final logo red (RGB)

 

Select Committee on the Electoral Registration and Administration Act 2013

Corrected oral evidence: Electoral Registration and Administration Act 2013

Tuesday 4 February 2020

4.35 pm

 

Watch the meeting 

Members present: Lord Shutt of Greetland (The Chair); Lord Campbell-Savours; Lord Dykes; Baroness Eaton; Lord Hayward; Lord Janvrin; Lord Lexden; Baroness Pidding.

Evidence Session No. 9              Heard in Public              Questions 101 - 110

 

Witness

I: Dr Christopher Prosser, Presidential Fellow in Politics at the University of Manchester and co-director of the British Election Study.

 

 


12

 

Examination of witness

Dr Christopher Prosser.

Q101       The Chair: Good afternoon, Dr Prosser. Welcome to this evidence session. The meeting is being broadcast live on the parliamentary website. A transcript of the meeting will be taken and published on the Committee website. You will have the opportunity to make corrections to that transcript where necessary. Only one member of our Committee, Lord Lexden, has declared an interest, as a trustee of the Hansard Society, so you are aware of that.

It would be helpful if you could describe the work of the British Election Study and tell us more about the data that you collect and what it is used for.

Dr Christopher Prosser: Thank you for having me along, on behalf of the British Election Study.

The BES is a very long-running academic survey and research project that examines political attitudes and electoral behaviour in Britain. It was founded by Sir David Butler and Donald Stokes in 1963, and it has been conducted around every election since 1964, so we have a very long-running series of data. We are publicly funded by the Economic and Social Research Council. We are strictly non-partisan in our research, and all the data we collect is publicly available at no cost to the end-user; all our data is freely available and replicable.

The primary purpose of the data we collect is academic research, which is done by me and my colleagues on the BES team, and other British and international scholars. It is also used widely outside academia. The media make quite a lot of use of it, polling companies make use of it, political parties make use of it in some of their research, and it is used by public bodies such as the Electoral Commission.

The current team at the BES took over the running of the project in 2013. We collect two types of data. The first is an internet panel survey, which is fielded for us by YouGov, and is a large-scale survey of about 30,000 people per wave. The important thing about that survey is that we try to talk to as many of the same people as possible in repeated waves of the survey as time goes on; we have now conducted 19 waves of the survey, going back to 2014, covering all the elections that have happened since then. The main advantage of it is that we can look at the same people. When it comes to electoral registration, we can look at who is registered at one time point and then drops off at another time point, and who is unregistered and registers at a later time point.

The large-scale panel element comes at a slight cost to representativeness, as regards the sort of people who end up in these surveys. It is slightly biased towards people who are more politically engaged than average. Because of that, and because we are particularly interested in things such as turnout and the political attitudes of people who are less engaged with the political process, we also run a random probability face-to-face survey. We do that by randomly selecting thousands of addresses from across Great Britain, sending letters and inviting people to take part in the survey; then we knock on their door and try to get them to answer our questions.

The key thing with the face-to-face survey is that, after we have asked people about their attitudes and what they did at the last election, we ask them if we can have permission to link their name and address to the electoral register. Then we go to the Electoral Commission, look at the marked electoral registers from the previous election, and look up whether our respondents were registered and whether they voted or not. As far as I am aware, this makes our data unique in actually examining levels of registration among particular sets of individuals who have been scientifically sampled at the time of the election.

The Electoral Commission conducts its own exercises with big survey research, but, as far as I am aware, it conducts them only at different points in the electoral cycle. There was a big one in 2015, after the introduction of IER, but that was conducted in December. As we say in our written submission, and I am sure we will go into it, the point we look at in the electoral cycle has important consequences for the answers we give about levels of registration, which groups are unregistered, the effects of IER, and that sort of thing. I hope that gives you a general idea of what we do.

Q102       The Chair: You talked about research, study and so forth, but there are some certain facts, I think. It occurs to me that the registers that were published at some time in November last year, or most of them, as I understand it, were as up to date as they could be. Then there was a great rush of applications to register, some of which were already there, when people who were trying to register were already registered. Do you know yet what the extent of the register as published was, and what the extent of the enhancements was?

Dr Christopher Prosser: We cannot yet say from our survey data what registration is. One of the downsides of our face-to-face survey, which is very important for looking at registration, is that it takes a very long time to conduct. It is still in the field and will be in the field for several months yet, so we will probably not be able to use our data until the summer.

The Chair: But is it not a fact? The Electoral Commission ought to know it, and you ought to be able to get it from them.

Dr Christopher Prosser: We know the aggregate level of registration. We know that, although there was a massive rush of registration in the run-up to the 2019 election, the overall level of registration had only a very small increase—of about the same level we would expect if it was just an increase in population growth. Despite the fact that there was an enormous aggregate level of application, it does not look as though there was a particularly large surge in the total level of registration.

A very important point that our data suggests is that, if we examine registration at the time of an election, it does not look as though IER has had a particularly large or disproportionate effect on registration, but if we look outside the immediate run-up to an election, people are much more likely to be unregistered. Because they are much more likely to be unregistered, they are much more likely to say, “Oh no, I need to register”, when an election is called, so we end up with a massive surge of enrolments just before elections, as we have seen three times, I think. In 2017, we saw a similar thing, and before the referendum in 2016 we saw a similar pattern.

Each time, people have pointed to that as evidence of a great surge in democratic engagement and said that this time there would be a huge change in the number of people taking part in the election, but it does not seem to come to pass. It largely seems to have stayed fairly stable over the last few years, despite the noise beforehand.

The Chair: Do you have a feel for the percentage of people who ought to be registered, if we had a 100% register—as to what that now is?

Dr Christopher Prosser: We have not tried to estimate what in the academic literature we call the voting-eligible population. We have not yet tried to work that out for the 2019 election, for the simple reason that it is very hard to work out what it is. There is no official number of people who should be registered, who are eligible to be registered. We need to rely on the annual population survey, which is conducted by the Office for National Statistics, to try to work out what has changed in population growth, and that data is not yet available, so we cannot do it yet.

Q103       Baroness Pidding: You may have answered part of this question. Have you made any early assessment of registration and administration in the 2019 election? Is there any issue from that election that you think this Committee should take note of in particular?

Dr Christopher Prosser: As I was saying, we cannot yet use our data to look at that; it will be something we look at eventually, but I suspect it will be outside your timeframe, unfortunately. From the aggregate data, it is fair to say that early indications are that registration levels, at least overall, were probably fairly stable and grew more or less in line with what you would expect from population growth.

One thing that might be worth thinking about in your recommendations is the burden that falls on electoral registration officers when there are massive surges in registration in the run-up to elections. If what our research suggests is true, that outside election periods people are more likely to drop off the register, and if we next have an election in 2024, as is scheduled—for once, it seems like a plausible length of time until the next election—considerably more people than in 2017 and 2019 are going to be unregistered, and we can probably expect an even bigger wave of registrations. You could probably plan for that to some extent, but making sure that local electoral registration officers are aware of it and are equipped to deal with these things is probably important. Otherwise, when they are put under burdens, mistakes are made and people do not get registered properly, and that sort of thing.

Q104       Lord Dykes: Thank you again for this fascinating research from the British Election Study, which I have read several times, as carefully as possible. It is very interesting indeed. Could you talk us through and highlight what you consider are the main points in this research on the impact of individual electoral registration, and the findings?

Dr Christopher Prosser: I certainly can. To go back to the motivation for why we conducted the research in the first place, when IER was introduced, there were two primary concerns. The first was that it would decrease overall levels of registration and the second was that it would disproportionately affect particular groups, such as young people and students—I know that you were just talking to student representatives—people from minority communities and in private rented accommodation, as well as disabled people.

We can use the BES data to address the question of the overall level of registration and use it to look at age and registration, and at people who are private renters compared with home owners, and that sort of thing. Unfortunately, we cannot use it to look at electoral registration for ethnic minorities and disabled people. That is not because we do not think that those are important issues; it is simply a limitation of the data. Those are quite small communities, and we do not have enough people in our surveys to say anything reliable about them.

Our key finding is that if we look at the bit of the impact of IER that is not at an election, and compare levels of registration in May 2014 with levels in April or May 2016, before the local elections and the big surge in registration in the run-up to the EU referendum, it looks as though there is quite a large drop in the overall level of registration. There is also a disproportionate impact on the types of demographic groups that are more likely to be unregistered in general, such as young people and people living in private rented accommodation.In that sense, it looks as if the fears about the impact of IER may have been justified.

To complicate matters slightly, if we look at what happens at elections and compare 2015, before people were kicked off the register at the end of that year, and 2017, after IER came into force, we find no evidence that there was a drop in overall levels of registration. If anything, there may have been a very small increase, but it is certainly within the margin of error. Furthermore, we do not find disproportionate impacts on young people or private renters. They are still less likely to be registered, it is important to say, but they are no less likely to be registered than they were in 2015.

Although those two things sound contradictory, we think they can be reconciled, if we understand the impact of IER to have produced a sort of volatility in levels of electoral registration. It has made people less likely to register to vote if they do not actually have to, if politics is not particularly on their minds; but the sorts of people who register to vote tend to do it before elections, so the falling-off effect tends to be counteracted by the joining effect.

Although that is good news for levels of participation in elections and, as far as we can tell, no one is being systematically disfranchised, it has important consequences, because registers, as you all know, have other uses than deciding who gets to vote. They also decide where constituency boundaries get drawn. Because, at least in the last proposal, the boundaries were going to be drawn with a non-election period set of registers, a lot of people, particularly young people and private renters, are likely not to be on the rolls that would be used for boundary drawing, and there would be some systematic biases in the sort of data that was used to draw constituency boundaries. That is the key point, but I am happy to go into more detail, if you would like.

Q105       Baroness Eaton: You have touched on this as well, but from all that you have observed and discovered can you tell us what the main factors are that account for the registration gap between the different demographic groups? Has it been affected by IER?

Dr Christopher Prosser: The main factor is probably to do with residential mobility. If you move around a lot, and you have to re-register at each house you live in, you are much more likely to fall off the electoral register, simply due to the fact that, once you have registered in one place, you need to keep doing it. If everyone in the population had a completely equal propensity to register to vote, people who stayed put would end up being more registered over the long run, because they would not have to re-register, whereas if you are constantly moving around you will have to re-register. We know that residential mobility is correlated with things such as being young; you move around when you are a student, and you move around private rented houses when you are a young adult, getting out into the workforce. That is why the age correlation with registration is particularly strong.

An additional factor that affects registration within broad demographic groups is, simply, political engagement. People who pay attention to politics are much more likely to be registered and much more likely to vote. This is something that is true not just in the UK now but across a long time period in the UK and, largely, in other countries.

There is an important thing to say about student registration. There is a lot of focus on the effect of IER on students, but some of that emphasis is slightly misplaced, to be honest. Students are less likely to be registered to vote, but that is largely because they are young and they tend to live in private rented accommodation. In general, university students are much more politically engaged than the non-student part of the population. If you pluck out an 18 to 21 year-old at random, a student is much more likely to be registered than a non-student. That is something that perhaps gets slightly lost in conversations about this, which are often dominated by the student/university side of the conversation.

Lord Hayward: You were not present at the previous session.

Baroness Eaton: That is why we are laughing.

Q106       Lord Hayward: You touched on a number of things that you may want to elaborate. Have IER and online registration changed the relationship between eligible voters and registered voters? As I say, to a large extent, you may feel that you have commented on that. Are there implications for the electoral process other than boundaries, or do you want to elaborate on what you said in relation to boundaries?

Dr Christopher Prosser: It depends on how you count it. One of the very confusing things about dealing with this area of research is that if, for completeness, you count the number of people who should be registered and how many people are actually registered, the level looks quite stable. One thing that has been improved by IER is the accuracy of the register, so there are slightly fewer registry entries that should not be there. If you just look at numbers, the numbers on the electoral register are slightly lower than they used to be, compared with the number of people who are eligible, because there used to be more inaccurate entries, which cancelled out the fact that there were missing people on the register.

One consequence of that is when it comes to measuring turnout. We have done research that suggests that we have slightly mis-estimated turnout at elections, because we have not been taking into account that there is a large number of inaccurate entries on the register. If you take that into account, there can be a difference of about 10 points in the number of people registered to vote who actually turn out to vote. IER coming in has changed that.

If you take the raw estimate, it looks as though turnout increased from 2015 to 2017 by 2.4%, but if you take the change in accuracy of the registers into account, the actual change was probably about 2%. There is an extra 0.4% of change, which is just because the registers have become more accurate. I do not know whether you care about 0.4% in turnout; it might be a slightly academic concern. But if we are measuring the health of a democracy by the number of people who are actually taking part in it, getting those numbers right is, we think, quite important.

I have touched on boundaries. The key thing is that, as far as we can tell—again, we do not have perfect data on this—the levels of inaccuracy after IER are not evenly distributed across the country. Places with more inaccurate entries on their register would be advantaged in a re-districting situation. I do not know how you would define it, but they would get more constituencies per actual person in their constituency, because they would be counting people who were not actually there. There are some very serious consequences with regard to the accuracy and completeness of electoral registers.

Q107       Lord Janvrin: We risk going over some of the same ground, but, given that we have had three general elections since IER was introduced, what conclusions can be drawn about how the system is working? You have touched on some of them. Specifically, what deficiencies, in your view, now need to be addressed?

Dr Christopher Prosser: It is probably worth emphasising that some of the things I have talked about definitely apply. One deficiency in the current registration system that I have not talked about so far is that there is no way of checking whether you are registered to vote. As I mentioned before, the increasing volatility of registration is going to place an increasing burden on electoral registration officers. Part of that burden is having to sift through a large number of erroneous registration applications.

It is not particularly surprising that people are making those applications, because it is better to be safe than sorry with that sort of thing, and there is no simple way of checking. There is no timely way of checking whether you are actually registered. You can write to your local authority and say, “Am I registered?” I did that once, and it took quite a lot of time to get an answer. That is one thing that is probably not too difficult and could be added without too much difficulty to our current registration system.

Lord Janvrin: Are you thinking of something national as opposed to local?

Dr Christopher Prosser: It might be something akin to the current system, where there is a centralised point of registration. If there was a centralised point of checking, that would be quite useful. Other countries such as Australia have a system much like that. The practicalities might be quite difficult, because local authorities are the ones that hold the register; there is no national, centralised register. But I understand from colleagues in computer science that there are very complicated ways to do it while still maintaining ownership of data and security, and that sort of thing.

Q108       The Chair: We have talked a bit about boundaries, and I would like to return to that. What impact does a more volatile register have on the work of the Boundary Commission in deciding electoral boundaries? I do not know what is happening on boundaries at the moment. I suppose, if the reduction from 650 to 600 is still in vogue, we may be looking at something that has already been agreed. On the other hand, they might still want 650, but, on the other hand, they might say, “It’s a bit dated anyway”.

Personally, my view is that 95%/105% is too narrow. I could just about cope with 90%/110%, and I think we would have got more sensible boundaries had that been the case. Through our discussions earlier, I just wonder, on any of these criteria, whether things like second homes or students being double-booked have any effect on boundaries as well. It occurs to me that it could have an effect in overenhancing places where people have second homes and, perhaps, overenhancing university towns. It could be the reverse: overenhancing towns that are not university towns but are where the students come from, whichever way you look at that. Do you have anything to say on that area, which seems to me quite a difficult one?

Dr Christopher Prosser: I agree that it is a very difficult question. There is no way of allowing multiple registrations, as the current system does, and having more than one national constituency, that can perfectly solve that problem. You would have to have flexible constituency boundaries, depending on how many people were living at home at a particular time, which is not a practical way of doing things. We have done an estimate of the number of dual-registered people and the number of students registered twice on the electoral roll. I am afraid that I do not have those numbers with me, but I can look them up and give them to you.[1]

The Chair: That would be helpful.

Dr Christopher Prosser: From memory, compared with the problem of inaccuracy in electoral rolls, the problem of dual registration is quite small. What we think is a much more serious issue in drawing constituency boundaries is the fact that there are a large number—10% or so—of entries on the electoral register that are not supposed to be there, because people have died or moved and were not taken off the electoral roll when they should have been. As I mentioned briefly before, we do not think that that is uniformly spread across the country. If some places are more inaccurate than others, they will get a higher proportion of seats than they should.

Without accusing anyone of actually doing this, it creates a bit of a perverse incentive for local authorities to maintain accurate electoral registers. If they have a few more extra people on their register than they should, they will get slightly larger parliamentary constituencies than they should, which might have knock-on consequences. I should say again that I am not sure that anyone has actually done that, but you can imagine that the incentive is there.

We did some calculations, which I have somewhere, of the effect at regional level of inaccuracy. According to our calculations, the levels of inaccuracy and the differences in levels of inaccuracy between regions, would fall outside the 95%/105% threshold and would result in some seats being re-districted between different parts of the country. According to our estimates, London would lose 3.5 seats compared with what the apportionment would be if the registers were accurate, and the north-west would get an extra two seats. The south-east might get an extra seat, and Wales might get an extra seat. There are actual consequences to the levels of inaccuracy and the differences in levels of inaccuracy between places. I agree that dual registration is a conundrum—trying to have two things at once, basically—but compared with the problem of inaccuracy it is quite a small problem.

Baroness Eaton: What about second homes? I do not think all second homes are allowed a second vote.

Dr Christopher Prosser: To be honest, I am very hazy on the details, but I gather it is an issue.

Baroness Eaton: Ryedale does not allow second home owners to exercise their vote, so if it is like that in many places, it will not necessarily skew things. You said it was a smaller effect.

Dr Christopher Prosser: Yes.

The Chair: Are you aware of whether some authorities allow registration or not? I have not heard of it.

Dr Christopher Prosser: I have not heard of differences between different authorities making different rules about who is allowed to vote and who is not. I am a bit hazy on the detail, but I gather that there are particular rules about residency, and that sort of thing. It is not that if you own a holiday home somewhere you can register if you want. There has to be something legitimate; you have to live in two places half the time, or something like that.

Q109       Baroness Pidding: In your written submission, you conclude that the use of the register for determining boundaries fails in both normative and legal standards. What are the implications of that failure, and how might it be addressed?

Dr Christopher Prosser: I should probably preface my statement by saying that I am not a lawyer and my colleagues are not lawyers. We have taken what we think is a fairly common-sense reading of the Act, but we also recognise that common sense and the law do not necessarily see eye to eye all the time.

The Parliamentary Voting System and Constituencies Act defines the electorate, which is the basis for drawing boundaries, as the total number of persons whose names appear on the relevant version of a register of parliamentary electors”. The problem is that the number of entries on the electoral register is not the number of persons whose names appear; it is strictly more than the number of persons whose names appear on the register, as it also includes entries that should not be there, because the people involved are not there, either because they have died or because they have moved. From a legal standpoint, it is not doing what it said on the tin. That is our problem with it. The number that it purports to use is not actually the number.

It is the same problem with turnout, which I covered before. We think about turnout as the proportion of people who are registered to vote who actually voted, but what we are measuring is the proportion of people compared with the number of entries on the register, and the people and the entries are not the same thing. From a normative point of view, the goal of equal constituency boundaries is that it is supposed to equalise the electorate in the sense of who is eligible to vote, not just those who are registered. If registration was perfect, which we know it is not, those would be the same things. But because there are people who are eligible but unregistered, they are excluded from the boundary-drawing process. Depending on which registers we use, that exclusion could be greater or less. If we are using an off-election register, the number of people excluded from the re-districting process will be much higher, and it will be much more systematically biased against young people, private renters and other demographic groups that are less likely to be registered.

It is a very difficult problem to solve, because registration is a very difficult problem to solve. If it was purely up to me, I would probably not use electoral registers; I would probably use the census and make sure that the census had the necessary information for determining who is an eligible elector and who is not, and use that information to redraw boundaries. There is the problem that the census is done only every 10 years, but, at least at the current rate, boundary redrawing does not seem to be happening much faster than that, so it might be less of a problem in practice than it is in theory.

There are other solutions. Other countries have much better registration systems, but they involve population registers that are politically unpalatable in the UK. They certainly result in much more accurate electoral registration; there is no denying it. The boundary drawing question falls out from the accuracy of the electoral registers.

Q110       Lord Janvrin: We are back to turnout, and I think you have answered my question. You said that voter turnout based on electoral registers has been underestimated in Britain because the electoral register is inevitably more than it actually is, as I understand it.

Dr Christopher Prosser: Exactly.

Lord Janvrin: Can you indicate how measurements of turnout might be improved? How can we address the issue, as I understand it from what you have said already?

Dr Christopher Prosser: The key thing when we are interested in measuring turnout is looking at which set of people who might have voted actually voted. There are two ways of doing that. We can look at the registered voter population, just among those who are registered to vote, and what proportion of them turned out. For practical reasons, we are often interested in that.

More substantively, we should be interested in the number of people who are eligible to vote, whether or not they are registered, and what proportion of them actually turned out to vote. In the UK, that is very difficult to do at the moment because, as I said before, that number does not exist in any official capacity. We tried to estimate it for the last couple of elections; we are doing some preliminary work to extend it backwards, and I am sure we will do it again this time. It is far from straightforward, partly because we have quite complicated eligibility rules about people from Commonwealth countries who are allowed to vote, and so on.

The level of detail in the data is not available publicly, because the ONS will not release the fact that there is a small number of people from a particular country who are eligible to vote. You cannot see that in the data, because there is a risk of disclosure. If the ONS was mandated to produce that number, it could give an estimate much more easily than we could, from the outside.

I mentioned the census. If we had the necessary questions, which we probably do, we could use that to look at the voting-eligible population. That would give us a much better idea about how electoral participation is going, compared with the potential pool of voters, and not just among people who are interested enough to register to vote in the first place.

It sounds a bit like academic hair splitting on measurement, but these things have quite important consequences. You might think that the problem is that people are registered to vote but are not voting, which is what a 60% official turnout suggests; it suggests that about 40% of people registered are not bothering to vote. The real number of people who are voting is more like 70%. Okay, 30% is still a large number of people who are not voting, even though they are registered, but it is not quite as bad as 40%.

It also suggests that we are forgetting about the 10% of the population who are so disengaged from politics that they have not even bothered to register. Those people do not get much say in political conversations. We do not know much about them. There is very little from party canvassing and the like, because parties canvass from the electoral register, so their views are not getting fed back into the broader political conversation.

Lord Hayward: Can I ask two questions about turnout? I admit that I am sad enough to include the BES as part of my bedtime reading. First, we have already heard that Ipsos has done a calculation that turnout among the younger population was X. Does your study confirm that the pollsters’ figures are accurate or, because many young people are registered in two different places, is the turnout in fact higher—the level of voting—than initially indicated by the pollsters?

Secondly, on a group you might have studied, I have never understood why, if you receive a postal vote, you do not return it. Even in a general election, something like 20% of postal votes are not returned. Have you done a study on that?

Dr Christopher Prosser: On the first question, I do not know about the 2019 election, because we have not done our assessment of turnout; we need our face-to-face data. In 2017, the Ipsos data and the BES data disagreed by quite a lot about what turnout among young people was. The Ipsos data suggested that there was a very large rise in youth turnout in 2017, compared with 2015, whereas our data suggested that there was not, and that it was probably more or less on a par with what had happened in 2015, with a small rise along the lines of the overall small rise in turnout.

That is not to say that Ipsos is wrong. I should not say that. Rather, it is not to say that we will disagree this time. We may well come to a similar figure. I think 2017 was a particularly hard election for Ipsos to calculate turnout, because it takes the pre-election surveys and then does a complicated reweighting to try to figure out what happened. As we know, in 2017 there was a lot of change over the electoral campaign, which makes that exercise particularly difficult to do.

I am not sure that anyone has looked at postal voting turnout in great detail, to be honest. The usual thing that we think about postal voters is that they vote in much higher numbers than everyone else, so we take it as that: “Oh good, postal voters, they’re an engaged bunch of people”. We do not think about the small number who are not.

In general, a lot of what we know about turnout among people who are regular voters is that often they do not vote for slightly incidental reasons—for example, they were on their way back from work and their train was delayed. That does not quite apply to postal voters, but maybe they were on holiday at the wrong time and were away when their postal vote arrived, and they got back too late. There are probably a lot of mundane reasons for that sort of thing.

The Chair: I think we are there. Thank you very much indeed for coming and giving your evidence to us.

 


[1] The witness subsequently submitted the following information to the Committee:

The Electoral Commission states that people who own a second home can be registered to vote at both addresses if they have a “considerable degree of permanence” at both addresses. Additionally students who have a permanent home address and a term-time address can be registered at both addresses. The Electoral Commission does not collect data on dual-registrants, but we estimate an upper bound of 2.3 million students (Annual Population Survey) and 5.2 million second home owners or 16% of the 2017 register. That figure is a substantial overestimate because both groups contain ineligible residents, and many will not be registered twice.

 

To estimate the number of dual registrants more precisely, the British Election Study Internet Panel asked respondents a series of questions to ascertain whether the respondent is dual-registered. After asking respondents whether they are registered at their current address, they were also asked if they are currently registered anywhere else and what the postcode of that registration is. 7.2% of registered respondents claimed to be dual registered. However, when there postcodes were checked, only 1.6% of registered respondents actually gave a different postcode, which would be about 750,000 people.