Key findings

Evidence from the pilot scheme indicates that there was a headline increase in voter confidence and in perceptions of the safety of voting in the pilots. We do not see the same change in non-pilot areas with May elections.

However, we should be cautious in drawing clear conclusions about what the impact on confidence of a Great Britain-wide scheme (based on any of these models) would be.

This is partly because the ten pilots are not representative of Great Britain as a whole and partly because the changes in attitude are not consistent within each pilot model.

People who voted in polling stations in the pilot scheme, and therefore experienced the process of showing identification, were more confident in the security of the voting system than non-voters in those areas.

Overall perceptions of electoral fraud

People in the pilot scheme were significantly less likely to say that fraud took place at the elections than those in other areas with elections in England in May 2019.

How much electoral fraud or abuse, if any, do you think took place at the 2 May elections?

Chart showing perceptions of electoral fraud at the May elections
Perceptions of electoral fraud at 2 May elections. Figures may not equal 100% due to rounding.

However, this is only a snapshot of opinion and we have also considered how opinions may have changed before and after 2 May in both pilots and those areas with elections but no pilots.

We asked respondents to say how much of a problem they think electoral fraud is on a scale of 1-5 (with 1 being not a problem and 5 being a serious problem). In the pilot scheme there was a notable increase in the proportion of respondents saying that electoral fraud is not a problem between January and May 2019 (30% compared with 39%).

This change was not seen in non-pilot areas, where the number of people who thought that electoral fraud is not a problem remained relatively stable at 27% pre-poll and 26% post-poll.

The clearest increase was in the photo only pilot model (moving from 13% to 32% saying it was not a problem) and in the mixed model (13% to 27%), while in the poll card model the change was more polarised with increases in both those saying it is not a problem and those saying it is a serious problem.

However, we should be cautious in drawing clear conclusions about how the experience of a specific pilot model may have affected levels of confidence. Although this data appears to present a clear pattern, the changes between before and after polling day are not always consistent across the individual pilots. This is particularly the case for the mixed model. For example:

  • In the photo model, in both Pendle and Woking there were increases in the proportion of people after polling day saying fraud is not a problem, and a decrease in those saying that it is. The change was significantly more pronounced in Woking.
  • In the mixed model, we only have data for four of the five pilots (because Craven joined the pilot scheme too late to be included in the January public opinion fieldwork). In two pilots (Braintree and North Kesteven) there was an increase after polling day in the proportion saying ‘not a problem’ and a decrease in ‘a problem’. However, in the other two pilots (Broxtowe and Derby) the results were closer to the poll card pilot (a polarisation to each end of the scale).
  • In the poll card model, one pilot (Mid Sussex) showed no change in attitudes at all after polling day. Data for the two other pilots showed, to varying levels, a polarisation of attitudes towards either end of the 1-5 scale.

The data does not help us to explain why we see these variations across pilots. We cannot know if we have seen genuine effects from the pilot models, or if the shifts are due to particular issues or concerns in each individual pilot.

Voting in a polling station

We asked respondents how safe from fraud and abuse they thought it was to vote at a polling station.

Overall, in the pilot scheme, the proportion of people saying it is safe did not change significantly between January and May 2019. However, the strength of confidence increased slightly between January and May; the proportion of people saying it was very safe increased from 48% to 55%.

This change varied in scale across the individual pilots but was seen to some extent in all but two (Derby and Mid Sussex). Respondents living in non-pilot areas were less likely to say that voting in a polling station was very safe (39%).

Our research also found that people who voted in the pilots, and therefore experienced showing their identification, were more likely to say voting in a polling station was very safe from fraud and abuse than those who didn’t vote (61% compared with 50%).

Confidence in the voting system

We asked respondents if the requirement to show identification at the polling station made them more or less confident in the security of the voting system.

Across all pilot models the majority of respondents said it would make them more confident or make no difference, although the balance between the two varied by model.

In the poll card model, a majority (57%) said that an identification requirement would make them more confident in the security of the voting system, in the mixed model, results were evenly split (46% each) and in the photo only model the majority said it would make no difference (51%).

However, in each case a substantial proportion felt it would improve their confidence in the security of the system.

Did the requirement for voters to show identification at the polling station make you more or less confident in the security of the voting system, or did it make no difference?

Chart showing levels of confidence in the voting system as a result of ID requirement
Levels of confidence in the security of the voting system as a result of the ID requirement.