Professor Alistair Clark, Professor of Political Science, Newcastle University – written evidence (VID0004)

House of Lords Constitution Committee inquiry into voter ID

Introduction

1. My expertise is in electoral integrity and administration, with numerous published research articles and reports on these themes. I write in a personal capacity (http://www.ncl.ac.uk/gps/staff/profile/alistairclark.html#background).

Overview/Summary

2. The introduction of voter ID in the Elections Act 2022 has been a solution in search of a problem. It constitutes an additional administrative burden or obstacle to otherwise eligible voters casting their ballots. The main justification often provided to support its introduction – the need to secure the ballot because of voter personation – has negligible evidence to support it. Research I have conducted with Prof. Toby James (UEA) has consistently noted the biggest difficulty in polling stations pre-voter ID was people being turned away because they are not properly registered.[1]

3. Voter ID’s first use in major public elections, the 2023 local elections in England, was problematic, for reasons explained below. The main factor that prevented difficulties with those elections from developing into a wider democratic problem was that these were low turnout local elections. In a general election, turnouts will be much higher, magnifying the potentially negative effects of voter ID.

Voter Authority Certificates

4. Efforts to raise awareness of the voter ID requirements must deal with a paradox. The paradox is that voters tend only to pay attention to issues around electoral process close to deadlines, while, in order to ease administrative pressures, election administrators need voters to know about these requirements well in advance, and, where necessary, to have applied for voter authority certificates (VACs) as far before the deadline as possible. The voter registration website crashing in advance of the 2016 EU referendum registration deadline is a good example of what can go wrong when dealing with this paradox is not at the centre of decision making.

5. Government has communications channels open to it beyond those available to the Electoral Commission. These include mailings (for various purposes), advertising, and electronic channels such as email and social media. These government channels should also be utilised. This would give the Electoral Commission, assisted by DLUHC where appropriate, access to additional communications channels to inform voters of the voter ID requirement and possibility of getting a VAC if necessary.

6. Another way of raising awareness of the voter ID requirement and the possibility of obtaining a VAC, would be to empower other public services – local government most obviously – to promote awareness of the voter ID requirement, so that in most engagements with public services, voters are reminded on the need to act early if they do not have appropriate ID.

7. Given the relative imminence of the general election and also the local elections in May 2024, any publicity campaign needs to begin forthwith.

8. If, and where, it is not already, the voter ID requirement needs to be prominently displayed on polling cards. This may require legislative change. It should also include a reminder about VACs, although care will need to be taken that the VAC deadline is sufficiently after the registration deadline to give voters time to apply. Polling cards however would be difficult to use as identification because of the potential for them to be picked up by someone other than the voter in multiple occupancy flats and buildings.  

9. I was struck during the May 2023 local elections by several conversations I had with voters, observers and otherwise interested parties about the VAC. Almost universally, and to paraphrase somewhat, these conversations highlighted the fact that the VAC was ‘just a printed piece of paper’ which, as such, was quite flimsy for the weight of democratic responsibility that it enabled. It was clear to me from these discussions that people expected something more than this from the VAC. Northern Ireland has, for example, a Voter Identity Card (VIC).

10. This suggests a potential mismatch between voter expectations over VACs, and the system the government decided to fund and implement. It is difficult to know how this might have affected take-up given the anecdotal nature of these observations. However, if voters were getting something like NI’s VIC which they might be able to use for other identification verification purposes, then it is possible that VACs might see increased take-up.

11. I would recommend revisiting the issuing process of VACs as a consequence. While there will be inevitable cost / benefit implications and considerations, and there would be considerable time pressure on making any changes, it is imperative that voters are given every incentive and opportunity to cast their vote. A revised and more substantial VAC may be one way of improving this feature of the system and providing voters with additional incentives to engage with this issue early.

12. A further aspect that ought to be reviewed is the deadline for applying for VACs to ensure that this is fit for purpose. There is a delicate balance to be struck. This should be late enough that it allows voters who only realise close to the elections that they need such documentation to apply and have it successfully issued. It should not, however, be so late that it introduces unmanageable demands on the electoral registration administration across the country.   

Turnout and participation

13. In places, general election turnout will potentially be around double or more than that experienced in local elections. This will magnify the impact of voter ID on turnout considerably, although it is difficult to say with any degree of precision what the suppressive effect of voter ID on turnout and participation might be in the forthcoming general election and where this might happen. The Electoral Commission’s report on the 2023 local elections estimated this at 7% of non-voters.[2] 

14. Even prior to voter ID, general election turnout shows considerable variation between different age groups, social classes, ethnicities, housing tenure, educational status (i.e. students, those with no qualifications) and urban, suburban and rural areas for example. It is not unreasonable to think that the communities already demonstrating lower turnout, are those most likely to be negatively affected by voter ID. Such communities also consistently show lower levels of electoral registration in the first place.[3]

15. There are additional groups which might be negatively affected by the voter ID rules. The first are older voters, who traditionally tend to turnout to vote, but who may not travel widely and as a consequence may not have ID such as passports. Many older voters also do not have internet access or skills, rendering the predominant requirement to apply for a VAC online a potential form of digital exclusion. Non-online application processes ought to be made more prominent.[4]

16. Evidence from a poll worker study during the 2023 local elections that I conducted with Prof. Toby James (UEA) provided evidence that other groups were potentially impacted. One poll worker highlighted potential impacts on female voters, for example, in the following quotation:

(That) ‘Women were turned away because they got married and changed their names then their ID and register names were different is gender discrimination!! I'm quite upset that I've turned voters away and particularly (since that has been) discriminating against women.[5]

17. Another poll worker pointed out the potential to cause offence to otherwise engaged and dutiful voters:

Be mindful that not all ID is recognizable for the person especially for expired Passports and Driving license's that show their picture to have been taken decades previously. It is easy to offend elderly voters whose ID is not recognizable with so many years having passed yet they do provide their address and Poll Card to confirm they are the voter.

18. There was also significant concern from charities representing blind and disabled voters about the effects of voter ID. The Electoral Commission found that voter ID had introduced further obstacles with some disabled people finding it harder to cast their vote. Concerns about disabled and blind voters therefore remain, and will be magnified in a higher turnout general election.[6]

19. The fundamental difficulty with estimating the overall impact of voter ID on turnout is less with counting those that go to vote and are turned away, but that of capturing those who simply decide not to attend the polling station in the first place. Such non-participation is already at quite high levels. Any attempts to distinguish the impact of voter ID would need to separate that from more general aspects (e.g. apathy, disillusion, work and various other factors) which lead to non-voting.[7]

20. This is likely to need an election study of individual potential voters, or a post-election opinion poll, to estimate the level of the problem. A project like the British Election Study would seem to have an appropriate methodology to reach non-voters, although it would take time for any results to be known. This could not be done with an exit poll. These only capture those voters who have attended the polling station. They could not capture potential voters who stayed away from the polling station because of voter ID.                  

Electoral Administration

Poll Workers, Recruitment and Experience of Voter ID

21. There are potentially significant impacts upon the ability of councils to conduct the next general election under voter ID and the provisions of the Elections Act 2022. I have conducted, along with Prof. Toby James (UEA), several surveys of polling station workers, including under voter ID circumstances in the 2023 local elections.[8] Relevant findings from these surveys are summarised here.

22. Even prior to voter ID, councils found it difficult to recruit sufficient staff to work in polling stations.[9] The vast majority who do so are volunteers. Although they do receive a small amount of pay for a long (roughly 6am-1030pm) work day, nevertheless, most give up their time to assist their fellow citizens in exercising their democratic rights by taking up these temporary but highly responsible positions.

23. Evidence from pre-voter ID poll worker studies in England has suggested that over half took time off work to serve voters on polling day, while around a third were retired and their average age was 53. More than three fifths (63.2%) of polling station workers were women. Most who worked at other jobs were in a routine technical or administrative role.

24. Our research suggests three main motivations for people volunteering to work on polling day. The first was financial. Although most receive only a small amount of pay for working on polling day (typically around minimum wage for a polling clerk), in the 2015 general election, 89.5% said making some extra money was either somewhat or very important.

25. The second motivation came from social incentives, such as being asked to do so by an acquaintance in a local group, being with people who share their ideals, receiving some form of recognition, and being with like-minded people. The third motivation came from more purposive incentives. These included: experiencing the democratic process, learning more about the political process, thinking it their duty as a citizen and being the kind of person who does their share.[10]

26. I recommend that in designing efforts to recruit polling station workers in the aftermath of voter ID, that each of these three sets of incentives are included in recruitment campaigns. Social and more idealistically motivated reasons have substantial impacts on why poll workers volunteer. With polling day pay, this may necessitate councils having to increase what they pay poll workers. Although this may be difficult at a time of council budget squeezes, this should be the sort of budgetary item that DLUHC covers in the Maximum Recoverable Amount (MRA) they allow councils to recover for general election costs.

27. There is a potential virtuous circle here. Once volunteers have experienced working at elections, they are likely to do so again in future, thereby underlining the social and purposive motivations mentioned above. In 2023, 81.5% had worked at elections previously, while the remainder were doing so for the first time. In the aftermath of the 2023 local elections, over 90% of poll workers said they would either definitely or probably be willing to do so again.[11]     

28. There is reason to be concerned about the training that temporary polling station staff receive in advance of polling day. Prior to the introduction of voter ID, our polling station staff surveys have consistently shown that between 15-19% thought election law was already too complex to understand quickly and easily. The introduction of voter ID introduces further complications and complexities for polling staff to deal with, including the ability to identify between numerous different types of identification.

29. In common with other training and learning during the pandemic, training of poll workers for the 2021 elections was delivered almost exclusively online. Previously, it had been held in person, with some training including aids such as role-plays and mock polling stations. Attending training is generally unpaid.

30. The findings on how poll workers were trained in the 2023 local elections where voter ID was used for the first time are clear. Only 12.6% of poll workers received in-person training. Over four-fifths received training by remote means: 25.9% by Zoom or other video-conferencing facility; and 55.7% via a link to watch a recorded presentation (N=2582).[12]

31. While 93.7% thought that their training prepared them well for polling day (N=2,577), there is an important question about the effectiveness of this in practice. Much research on online training finds positive effects. Yet, on a much larger scale, educators forced to teach online during the pandemic struggled with student engagement. Some studies found that learning gain with online teaching depends on the quality of internet access and the ability of the recipient to adapt to this mode of training.[13] Others have found in-person training more effective at imparting knowledge than online.[14] Online training should not therefore be a panacea where electoral rights are at stake, particularly where a consistent minority already showed some concerns about the clarity of electoral law.[15]

32. Sample qualitative comments from our survey underline difficulties with training and understanding the Election Act’s requirements:

‘train staff better to know the guidance on ID. Most did not know expired ID could be accepted meaning people were turned away.  Others thought they had to have the correct address on driving licenses, this meant some were challenged.’

‘The training on identifying fake ID was not adequate to honestly guarantee I could recognise one.’

33. Increased time taken to process voters can lead to queues developing. Queues in polling stations have, in American research, been highlighted as a potential disincentive towards voting for various categories of electors who might be, for example, time poor.[16]

34. In the 2023 local elections, voter ID led to it taking longer to process voters. Over 60% of respondents to our poll worker survey indicated this had been the case where they were working, although only 1.4% (n=2548) indicated that ‘long queues caused voters considerable inconvenience at my polling station’.

35. While a longer processing time might be manageable in low turnout local elections, in a general election where turnout is often almost double (or more) than for local councils, this will almost inevitably lead to queues, particularly at busy points during the day such as the end of the working day, or close of poll.

36. Polling station staff were asked to compare their experience of working at the 2023 local elections with voter ID with previous elections they had worked at. The main responses were relatively evenly divided between ‘about the same’ (44.6%) and either a lot or a bit more difficult (44.7%; N=2009).

37. The potential for any general election to be run concurrently with local elections already scheduled for May 2024 should be of concern to the Committee. My research has shown that while holding different levels of election together is often argued for in terms of cost savings, in practice this can lead to poorer quality election administration.[17] This would add another complication for polling station and other election staff and suppliers to deal with, since on top of voter ID, they would have to issue multiple ballot papers. This would inevitably add time to processing voters, and introduce the potential for unnecessary additional confusion and administrative difficulties. While I understand the constitutional position with regard to calling general elections, I would suggest that the Committee caution against holding general and local elections together in 2024.      

Other Impacts of the Elections Act 2022

38. It is often not understood just how small electoral services and electoral registration teams are. It is not unusual for electoral services teams to have as little as three or four full time members of staff. Under those circumstances, illness, stress and high workloads where there is no room for failure put extreme pressure on electoral services teams. The Association for Electoral Administrators has warned on numerous occasions about the pressures this puts on the ability to deliver elections.[18] It is not unusual to hear of electoral services staff working upwards of 12-15 hours daily, including weekends, in the run up to elections.

39. The alternative is for councils to take a ‘full council’ approach. In this, available staff are diverted or seconded to work on elections as necessary. This seeks to dilute those stresses on small electoral services teams. However, this has potential to introduce other difficulties and mistakes which have the potential to compound to become a bigger problem. For example, the very complexity of electoral law needs expertise, familiarity with and clarity about statutory deadlines and requirements. Introducing non-expert staff risks errors in administrative processes, which potentially go on to negatively impact voters’ democratic rights.

40. Whichever approach councils take, the tranche 2 reforms contained in the Elections Act 2022 would seem to worsen the situation. They would introduce several new and time consuming requirements adding to the stress of small electoral services teams already under pressure to deliver, while also potentially introducing further scope for error where a ‘full council’ approach is taken.              

41. Extending the franchise to overseas electors who have lived abroad for more than 15 years could helpfully be postponed. This is for two reasons. Firstly, the real problems experienced by overseas voters are to do with cumbersome processes and tight deadlines for despatch, receipt and return of postal ballots. Extension of the franchise as per the Elections Act 2022 does nothing to deal with these very real difficulties already experienced by overseas electors.

42. Secondly, the extension of the overseas franchise will lead electoral registration staff into lengthy verification processes for new overseas applicants. Given the length of time that has passed since new applicants lived in the UK, their details are more likely to be held in analogue or hard copy form than electronically, leading to complications in verifying applications. While demand is unknown, this risks adding considerably to workload at a time when registration staff are dealing with registration (re)applications and also those for VACs.    

43. With the shortening of the period for postal vote (re)applications and other aspects of the tranche 2 reforms, there is no obvious or compelling reason (beyond the legislation) why they need to be introduced for the next general election. They add further processes to be completed under a very tight timescale (for instance, the increased levels of signature matching and ID verification necessitated by postal vote reapplications).

44. I would recommend that these tranche 2 reforms be postponed until after the general election and a realistic revised timetable for their introduction be published and legislated for where necessary.  

Are There Other Potential improvements?

45. Respondents to our poll worker study from the 2023 local elections made numerous suggestions for improvements. Many of these were linked to the overall electoral process, while a sizeable proportion also related to voter ID and its implementation.[19]

46. One experienced presiding officer suggested polling stations being given shorter electoral registers to administer, which would make difficulties with voter ID potentially more manageable.[20] However, shorter registers to administer would most likely mean a need for more polling stations as well as staff. This would likely be a reform for the longer term, post-2024/5 general election.

47. Other respondents suggested additional forms of identification be permitted. Suggestions included student identification, young people’s travel cards, and gun licences. The feasibility of these alternative IDs should be investigated.

48. A significant number of respondents who provided qualitative comments on voter ID (126/196) suggested that the voter ID requirements should be abandoned altogether and the previous system be reverted to.    

Adequacy of Data Collection and Reporting

49. Data collection on the effects of voter ID after the 2023 local elections was problematic for several reasons. Although the Electoral Commission have provided an estimate of the number of voters that did not return to vote in 2023 (0.25% or approx. 14,000), this is likely to be an underestimate. As noted above, it is extremely difficult to establish the motivations of those who do not turnout to vote, among which, voter ID may only be one among several factors.

50. A significant difficulty in the aftermath of the 2023 local elections was the absence of any early information on voter ID’s effects, despite councils collecting the numbers of voters who were turned away at the polling station desk. Councils were not required to publish this publicly, although some eventually did so.

51. This lack of information meant it was (and is) difficult to assess and scrutinise how the Elections Act 2022 has worked in practice. While it is certainly important that wider evaluations are done post-election, it is also vital that the effects of voter ID plays a role in the immediate discussion of results when public attention is at its highest.   

52. Councils should continue to be required to collect data from polling stations on how many voters were turned away. I would recommend two reforms to this process however to improve public scrutiny of the impact of voter ID. Firstly, these data should be published by councils at the constituency level (or ward in local elections). Secondly, and taking this further, these data should be reported as part of the returning officer’s statutory declaration.

53. Given that collation of figures from polling stations already happens for the overall vote count, and for the number of rejected ballots, there is no obvious reason why adding this to ROs’ statutory declaration could not also happen for the number of voters turned away for identification reasons. This would need added to count processes. It is likely also to need further legislation to enable returning officers to comply with such a requirement.

54. The impact on voters will be hard to estimate, for reasons given above. It ought nevertheless to be possible to gain a wider appreciation of how voter ID and the other requirements of the Elections Act 2022 have impacted upon administration of the general election process through wider studies of polling station workers, and also on electoral services teams and / or returning officers. 

55. I would recommend that such wider research be undertaken in the aftermath of the 2024/5 general election. I would hope that the Committee would support such research being conducted. Such studies have provided vital independent academic evidence about the operation of the electoral administrative system. They have contributed to the evidence in this paper, in addition to knowledge of variation in how the electoral system works under various other high-pressure situations.[21]

56. Finally, DLUHC ought to be pressed to publish data on the financial costs incurred by the implementation of voter ID and other electoral administrative aspects of the Elections Act 2022. Its predecessor department – the Cabinet Office – promised to publish general election cost data two years in arrears. To date, only data up to the 2015 general election and the 2016 Police and Crime Commissioners have been published.[22]

57. Such delays in publication are woeful when trying to scrutinise government for the cost effectiveness of its electoral administration policies. This is doubly important with the implementation of voter ID if an informed analysis is to be made of the policy’s overall effectiveness. Importantly, this also needs to include data on the shortfall between what councils actually claim and what government are willing to pay for through the Maximum Recoverable Amount (MRA).

58. I recommend therefore that the Committee request a report and publication of the actual data regarding the cost of implementation of voter ID in the 2023 local elections, and for this to be repeated as soon as possible in the aftermath of the 2024/5 general election. More generally, this should also be brought up to date for the other elections that remain outstanding.  

3 November 2023