary class-dealignment phenomenon. This implies that our political system is now far more susceptible to events and ‘electoral shocks’ that may have seemed unthinkable in the past. The collapse of the so-called ‘Red Wall’ in 2019 is an excellent illustration of such a phenomenon.[5] Therefore, at this juncture, it is always possible that the British electorate can defy expectation and either: (a) uphold the status-quo, or, (b) engage in some manner of unprecedented tectonic shift.

So, what are we looking at today? On the 4th May 2023, 230 councils are undergoing elections – as is the Northern Irish Assembly. Of these 230, 94 are electing one-third of their council seats and 136 are electing the whole set. This means that 8,058 seats will be fought for on the local level in the UK this year. It might be tempting to compare this election with last years, statistically speaking, given that it was the most recent. Nonetheless, we cannot forget that Local Elections work in cycles, i.e. the councils electing all of their seats this year will not have been those that elected all of their seats last year. Subsequently, it is worth reminding the reader that the year we should be comparing to is the last one that was on the same cycle – and that is 2019.

Using the ‘bible of local elections’, The ‘Local Elections Handbook 2019’ by Colin Rallings and Michael Thrasher, the total state of the parties was as follows:

 

Con

Lab

LD

Other

Nat

NOC

Total

Councillors

7507

6332

2531

2655

622

-

19,647

Councils Controlled

144

98

23

11

1

118

395

Vote Share (%)

31.1

26.5

16.9

25.6

-

-

100

Figure 1 – Councillors, Council Control and Vote Share (%) after the 2019 Local Elections[6]

These are some really significant figures to keep in mind, especially Vote Share (%). As has already been discussed elsewhere on the poLit blog, this is an election where the context is one of expected unpopularity for the conservatives. As we go through the night, I will be live blogging some statistical information as the first results come in. The purpose of this will be to lay out some statistical intrigues and predictions concerning: (a) Turnout at Local Elections, (b) Vote Share (%), (c) Councillor and Council Control, and lastly (d) Ward control. This will all occur before turning over to some Kirklees Council specifics as poLit live blogs from the count in Huddersfield. This particular council will be explored as a case study because  it was the heart of the class-dealignment phenomenon on the constituency level in 2019 – specifically with Working Class dealignment in Colne Valley.[7] Thus, it will be the topic of investigation as to whether or not if the results of this Local Election will indicate a reversal of that very phenomenon in the ‘Red Wall’, and the way that could adapt national policy as the potential epicentre of that very indication. Let’s turn, fittingly, to Turnout.


Turnout – What should we expect?

3:35am - 5th May 2023

Ok let’s turn out for turnout. Turnout is always significant because it’s a good measure of political participation and whether or not our polity is experiencing a democratic deficit in such a manner. Turnout can have dramatic effects on legitimacy, which may be defined as the right to govern. There are a number of ways in which we can ‘operationalise’ or measure turnout. Operationalisation is always important as it can change the way one observes a phenomena. For example, one can operationalises temperature by measuring degrees with a thermometer or by measuring the comfortability of a space in its effect on one’s constitution on a sliding scale from -5 to +5 – for instance. The way you operationalise something changes how you see the phenomenon itself.

Operationalising Turnout is no different. Simply put, turnout is the percentage of the eligible electorate that turnout out to vote – often reported in this manner as Voting Eligible Population Turnout (VEP).[1] It would be nice if at Local Elections in the UK, it was this simple. As local elections are conducted by local authorities turnout is reported by ward and then aggregated onto the council level. For this investigation I have chosen to adopt focussing on cumulatively reported average turnout for all authorities, as this is more than likely to be the one reported on. This has been chosen despite the fact that aggregate level turnout allows for a cancelling out of individual level idiosyncrasies.[2] Also, in presentation of this data, I have chosen to exclude those years with May general elections, as this mystifies applicability to this election.

What does the data tell us about trends and probabilities that can inform us about 2023. Let’s begin with the long term data, which is observable below in Figure 2.

Year

Turnout

1999

34.4

2000

29.1

2002

32.8

2003

35.2

2004

40.2

2006

36.3

2007

36.3

2008

36

2009

39.2

2011

41.8

2012

31.9

2013

30.9

2014

36.1

2016

34.1

2017

35.7

2018

38.9

2019

33.3

2021

36.1

2022

34.4

μ

35.41

σ

3.0850

R

0.1076

r2

0.0116

t

0.4461

p

0.66115

ta

0.0477

pa

0.96250

Figure 2a – A Long-Term view of Turnout (%) for Local Elections in the UK (1999-2022)

Figure 2b - A Long-Term view of Turnout (%) for Local Elections in the UK (1999-2022) in Graphic form.

What can we learn from this. First of all, the average (μ) turnout in this view has been 35.41%, retuning a standard deviation (σ) of 3.085. The highest point reached was in 2011, wherein 41.8% of the electorate turned out, still below an absolute majority of the number of eligible voting public. The lowest depth reached was in 2000 where only 29.1% turned out to vote.

 This latter figure is the one to recall, as fears turnout will be negatively affected by (a) voter fatigue and (b) newly introduced Voter ID legislation is widespread, with reports across the country repeating these concerns.[3] The data will speak for itself, but for the time being at least, there is plenty of data that we can compare from US elections that show Voter ID does not necessarily have to negatively affect turnout, even if it may indeed have an influence in other ways.[4] Saying this, we can remain sceptical of such a comparison, especially given normative distinctions between political cultures in the UK and the US.

Returning to the data in Figure 2, there is however a lack of statistical evidence of any kind of correlation or trend with time. Therefore, using a normal distribution curve, statistically speaking, what is the probability that turnout will go up, down, break ceilings or smash through basements? There is only a 2% chance that turnout will drop lower than the 2000 rock-bottom point of 29.1%. Nonetheless, there is already indicators coming in that turnout will be lower than its 2019 counterpart. For instance, Hull City Council declared that their turnout sat at 22.01%, a decrease of some 1.4%.[5] Thus as the night goes on, we may find this 2 in 100 probability slowly becomes more of a realistic outcome.

Inversely, what if this is just fear, scepticism and cynicism rolled together? What is the likelihood that turnout will break through the ceiling? There is only a 1.9% chance that the 41.8% high of 2011 will be breached. Consequently, following on from this, the statistical likelihood that turnout will fall at some point between these two polls returns a 96% probability. How can we close this 12-point aperture and find a smaller window of probability? The answer – look at the short-term view with this long-term view in mind.

Year

Turnout

2016

34.1

2017

35.7

2018

38.9

2019

33.3

2021

36.1

2022

34.4

μ

35.42

σ

1.8225

R

-0.0771

r2

0.0059

t

-0.1547

p

0.88455

ta

0.0119

pa

0.99108

Figure 3 - A Short-Term view of Turnout (%) for Local Elections in the UK (2016-2022)

Figure 3 lays out the short-term view of average turnout (reported)(%) at UK Local Elections across all authorities, excluding years with May general elections from 2016-2022. The average (μ) turnout in this view fell at 35.42%, not too distinct from the long-term view laid out above, retuning a standard deviation (σ) of 1.823 and so this tells us the data is more closely clustered than the long-term figures. The highest point reached was in 2018, returning in 38.9% and the lowest occurred in 2019 where 33.3% turned out to vote, less than -1σ from the mean in either a long-term or short-term overview.

Much like the case above in Figure 2, there is no correlation or concrete level of determinability with the passage of time. What about closing that aperture however? How can we turn that 12-point gap of 96% probability into something a bit more focussed? Firstly, before doing precisely this, the notion that turnout will remain above the 2019 low – important because this is the election that we are cyclically comparing to in 2023 – occurs roughly 88 times out of 100; 12 in 100 that it will fall below 33.3%.[6] Interestingly, there is only a 3% chance that turnout will rise beyond 38.9%. If we close our scope of focus slightly, we can see that the likelihood of turnout in 2023 resting between the highest (38.9%) and lowest points (33.3%) of this short-term view sits at 85%.[7]

An increase is statistically implied in this data, although as discussed above this should be taken with a shaker of salt. In both cases there is only a minimal correlation coefficient and coefficient of determination. Nonetheless, statistically, probability favours: (a) national turnout not decreasing below 29.1%, (b) an increase above the short-term high of 38.9% occurs in only 3 times in 100, and so (c) it is not unreasonable to expect national average turnout to rest somewhere between 33.3% and 38.9% of which there is an 85% probability.

Heed this warning however – these probabilities are only projections given the past data we have. As discussed above, voter fatigue and new voter ID legislation could collapse turnout entirely. Equally, the context of unpopularity surrounding the Conservatives may compound this, leading to the observation that turnout will vary. The expectation perhaps should be that the safer the council is for the Tories, the lesser the turnout, with the inverse holding, i.e. the more marginal the seat, the higher the turnout – especially if they are Conservative held currently. Perhaps the context of this election will simply deem the statistical likelihood and history of turnout data irrelevant this time around.

A Slice of The Pie? On Vote Share (%) 

9:30 am - 5th May 2023 - Kirklees Election Centre.

Today, in the daylight, the final three-quarters of seats of the 230 holding elections today will be counted for this election. I shall be blogging about Kirklees’ election as the day goes on, but for now, lets finish with the national level, speaking firstly about vote share (%) and any relationships with the passage of time and other variables that are significant.[1] Share of the popular vote, or Vote Share (%), concerns the percentage of valid votes won by the different parties at the election. Below, in Figures 4, Vote Share (%) is laid out in two forms: a long-term view and a short-term view.

Year

Con

Lab

LD

Other

Green

UKIP

Lab/Con

1999

27.6

33.8

21.5

16.3

0.8

0

61.4

2000

37.2

32.4

24.3

4.9

1.1

0.1

69.6

2001

39.4

30.9

25.1

3.5

1

0.1

70.3

2002

32.2

35.3

23.7

5.8

2.7

0.3

67.5

2003

31.5

27.9

24.2

14.4

1.6

0.4

59.4

2004

29.6

28.6

23.9

13.5

3

1.4

58.2

2005

39.8

25

27.6

4.3

2.3

1

64.8

2006

34.5

28

23.4

8.6

4.6

0.9

62.5

2007

38

22.2

23.6

11.6

3

1.6

60.2

2008

34

26

21.3

14.7

2.6

1.4

60

2009

44.4

12.7

24.8

8.9

4.6

4.6

57.1

2010

32.5

31.9

23.8

6.7

3.5

1.6

64.4

2011

37.8

31.6

16.1

8.5

3.6

2.4

69.4

2012

25.3

42

13.2

12

3.7

3.8

67.3

2013

34.4

21.2

13.9

7.1

3.5

19.9

55.6

2014

25.9

35.8

11

5

6.6

15.7

61.7

2015

35.8

28.4

10.3

6.2

6.5

12.8

64.2

2016

27

38.5

12.8

5.1

6

10.6

65.5

2017

43

21.3

16.5

11

4

4.2

64.3

2018

31.8

41.2

14

5

6.7

1.3

73

2019

31.1

26.5

16.9

11.7

9.3

4.5

57.6

2021

40.5

27

15.3

8

9.1

0.1

67.5

2022

27.7

37.6

15.5

8.7

10.5

0

65.3

μ

33.96

29.82

19.25

8.76

4.36

3.86

63.77

σ

5.3316

6.8399

5.2233

3.6462

2.6464

5.3920

4.5634

R

-0.0816

0.1029

-0.7687

-0.1486

0.8991

0.3540

0.0589

r2

0.0067

0.0106

0.5909

0.0221

0.8084

0.1253

0.0035

t

-0.3753

0.4740

-5.5074

-0.6885

9.4119

1.7348

0.2702

p

0.71122

0.64039

0.00002

0.49866

0.00000001

0.09743

0.78967

ta

0.0305

0.0485

3.3564

0.1012

6.2927

0.5790

0.0159

pa

0.97593

0.96177

0.00299

0.92036

0.000003

0.56876

0.98749

Figure 4a – Vote Share (%) For Different Parties at UK Local Elections (1999-2022)[2]


Figure 4b – Vote Share (%) For Different Parties at UK Local Elections (1999-2022) in graphic form

 

Year

Con

Lab

LD

Other

Green

UKIP

Lab/Con

2016

27

38.5

12.8

5.1

6

10.6

65.5

2017

43

21.3

16.5

11

4

4.2

64.3

2018

31.8

41.2

14

5

6.7

1.3

73

2019

31.1

26.5

16.9

11.7

9.3

4.5

57.6

2021

40.5

27

15.3

8

9.1

0.1

67.5

2022

27.7

37.6

15.5

8.7

10.5

0

65.3

μ

33.52

32.02

15.17

8.25

7.60

3.45

65.53

σ

6.1072

7.3936

1.4067

2.5889

2.2316

3.6637

4.5485

R

-0.0153

-0.0052

0.3847

0.2451

0.8688

-0.8142

-0.0289

r2

0.0002

0.0000

0.1480

0.0601

0.7548

0.6629

0.0008

t

-0.0305

-0.0103

0.8335

0.5055

3.5087

-2.8049

-0.0578

p

0.97710

0.99227

0.45142

0.63977

0.02470

0.04857

0.95670

ta

0.0005

0.0001

0.2993

0.1203

2.3012

1.7710

0.0017

pa

0.99965

0.99996

0.77963

0.91003

0.08283

0.15127

0.99875

Figure 4c – Vote Share (%) For Different Parties at UK Local Elections (2016-2022)

What can we see? First and foremost, let’s limit scope to Figure 4a and 4b. Between 1999 and 2022, national vote share at elections is highest for the Conservatives, who returned 33.96% on average in the period. Labour follows with 29.86%, then the Liberal Democrats with 19.25%, then the Greens with 4.36%, UKIP with 3.85%, and finally the ‘Others’ with 8.76%.

There are only two statistically significant correlations with the passage of time. The First concerns the decrease of the Liberal Democrat’s share over time. It seems that since their best performance in 2005 (27.6%), the Liberal Democrats have been unable to recover such popularity – although the same could be said of the national level for parliamentary elections to some degree. Interestingly, we should, statistically speaking, expect the Liberal Democrat popular vote to decrease even further at this election – decreasing just twice in every 10,000. As a result of this statistical likelihood, if this transpires as not to be the case, then it will make this election all the more fascinating, especially if their final share of the vote is higher than 24.4% - an increase of over +1σ in the face of this.

The second concerns the rise of the Green Party. Although holding a period average of just 4.36%, in 2019 the Greens returned 9.3% and 10.5% last year, in 2022. To have risen from 0.8% and 1.1% in 1999 and 2000 respectively has been a feat by any standard. We should, absolutely expect the Green Party’s vote share to rise this year. Indeed, the odds that the Green Party share decreases this year sits at only 3 in 100,000.

Now Turning to Figure 4c, we can see that, in the short term, the order in which the parties fall remains the same, however their averages differ. The Conservatives are at the top with 33.52%, Labour with 32.02%, Lib Dems with 15.17%, the Greens with 7.6%, UKIP with 3.45% and the Others with 8.25%.

There are also only two statistically significant correlations here. One concerns the rise in votes over the period for the Green Party, and the other the near total wipe-out of UKIP. Speaking of the latter, in 2016 UKIP returned 10.6% of the popular vote. By 2022, this was reduced to 0. The likelihood of UKIP gaining anything more than 0.1% of the vote share nationally is very, very slim indeed. The other correlation concerns the Green Party, wherein a period wide positive correlation is clear; a reduction to less than 6% occurs only twice in 100 times.

Like always, however, these are just the overarching trends, and they tell us very little about the two major players in the UK’s Two-and-a-Half Party System. Thus, although they do reveal quite a bit about the minor parties, they tell us little about Labour or The Conservatives beyond the already known statistic that they magnetise the greatest number of votes. Equally, context could change everything and once again we could see realignment or even the beginnings of an ’Electoral Shock’ in the making.



[1] In the use of the phrase ‘significant’, what I mean is ‘statistically significant’, wherein a correlation of determination or correlation coefficient returns a p-value of less than 0.05.

[2] Rallings and Thrasher throughout



[1] A great reference point for this discourse is: Thomas Holbrook and Brianne Heidbreder (2010) ‘Does Measurement Matter? The Case of VAP and VEP in Models of Voter Turnout in the United States’, State Politics and Policy Quarterly, 10(2), pp. 157-179; Michael P. McDonald (2022) ‘What is the voting-age population (VAP) and the voting-eligible population (VEP)?’, United States Elections Project, https://www.electproject.org/election-data/faq/ denominator [Accessed 28th October 2022].

[2] Benny Geys. (2006). ‘Explaining voter turnout: A review of aggregate-level research’. Electoral Studies. 25(4): 637-663; J.G. Matsusaka and F. Palda. (1999). ‘Voter turnout: How much can we explain?’. Public Choice, 98(4): 431–446.

[3] Joanna Taylor. (4th May 2023). ‘Local elections 2023: Voter ID effect on turnout to be monitored at polling stations ‘. Peterborough Telegraph. https://www.peterboroughtoday.co.uk/news/politics/local-elections-2023-voter-id-effect-on-turnout-to-be-monitored-at-polling-stations-4130254. (Accessed 05/05/23); John Cooper. (4th May 2023). ‘Councillors across Derbyshire reflect upon voting turnout and how much influence regional results may have on national politics’. Derbyshire Times. https://www.derbyshiretimes.co.uk/news/politics/ council/councillors-across-derbyshire-reflect-upon-voting-turnout-and-how-much-influence-regional-results-may-have-on-national-politics-4130994. (Accessed 05/05/23).

[4] Jason D. Mycoff, Michael W. Wagner, and David C. Wilson (2009) ‘The Empirical Effects of Voter-ID Laws: Present or Absent?’. PS: Political Science and Politics, 42(1): 121–126; Rene R. Rocha and Tetsuya Matsubayashi (2014) ‘The Politics of Race and Voter ID Laws in the States: The Return of Jim Crow?’. Political Research Quarterly, 67(3): 666–679; Rick Hassen (2020) Election Meltdown: Dirty Tricks, Distrust, and the Threat to American Democracy. New Haven, CT: Yale University Press, p. 41.

[5] Hull City Council Twitter Post: https://twitter.com/Hullccnews/status/1654264376252145666 (Accessed 05/05/23).

[6] Where p = 0.8776 and p = 0.1224.

[7] Where p = 0.8495.


[1] Anthony Wells (1 May 2017) ‘Why local elections are not useful indicators of national votes’. YouGov. https://yougov.co.uk/topics/politics/articles-reports/2017/05/01/why-local-elections-are-not-useful-indicators-nati. (Accessed 05/05/2023).

[2] David Denver and Mark Garnett. (2021). British General Elections Since 1964: Diversity, Dealignment and Disillusion. Second Edition. Oxford: OUP. pp. 176-177.  

[3] See Figure below detailing Vote Share (%).

[4] Fieldhouse, E., Green, J., Evans, G., Mellon, J., Prosser, C., Schmitt, H., and Van Der Eijk, C. (2020) Electoral Shocks: The Volatile Voter in a Turbulent World. Oxford: OUP, p.9; Ascher, W. and Tarrow, S. (1975) ‘The Stability of Communist Electorates: Evidence from a Longitudinal Analysis of French and Italian Aggregate Data’. American Journal of Political Science, 19(3): 475-499.

[5] For a discussion and definition of the ‘Red Wall’, see: Kanagasooriam J, Simon E. (2021). ‘Red Wall: The Definitive Description’. Political Insight, 12(3): 8-11.

[6] Colin Rallings and Michael Thrasher. (2019). Local Elections Handbook 2019. Plymouth: Local Elections Centre. pp. viii-ix.

[7] Gavin Hart. (2022). ‘Working-class conservative voters in 2019: voices from a valley in northern England’. British Politics, 17(3): 298–314.