Who lacks voter identification? The electoral implications of the Elections Act 2022

Abstract

The Elections Act 2022 requires voters to present photo identification at general elections in Great Britain and local elections in England. In this article, we use data from the British Election Study Internet Panel to identify who does and does not own the necessary photo documents and how this affected turnout in the 2023 and 2024 local elections, as well as the 2024 General Election. We find that around 5% of the voting-age population lack valid voter identification and that this is related to age, education, socio-economic status, and social grade. The data also give us a reliable indication of the party-political impact of the photo ID requirement. We find that conservative voters are more likely to hold valid photo identification than supporters of other parties. We also find that a small but significant proportion of registered voters (between 1.2% and 2.4%) reported either being turned away or dissuaded from voting at these elections because of the requirement.

The Elections Act 2022, passed into law by the Conservative government, introduced compulsory photo identification for in-person voting for the first time in Great Britain, as part of a tranche of reforms governing electoral practices.¹ The photographic identification provisions of the act came into practical effect at the 2023 local elections in England and will be in effect at all future Westminster elections in Great Britain, and local elections in England.² Acceptable documents include passports, photographic driving licences, and some travel passes/parking permits for people with disabilities.

The ostensible reason for the introduction of voter identification was to prevent electoral fraud and increase confidence in the integrity of elections. However, critics have charged that there was little evidence to suggest that electoral fraud and ‘personation’ are widespread problems (James and Clark 2020; All Party Parliamentary Group on Democracy and the Constitution., 2023) and that the true goal of the stricter voter identification rules is to secure the Conservatives an electoral advantage.³

In this article, we use four waves of the British Election Study Internet Panel (BESIP) (Fieldhouse et al. 2024) to examine who is most likely to be affected by voter identification requirements by measuring who does and does not have access to the necessary photo documents, and how that affected voter turnout in the local elections in England in May 2023 and 2024, and the general election in July 2024. By virtue of its large sample size and wide array of demographic and political measures, the BESIP allows us to examine detailed characteristics of those lacking valid voter identification including their social background and their party preferences. With a sample of over 30,000 British electors in each wave, we are able to examine which social and demographic groups lack voter identification and their political preferences.

We find that, overall, at the time of the 2023 and 2024 local elections in England, approximately one in twenty electors did not own any recognized voter identification, and this was concentrated in poorer and less educated groups and among supporters of left-leaning political parties. While not all those lacking voter identification would have voted in the absence of photo identification requirements, as many as four times as many people did not attempt to vote because of voter identification requirements than were turned away.⁴

1. Preventing electoral fraud or political opportunism?

As we remarked above, the stated reason for introducing stricter voter identification requirements is to prevent fraud and thereby improve confidence in the integrity of elections. Parliamentary debates leading to the passing of the Elections Act 2022 reflected these competing frames of voter fraud and electoral integrity versus voter suppression and discrimination (Alonso-Curbelo 2023). This debate mirrors the US debate over voter ID, where Republicans seeking more restrictive voter access have been widely accused of voter suppression for electoral gain, namely to disproportionally discourage minorities—who are more likely to be Democrats and less likely to possess the necessary documents—from voting (Hajnal, Lajevardi and Nielson 2017; Barreto et al. 2019; Fraga and Miller 2022).

In a letter to Chloe Smith MP, the Minister responsible, the Electoral Reform Society, and over 40 other organizations argued that the proposed requirements, which could affect 3.5 million voters, were disproportionate in the face of only a handful of allegations of personation (Electoral Reform Society 2018). In a report following the 2023 local elections in England, the first elections since the act was passed, the All Party Parliamentary Group on Democracy and the Constitution (2023) concluded that ‘the current voter-ID system is, as it stands, a “poisoned cure” in that it disenfranchises more electors than it protects to the extent that it could affect electoral outcomes’ (p. 4). Opponents of voter identification also point to data from the Electoral Commission that there is no evidence of large-scale fraud in Britain (Electoral Commission 2023a). For example, at the 2019 General Election, the police investigated 25 allegations of personation at polling stations, none of which led to a conviction. Indeed, between 2014 and 2022, there were only three convictions of personation in total (Johnston and Uberoi 2023). Reflecting this concern, it has been claimed that an increase in those discouraged from voting potentially undermines the right to representation in general elections under the European Convention on Human Rights and may be subject to legal challenges (Good Law Project 2023).

However, supporters of voter identification argue that the requirement for voter identification is widespread across most democratic countries (Barton 2022) and has been framed by supporters of the Elections Act not only as a necessary measure to prevent fraud but also a commonsensical and natural step in modernizing the democratic system (Alonso-Curbelo 2023). Moreover, concerns about voter suppression do not necessarily mean that attempts to prevent personation are without merit. In 2015, the Electoral Commission recommended the introduction of photo identification based on the model already used in Northern Ireland (Electoral Commission 2016b). They pointed out that whilst levels of detected fraud were low, polling stations could become vulnerable in the future especially as absentee voting and registration systems become more secure (Electoral Commission 2016a).⁵ The Organization for Security and Co-operation in Europe Office for Democratic Institutions and Human Rights—an international election monitoring organization—has also previously recommended the introduction of voter ID in the UK (OSCE/ODIHR 2010, 2015). The government’s own research and pilot studies suggest that most people do not find it difficult to provide identification. Following the piloting of voter ID requirements in ten local authorities in 2018 and 2019, research by the Electoral Commission pointed towards an increase in voter confidence in the pilots compared to other areas and little impact on turnout (Electoral Commission 2019).

Evidence from other countries that voter identification requirements negatively affect turnout is mixed and, because voter registration laws are not randomly implemented, it is difficult to assess the impact of such laws on aggregate-level voter turnout. However, individual-level survey research on the impact of voter identification in the USA suggests that stricter voter identification requirements—particularly the requirement of photo identification—do depress turnout, especially among less educated and lower-income populations (Alvarez et al., 2008). Nevertheless, a comprehensive review of US research on the topic concludes that the impact on turnout is only modest, finding only minor differences across politically relevant groups (Highton 2017), and demonstrating the importance of high-quality empirical data to assess the electoral impacts of these laws.

Political science offers strong theoretical reasons why even relatively modest barriers to voting might reduce turnout. Decades of research on voter turnout suggest that the most compelling motivations to vote are not instrumental calculations but expressive and normative (Fieldhouse and Cutts 2018). That is, individual voters are unlikely to judge that their vote will be pivotal in determining the election outcome. Rather, they vote because they wish to ‘have their say’ and show their support for their preferred party or candidate (expressive), or because they think it is a duty and that voting is expected of them by the people they know (normative). Moreover, the voting decision is based on a low-cost–low-benefit decision calculus (Aldrich 1993). This means that anything that raises the cost of voting (like photo identification requirements) is likely to reduce turnout amongst those whose turnout decisions are marginal. It also suggests that the incentives to fraudulently cast a second or third ballot are low: voting for a second or third time will not increase the expressive or normative benefits from voting and will only have a minimal impact on instrumental ones.

By contrast, there are obvious rational incentives for some political parties to support or oppose—stricter voting identification requirements, depending on the probability of their supporters possessing valid voter ID. In the USA, it is widely argued that Republican measures to restrict access to voting disproportionally affect African American voters, the majority of whom usually vote Democrat (Hajnal, Lajevardi and Nielson 2017). Reflecting the partisan split, strict voter identification laws (i.e. photo identification) have only been passed in states where the Republican party control state government (Hicks et al. 2015; Highton 2017: 150).

The UK government’s own figures, published before the legislation was passed, suggested that around 4% of the electorate have no photo identification (IFF Research 2021). If restrictive voting practices affect less educated and lower-income voters as elsewhere, we might expect that stricter access to voting would give an electoral benefit to the Conservatives. However, there is a caveat to this: in the UK, it is unlikely that the vast majority of people without photo ID would support the same party, especially as the ongoing realignment of voters triggered by Brexit means that those groups most likely to lack voter ID are not necessarily the most pro-Labour groups (Fieldhouse et al. 2023).

2. Data

We use data from waves 25, 26, 27, and 29 of the BESIP. Fieldwork was carried out by YouGov and each wave of the panel includes over 30,000 respondents. Wave 25 was conducted in May 2023 following the local and mayoral elections in England, and Wave 26 was fielded in May 2024 following local elections, shortly before the 2024 General Election was called (Fieldhouse et al. 2024). Wave 27 was fielded between 24 May and 7 June after the general election was called, and Wave 29 between 5 July and 19 July (post-election).⁶ To assess access to valid photographic identification, respondents were asked the following questions:

Which, if any, of the following forms of photographic identification do you currently hold? *Please tick all that apply*

Respondents were presented with a list of all forms of valid photographic identification according to the Elections Act, which was split over successive pages, with those forms of identification that we expected to be the most common appearing on the first page (UK passport, driving licence, Blue Badge). Respondents who answered ‘none of these’ or ‘don’t know’ were taken to a second page with additional forms of identification. In total, respondents saw up to four pages of alternatives until the list of valid forms of identification was exhausted. If, after the last page, the respondent still says ‘none of these’ then she/he is coded as having no valid photographic identification.

To ensure that the BES data is representative of the relevant population, we designed weights with bespoke targets (in addition to normal BES weighting targets) defined by the age profile of British passport holders according to the 2021 Census of Population. This ensures that the sample is representative of the extent to which different age groups in Britain hold passports. This reduces bias in the estimates of the proportion lacking voter identification because non-passport holders tend to be underestimated when using standard BES weights, and we know from our own data that passports are the most widely held form of identification (see below). In addition, when estimating the effect on voting behaviour in the local elections, we also include weighting targets for local election turnout for 2023 and 2024 and general election turnout in May 2024, as a larger proportion of respondents to the survey report voting in the elections than was observed in the whole population.⁷

Using these weights, we estimate the percentage of the electorate that holds different forms of identification. The most widely owned forms of photo identification are valid passports and photo driving licences, which are both owned by around three-quarters of respondents in both years, although there was a very slight (1%) increase in possession of both in 2024 (see Supplementary Fig. A1 for further details). Blue Badges were held by 6.4% of respondents. Beyond this, we cannot provide accurate estimates of the percentage of people showing each separate form of identification because, as noted above, the structure of the questionnaire means that respondents were only asked about other forms of identification if they did not possess a passport, driving licence, or Blue Badge. However, assuming that people who have another form of valid ID would not request a Voter Authority Certificate (VAC), we can be reasonably confident of our estimate that only 0.3% of voters reported having a VAC (in both 2023 and 2024). This reflects a low level of take-up in comparison with the numbers lacking voter identification (VID), which is likely to result from a low level of awareness at the time (Electoral Commission 2023b).

Combining these into a single indicator, we find that 4.4% of the eligible voter population lacked any suitable form of photo identification in May 2023, and this increased to 5% in May 2024 (though the difference is not statistically significant).⁸ We refer to this as VID poverty —not to imply a financial cause of lacking VID, but simply for want of a more easily understandable noun to mean ‘to be lacking’. The overall proportion we find without photographic identification is consistent with the Cabinet Office report published in March 2021, which also showed that 4% of respondents to a telephone survey lacked recognizable photo identification, with 9% lacking an in-date and recognizable photo ID (IFF Research 2021). As most people possess either a passport or a driving licence, the other forms of voter identification contribute a small but important addition to the total availability: overall in 2023, 96.3% of those with a valid form of identification had either a driving licence or a passport, while 3.7% of people relied on some other form of identification.

Of course, lacking photographic identification is not necessarily a permanent state. Some people will be in the process of renewing passports and driving licences or other forms of identification, while others may become eligible for a new type of identification or might apply for a VAC to enable them to vote (although uptake is not currently high as discussed above). One of the advantages of the BES panel study is that we can examine stability or continuity of access for respondents who answered the relevant questions in May 2023 (Wave 25) and May 2024 (Wave 26).

It is interesting to note that this analysis suggests a considerable degree of instability in lacking ID. While most people (more than nine out of ten) had some form of ID in both waves, only 2.7% of all respondents had no ID in both 2023 and 2024 (see Supplementary Table A1 for details). Of the 4.1% of respondents who lacked ID in 2023, around a third (1.4% of all respondents) had gained access to some form of valid photographic ID one year later. On the other hand, there was a significant number of respondents who had previously held ID but reported having none in 2024 (2.1%). This suggests a degree of fluidity and that, while VID poverty may not be permanent for many, it may affect a larger number of people at one point in time than the snapshot figure implies. Between 2023 and 2024, for example, nearly 7% of respondents were without ID at either one-time point or the other. However, because of the low overall rates of VID poverty, this estimated overlap between panel waves should be treated with caution as it may be disproportionately affected by misreporting on one or another occasion (Ansolabehere, Luks and Schaffner 2015).

3. Demographic and social differences

As we noted above, lack of photo identification is likely to affect some sections of the population more than others. This is what makes voter identification and registration both controversial and highly politicized. Despite the relative rarity of VID poverty, the size of the BES panel (Fieldhouse et al. 2024) makes it possible to compare levels across different demographic groups. Our data from May 2023 suggests that women are slightly more likely to lack photo ID (4.7%) than men (4.2%), and ethnic minority respondents are slightly less likely to lack ID (4.1%) than white British respondents (4.4%), though the differences by ethnicity are not statistically significant.^9,10

While gender and ethnic differences are small, we might expect different age groups to vary in their propensity to hold different forms of identification and therefore to display different levels of VID poverty. For example, we know from the Census (and from our data) that passport holding tends to decline with age (see Supplementary Fig. A2). In contrast, photo driving licence ownership increases until age 30 as people learn to drive and then declines before rising again at the age of 70 and falling again among the over-80s. This complicated pattern appears to reflect the introduction of photo licences in 1999 meaning that people between 50 and 70 are more likely to have a non-photo licence because most will have applied for a licence before photographic licences were introduced. Similarly, the requirements of those over the 70s to reapply for their licence every 3 years account for the rise among the over 70s. A closer examination of the data lends further credence to this explanation for the rollercoaster pattern: there is a sharp discontinuity in the probability of having a photo driving licence at age 70 (Fig. 1, left-hand panel). This discontinuity is repeated in the 2024 data (Wave 26), although the initial rise between ages 18 and 30 is less marked (Supplementary Fig. A3).

The combined effect of the age distribution of passports and driving licences together with the contribution of travel cards for older people produces a complex relationship between access to photo ID and age, which is also shown in Fig. 1 (right-hand panel). In 2023, access to any valid photographic ID declined very slightly with age but rose sharply at pension age as access to older person’s travel passes and new photo driving licences took effect. Overall, the opposing effects of passport and driving licence ownership to some extent cancel each other out, meaning age differences are relatively modest. In 2024, we see a similar discontinuity at pension age, although the decline with age for the under 65s is no longer evident. This subtle difference by year is probably an artefact of the small number of people lacking voter ID when broken down by single years of age, even with the BES sample size and the effect of smoothing. The key feature—which is clear in both years—is the sharp rise in access to voter ID around age 70.

Perhaps not surprisingly, socioeconomic differences in access to photo identification are, by and large, greater than demographic ones. Differences by educational level are particularly notable. While in 2023, 7.8% of respondents with no formal qualifications lack any voter identification, the equivalent figure for those with a postgraduate degree is 2.4% (see Supplementary Fig. A4). The educational gradient was even steeper in 2024 ranging from over 9% for those with no qualifications to under 3% for people with postgraduate qualifications. A similar gradient is apparent for social grade: for example, in 2023, 7.1% of semi and unskilled manual workers (social grade D) had no photo identification compared to 1.9% of higher and intermediate managers and professionals (social grade A/B). In terms of economic activity, in 2023, the retired were most likely to have eligible photo ID, with only 1.9% in VID poverty, while full-time students were about as likely to lack photo ID (3.1%) as those in full-time work (3.2%). However, the unemployed and economically inactive were much more likely to be in VID poverty, with 10.4% and 12.6% lacking appropriate ID, respectively.¹¹ With respect to household income, in 2023, 14% of the poorest households (on an income of under £5,000 per year) had no photo identification compared to 1.4% of those of a household income greater than £60,000.

Finally, with respect to geography, notwithstanding small variations between 2023 and 2024, the highest rates of VID poverty were in Scotland, the North-East, and North-West and the West Midlands (Supplementary Fig. A5). The region with the lowest rates of VID poverty were the East of England and London. Looking at more political geography, we can also look at how VID poverty varies based on the marginality of the new seats contested for the first time in the 2024 UK General Election (Supplementary Fig. A6). This shows that, in May 2024, those living in Conservative seats with majorities larger than 20% (going into the 2024 General Election) were least likely to lack photo ID (4.3%), while those in more marginal constituencies (whether held by the Conservatives or other parties) were more likely to be in VID poverty (5.9% and 6.3%, respectively).

In summary, VID poverty is strongly related to social and economic disadvantage, education, age, and geography. Many of these factors are correlated with voter turnout (Smets and Van Ham 2013), so some of the potential impact of the introduction of VID on turnout may not be realized. This is supported by our data in which the percent lacking VID is much higher among those who say they are unlikely to vote (see Supplementary Fig. A7). However, the same social and demographic characteristics are also correlated with vote choice so they are likely to have important political implications, which we explore further below.

4. Party-political implications: vote choice intention

We showed above that access to VID is not equally spread across the population. Given that the political parties draw their support unevenly from different social demographic groups, these inequalities mean that voter identification legislation would be expected to have an uneven impact on vote shares at subsequent elections, if some electors are prevented from voting. As noted above, the Elections Act provides for electors to apply to the local electoral registration officer for a VAC if they lack one of the approved forms of photo identification. In other words, no eligible voter is completely prevented from voting because they lack the forms of photographic identification discussed above.

However, as previous research on voter identity requirements demonstrates, turnout of some groups tends to be negatively affected, and, because voting is a low-cost/low-benefit activity, a small increase in the cost (effort) of voting may act as a deterrent. This reflects the fact that not all voters will be sufficiently motivated or organized to apply for a voter identification document, and the requirement to do so acts as a disincentive to participate. We should therefore regard inequalities in the availability of VID between supporters of different parties as indicative—not conclusive—evidence of the potential impact of voter ID on the vote share of each party. Notwithstanding these caveats, any party-political differences are likely to be controversial because of the possibility of a political advantage for the government of the day, giving rise to public perceptions of political opportunism.

Figure 2 shows the percentage lacking photo ID by general election vote intention, as measured in May 2023 and in May 2024, shortly before the general election was called. Perhaps unsurprisingly based on the demographic patterns described previously (particularly social grade), we find that in 2023 those intending to vote Conservative at the general election were less likely to lack photo identification (2.4%) than Labour supporters (3.8%) and supporters of all other political parties. Moreover, although the relative popularity of the parties varies over time, if we look at the rates of VID poverty a year later in 2024, we see a very similar pattern. The one notable change is an increase in Liberal Democrats and non-voters with no photo identification, although this is almost entirely due to a change in the makeup of Liberal Democrat support and non-voters rather than in situ changes in photo ID possession.¹² The Liberal Democrat voters had the lowest proportion of supporters without voter ID in 2023 (1.3%), but in 2024, the Liberal Democrat rate exceeded that of the Conservatives (2.9%).

In short, based on general election voting intention measured in May 2023 and May 2024 (and assuming equal propensity to vote), if all those lacking photo identification failed to vote, it would have had a greater impact on the Labour vote than the Conservative vote. Figure 2 also shows that supporters of Reform UK, the Scottish National Party (SNP), and the Greens are the least likely to have voter identification along with non-voters (though there is a large degree of uncertainty about smaller parties because of sample size). Plaid Cymru and other parties are excluded because of small sample sizes.

Parties are not only interested in current vote intention or past voting behaviour but in the potential for winning future votes. Bearing this in mind, it is instructive to look at how much electors, with and without voter identification, like each party, as shown in Fig. 3. In each wave of the BESIP, respondents are asked ‘How much do you like or dislike each of the following parties?’ on a 0–10 scale, where 0 is strongly dislike and 10 is strongly like. Figure 3 shows the average like score for each party asked about in May 2024, broken down by whether the respondent did or did not have a voter ID in that wave. The Conservative Party was liked much less by those lacking photo ID, while there are no statistically significant differences between people with and without VID in their liking for Labour, SNP, Plaid Cymru, and Reform UK. In other words, people lacking voter ID seem to especially dislike the Conservative Party. The Liberal Democrats are also much more liked by people with voter ID than those who lack it, likely reflecting their more highly educated electorate.

Interestingly, Fig. 3 also suggests that parties are generally liked more by people with photo ID than without—reflecting that people without VID tend to be in groups more disillusioned with politics in general.

This analysis suggests that the preference for Labour over the Conservatives amongst people without photo identification runs deeper than simply current vote intention. This is illustrated in Fig. 4, which shows the smoothed probability of having no photo identification by the difference between Conservative and Labour-like scores.¹³ This suggests that there is no monotonic relationship between preferring Conservatives over Labour (or vice versa) and not having any identification. In general, those who are relatively indifferent about the parties tend to be the most likely to lack photo ID. These are also the people we would expect to be most easily discouraged from voting. We also see that VID ownership is generally lower among Labour preferers (negative difference scores in Fig. 4) than among Conservative preferers (positive difference scores). This provides further evidence that the legislation is likely to have a greater negative impact on Labour’s future vote potential than that of the Conservatives.

5. 2023 and 2024 elections

Local elections were held in England in May 2023 and 2024 in some but not in all local authority districts. BESIP respondents living in areas that held elections were asked a series of questions about their turnout behaviour including whether they voted, how they voted (e.g. by post, by proxy, in person), and, if they did not vote, the reasons for that. They were also asked whether their photographic identification was checked at the polling station and if they were turned away because of a lack of voter identification. These data were combined to create a classification of people who potentially could have voted in the 2023 and 2024 local elections and the 2024 General Election.

Figure 5 shows the percentage belonging to each category broken down by possession of valid photo identification for the 2023 and 2024 local elections, as well as the 2024 General Election. As we would expect, we find the proportion who did not vote because of identification requirements was much higher amongst those who reported having no identification. Over 80% of those lacking ID did not attempt to vote in the 2023 and 2024 local elections, and over 60% in the general election, but of these, the majority gave reasons other than their lack of valid documents. Nevertheless, sizeable proportions of those without identification reported not attempting to vote in these elections and gave that as the primary reason (25.2% in the 2023 local elections, 15.2% in 2024 local elections, and 12.5% in the 2024 General Election).

What is also clear from Fig. 5 is that there is no automatic relationship between lacking identification and voting. Not everyone without valid identification was unable to vote because it is not required for postal or proxy voting. In 2023, 12.5% of respondents with no photo ID reported voting by post or proxy, compared to 15% of those with identification. The equivalent figures in the 2024 local elections were 7.8% and 10.2%, and in the general election, they were 19.3% and 25.8%.

Previous research has indicated that those who vote by post tend to be motivated to vote through a sense of duty, are more likely to be mobilized by political parties, and live in competitive constituencies but are also more likely to be affected by the cost or inconvenience of voting (Townsley et al., 2021). This suggests that those electors who lack voter ID but who are already more motivated to vote are more likely to use postal voting as a ‘work around’ and less likely to be deterred by the requirement. It is also worth noting that the mitigating effect of postal voting on the impact of the introduction of voter identification requirements may be affected by other changes in the Elections Act concerning the handling of postal votes. These changes had not come into effect for the 2023 local elections but had come into force in 2024.

A small proportion of people who lacked voter ID claimed to have voted in person in the 2023 and 2024 local elections (3.9% and 2.6%, respectively) and a significantly larger percentage (14.9%) claimed to have voted in the general election. Assuming that, for the most part, voter identification requirements were implemented correctly—which may not uniformly have been the case (Stride 2023)—there are two likely explanations for this apparent anomaly.

First, it is widely recognized that survey respondents report having voted when in reality they had not and that over-reporting is greater when actual turnout is higher (Karp and Brockington 2005). Given the average rate of over-reporting in BES surveys is also close to 15%, this is likely to account for most of this anomaly.¹⁴ Second, data on possession of identification for the general election was collected either in May (for respondents who were included in Wave 26) or early June (Wave 27) for new respondents, meaning that some respondents may have obtained valid identification between answering the survey and election day. As we showed above, there is considerable churn in possession of identification as most forms (e.g. passports) require occasional renewal. Moreover, the Electoral Commission reported that 57,000 applications for voter authorization certificates were submitted between the election being called and polling day (Electoral Commission 2024). This suggests a real increase in overall levels of identification in the run-up to the general election, which may also account for part of the anomaly.

Overall, the total proportion reporting that they did not attempt to vote because they lacked identification or were turned away for that reason made up 2.4% of all those eligible to vote in the May 2023 elections. This constitutes a significant number of disenfranchised electors, comprising 1.9% who were discouraged and 0.5% who were turned away. The equivalent figures were slightly lower in the 2024 local elections: 1.2% did not attempt to vote because of lack of ID and 0.3% reported being turned away (making 1.6% overall with rounding). The proportions were slightly smaller again in the 2024 General Election, with 1% not attempting to vote due to lacking ID and 0.3% reporting being turned away (making 1.3% overall). These are very much in line with equivalent estimates made by the Electoral Commission.¹⁵

6. Conclusions

The need to produce photographic identification documents to vote has proven to be controversial. On the one hand, it is widely supported as a necessary measure to prevent personation. On the other hand, it has the undesirable effect of potentially disenfranchising large numbers of electors who are unable to produce valid identification. The impact of the introduction of voter identification is difficult to gauge, largely because it affects a relatively small proportion of people, which presents problems for typical sample surveys. At the same time, while administrative records can provide accurate numbers about how many people were turned away at the polling station, they tell us little about people who were discouraged from voting. There are also problems with estimating how many of these later returned.

The BESIP (Fieldhouse et al. 2024) provides unique evidence regarding which voters are likely to be most affected by the introduction of photographic voter identification requirements. Its large sample size allowed us to examine detailed characteristics of those lacking valid voter identification including their social background and their party preferences. We have demonstrated substantial age, education, and income differences in access to voter identification. Moreover, we find that at the time of the 2023 and 2024 local elections in England, around one in twenty did not possess valid voter identification, and, while only around 0.5% were turned away at polling stations, between three and four times as many people were discouraged from voting. The equivalent figures were slightly lower at the general election, but still a meaningful contingent, with around 1.3% having been turned away or discouraged by voter identification requirements.

Our evidence also makes clear that the 5% of the electorate lacking identification indicated, on average, that they would have been less likely to turn out to vote than other electors anyway. This reflects a correlation between the factors influencing the underlying propensity to vote, such as education, class and age, and the possession of photographic identification. Moreover, a substantial proportion of these electors did report having voted by post in each of the elections studied, which does not require photographic identification.

Notwithstanding the relatively small percentage of electors directly affected, the choice to vote—or not to vote—is an important democratic right, so we should not only be concerned about the direct impact on voter turnout but about the wider implications of making it harder for some groups to take part in elections. While we have shown that there is considerable volatility in access to VID, some voters appear to remain without VID over a prolonged period. Future research using the BES data will be able to track how their sense of political efficacy and trust in politics changes in comparison with the rest of the electorate.

There are other implications. An increase in the numbers turned away may lead to increased pressures on polling station workers and election administrators, with the possibility of long queues and incidents of disorder (James and Clark 2020; Stride 2023). We have shown that inequalities in voter ID access have party-political implications. Given that parties of the right have historically been assumed to benefit from more restrictive identification requirements, it is perhaps not surprising that our evidence suggests that the Conservatives are most likely to benefit from the requirement, based on recent patterns of support. The most obvious reason that VID access is high amongst Conservative supporters is that it is strongly associated with those social and demographic characteristics that are related to Conservative voting. Despite changes in social patterns of party support since the 2016 Brexit referendum, Conservative voters are (increasingly) older, and still more affluent and therefore more likely to own photographic identification. The selection of the types of identification that are allowed—especially the inclusion of travel passes for older people, and controversially not younger people—exacerbates these differences.

There are still opportunities to mitigate these risks, for example, by extending the forms of identification allowed (especially for young people and those living outside of London), allowing a system of vouching in the absence of valid ID and improving public awareness of both the requirements and the availability of VACs (All Party Parliamentary Group on Democracy and the Constitution 2023). While around 5% of the electorate lacking voter identification may sound like a small proportion, with a UK electorate of forty-nine million, that equates to almost two million potential voters. Those interested in building trust in our democracy should consider not only minimizing electoral fraud but also reducing this number by as much as possible.

Supplementary data

Supplementary data is available at Parliamentary Affairs online.

Acknowledgement

The authors gratefully acknowledge the support of the Economic and Social Research Council in funding this research (Grant No. ES/S015671). The views expressed herein are those of the author and do not reflect the position of the United States Military Academy, the Department of the Army, or the Department of Defense. 

Conflict of interest statement. None declared.

References

Footnotes