Analysis of electoral registration data

Overall

At the beginning of the canvass, all registers are matched against Department for Work and Pensions data and EROs then use the results to help them allocate properties to routes (as explained above). Following this national data matching exercise, our data shows that 19.5 million properties were allocated to Route 1 and 9.3 million to Route 2.

In addition, EROs have the discretion to carry out data matching using local datasets such as Council Tax in order to aid with the process of allocating properties to routes.

The majority of EROs carried out local data matching, with Council Tax being the most popular dataset used and respondents citing that this data was the most easily usable and accurate. The next most popular piece of data was housing benefit data.

The local data matching step had a significant impact on the total figures for route allocations as shown in the table below.

The proportion of properties allocated to Route 2 (26% after the local data matching stage) is broadly in line with that predicted in the UK, Scottish and Welsh governments’ statement of policy for canvass reform, which anticipated that around a quarter of properties would have to go down the Route 2 process nationally.

Whilst we are aware of some EROs successfully using a range of local data for data matching, we also received feedback during, and following, the 2020 annual canvass, indicating that other datasets were more difficult to access. In response to this, we have undertaken work to expand our guidance on local data matching ahead of the 2021 canvass. We have included additional guidance on the ERO’s power to obtain data, and have also highlighted some practical considerations for how to work with other teams in local authorities to access data. Additionally, over the course of this year’s canvass, we plan to work with EROs to identify examples of good practice that we can share to support other EROs with effectively accessing and using a variety of local data sources.&

Variation in route allocation

Generally speaking, the allocation of properties down each route suggests a relatively stable picture across the nations of Great Britain. Within England the allocations were similar across regions although, perhaps unsurprisingly, London recorded the lowest proportion of Route 1 allocations (66.4%). This reflects the greater degree of population movement in London, with more properties failing to match with existing data and being allocated to Route 2 in order to pick up changes in household composition.

The table below shows the proportion of properties allocated across each route for each region:

The reforms to the canvass were designed to allow EROs to ensure their resources could be directed at households where changes needed to be made to the registers. In previous years, because all households were required to be chased for a response, even where there were no changes to report, significant effort and resource was being expended on households where the composition had not changed, with no benefit to the quality of the registers. 

In the 2020 canvass, around 8.8 million responses were received from households across all canvass routes. This compares to 23.8 million responses to Household Enquiry Forms at the 2019 canvass. This lower overall response suggests that the increasing targeting of activity in the reformed canvass was successful in reducing unnecessary response. 

Also, as anticipated in the UK’s governments’ canvass reform policy, there was a significantly higher response rate from those households in Route 2, where data matching had indicated a greater likelihood of a change in residents, compared to Route 1 (66% compared to 18%). Route 2 is also the route where EROs are required to chase for a response, by sending reminders and by trying different contact methods.   

While this balance of responses between Routes 1 and 2 is positive, it is important to note that although all households in Route 2 are required to respond, more than one third did not, meaning details for these households have not been updated when the data matching suggests updates were required. This will be an area to continue to monitor in future canvasses, and EROs and the Commission will be able to use the new ERO performance standards framework to support an analysis of the data to help understand and, where appropriate, address the reasons for this.

The level of response for Route 2 varied significantly, as shown in the table below, with Scotland reporting the lowest response for Route 2 at 50% and England the highest at 67%. Within England, the North East was the region with the lowest response rate to both Route 1 and Route 2.

Households that respond to the canvass can record a major change (e.g. showing a potential new elector is resident), a minor change (e.g. amending the name of an existing elector) or no change (i.e. confirming existing details).

A total of 2.6 million households who responded during the 2020 canvass reported a major change in their registration details. This represented 30% of all 8.8 million responses received across each route. These major changes were distributed by route as follows:

• Route 1 major changes: 791,720 (30%)
• Route 2 major changes: 1,801,283 (69%)
• Route 3 major changes: 26,835 (1%)

This again shows that overall those properties in Route 2 that were identified by data matching as most likely to need changes to residents’ registration details, reported the largest proportion of changes during the canvass. However, nearly a third of the major changes reported came from properties allocated to Route 1 and where the data had indicated no change was needed. 

It is important to bear in mind that, of all the households allocated to Route 1 (21.4 million), the proportion reporting a major change (791,000) is small (around 4%). However, these figures do highlight that the matching process does not accurately identify all properties where changes will be needed. We will use the new ERO performance standards to support our work with EROs throughout the 2021 canvass to understand further the scale and impact of this.

A further 6.1 million households reported no or minor changes to their registration details, representing 70% of households who responded.  In order for this data to be more useful in future, we will work with electoral management software (EMS) providers to split this category up into no change and minor changes.

Canvassing activity with properties does not directly result in registrations. For example, where a household reports that a potential new elector is resident that individual still needs to submit an application to become registered.  

In the year to the publication of the annual register in 2020, a total of 5.3 million applications to register to vote were received. Of these applications, 2.8 million (53%) were received during the canvass period.
The annual volume of applications tells us about whether the registers are picking up expected levels of change in the population. While the number of applications in 2020 (5.3 million) was much lower than in 2019 (7.5 million) it was in line with 2018 (5.3 million).  The lower level of registration activity in 2020 is likely linked to the fact that no elections were held after being postponed due to the coronavirus pandemic. In 2018 there were no high profile national polls, with elections held in only some local authority areas in England. 

The proportion of applications in the canvass, compared to the rest of the year, tells us something about the relative effectiveness of the canvass itself in driving new registrations. In 2020, the proportion of applications during the canvass (53%) was lower than in 2019 (66%) and 2018 (62%), but much higher than in 2017 (37%).  

The position of 2020 relative to 2019 is likely explained by the 2019 UKPGE registration period taking place during the canvass in 2019, driving up applications. In 2017, the UK general election took place in June, which had the effect of driving applications outside the canvass period. 

The lower proportion of applications in 2020 compared to 2018 could be, at least in part, the result of a reduction in personal canvassing in 2020. At this stage it is not possible to disentangle the extent to which that reduction can be attributed to pandemic restrictions compared to the reformed canvass process. We will therefore continue to monitor this data in future years. 

Data on additions to, and deletions from, the registers provides a good understanding of whether registration activity is keeping pace with population change. Data relating to additions and deletions form a core part of the new performance standards to support EROs with measuring and monitoring the impact of their registration activities.

Changes to the registers are generated from several sources. For example, once applications have been received and processed, EROs are required to add new eligible electors; EROs can also use data gathered from households and other existing data sources to identify records relating to home movers and those who have died, which they will need to take steps to remove.  As population mobility varies across the country, so does the scale of the challenge of maintaining the accuracy and completeness of electoral registers faced by EROs. 

There were a total of 4.7 million additions to the registers and 4.9 million deletions from the registers in the year from publication of the 2019 annual registers to the publication of the annual registers for 2020. This represents 10% of the registers being added and 10% removed over the year.

The number of additions in 2020 (4.7 million) was lower than that recorded in 2019 (6.3 million). The number of additions recorded in the canvass period (2.7 million) was also lower than in 2019 (3.9 million).  However, this is again likely to be linked to elections - 2019 saw three sets of electoral events, including a UKPGE where much of the registration activity was during the canvass period, while 2020 had no polls due to COVID-19. 

The number of deletions in 2020 (4.9 million) was very similar to that recorded in 2019 (4.9 million). The number of deletions during the canvass period in 2020 (3.1 million) was also very similar to 2019 (3.0 million).

The proportion of additions and deletions across routes is largely in line with expectations, with more changes being picked up within the Route 2 households. Although as with the level of household major changes, which are directly related to additions and deletions, the data highlights that just over one third of the changes made to the registers were within households where the data matching did not suggest any change was needed. It is not possible from a single year of data to judge if this will continue to be a feature of the reformed canvass and, if so, what the longer term impact of it would be. This is an area we will continue to monitor.
 

The number of additions and deletions as a proportion that took place during the canvass over 2020 varied across Great Britain.  

The table below shows that the proportion of additions (56%) during the canvass in 2020 is down on 2019 (62%) and proportion of deletions during the canvass during the annual canvass in 2020 (64%) up on 2019 (61%).

The level of additions and deletions matters because it tells us whether the accuracy and completeness of the registers is likely to be increasing or decreasing. As with applications, the pattern of annual comparisons between additions and deletions may be mostly explained by external factors (particularly the timing of significant national elections and, last year, the pandemic). However, it is not possible to draw conclusions after only one year of the new canvass on what impact, if any, the changes in the canvass process have had. We will continue to collect and analysis this data in order to monitor the trends over time.