Representation of the People Bill — Written evidence submitted by Dr Sofia Collignon, Director of the Mile End Institute and Reader in Comparative Politics at Queen Mary University of London (RPB51)

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.1
Dr Sofia Collignon
Director, Mile End Institute
Queen Mary University of London
s.collignon@qmul.ac.uk

Written Evidence to the Public Bill Committee
Representation of the People Bill 2024–26
April 2026

About the submitter
I am Director of the Mile End Institute and Reader in Comparative Politics at the School of
Society and the Environment , Queen Mary University of London. I am a leading expert in
political representation , with a particular focus on the impact of harassment, abuse and
intimidation (HAI) on political candidates in established democracies. My research has
directly shaped policy and parliamentary debate in this area over the past decade.
My evidence draws on a sustained programme of research including the Representative Audit
of Britain (RAB) surveys, which track the experiences of all parliamentary candidates across
successive general elections (2015, 2017, 2019, 2024). I have previously submitted evidence to
the Committee on Standards in Public Life (2017), the Public Administration and
Constitutional Affairs Committee (2025), the Speaker’s Conference on the Security of MPs,
Candidates and Elections (2024–25), and the Task Force on Safe Participation in Political Life
(Ireland). I regularly advise the Local Government Association on resources and guidelines to
support councillors in dealing with HAI in public life.
This submission addresses the provisions in Part 6 of the Bill relating to the intimidation of
candidates, campaigners and election staff, and the provision removing the requirement for
candidates acting as their own election agents to publish their home a ddresses. It draws on
published research and on preliminary findings from the local and general election candidate
surveys.
Summary
My evidence supports the Bill’s direction of travel on candidate safety but argues that the
measures proposed are insufficient on their own. The Bill addresses the downstream problem
of how offences are prosecuted and sentenced. It does not address the upstream problem: that
the threat of HAI deters diverse candidates from standi ng in the first place, filtering out
exactly the people whom a representative democracy most needs to attract. The following
summary sets out my core arguments.
• HAI is worsening and the 2024 election was no exception. Preliminary results from
the 2024 RAB Survey show that 85% of 2024 general election candidates experienced
at least one form of violence during their campaign. Despite awareness campaigns and
legislative changes since 2019, the problem is intensifying, not diminishing.

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.2
• The gendered and intersectional dimension is severe and must be named explicitly
in legislation. Women, and particularly women from ethnic minority backgrounds
and LGBT+ women, experience significantly higher rates of HAI. This is not an
incidental feature of the problem: it is the defining characteristic of how HAI operates
as a tool of democratic exclusion.
• The Bill’s home address provision is welcome but far too narrow. Under current law,
candidates can already opt to withhold their home address from the ballot paper and
nomination documents, and by 2024 three -quarters did so. The Bill closes only a
residual gap for candidates who act as their own election agents. The Government
should remove the home address requirement entirely for all candidates, replacing it
with a mandatory correspondence address such as a PO Box or campaign office.
• Tougher sentences alone will not change the composition of the candidate pool. The
aggravating factor provision sends a welcome signal, but my research shows that HAI
functions primarily as a deterrent before the election, not simply a harm during it.
Legislative reform must be complemented by upstream measures that address the
conditions under which people decide whether to stand at all.
• Social media platforms require a clearer statutory duty. Online HAI has grown faster
than any other form of abuse in every election cycle. The Bill does not address the
structural conditions that allow it to flourish.
1. The scale and trajectory of HAI
Evidence from the Representative Audit of Britain surveys demonstrates a consistent and
statistically significant deterioration in the safety environment faced by parliamentary
candidates at UK general elections.
In 2017, 38% of all candidates reported experiencing some form of HAI during the campaign.
By 2019, that figure had risen to 49% , an increase of 11 percentage points, statistically
significant at p≤0.05. Preliminary results from th e RAB Survey, covering the 2024 general
election, paint an even starker picture. When candidates were asked directly whether they
had experienced HAI, 32% said yes. However, when presented with a structured list of
specific experiences of physical, psycholo gical and online violence, 85% of candidates
indicated that they had experienced at least one form during their campaign. This divergence
between self-reported and prompted measures is itself important evidence: many candidates
do not recognise what they experience as HAI, which reflects a process of normalisation that
is deeply corrosive to democratic health.
Analysis of the Statements of Persons Nominated (SOPNs) published by returning officers
provides independent corroboration of candidates’ growing security concerns. In 2010, when
candidates were first given the option not to publish their home address on b allot papers,
approximately one-quarter withheld their address. By 2024, three-quarters did so. This is one
of the most dramatic and unambiguous trend lines in our data.
Polling conducted by Savanta for the Mile End Institute in June 2024 revealed troubling public
attitudes. Among a representative sample of Londoners, 18% considered it acceptable to
follow political figures on the street, 14% found it acceptable to loiter near political figures’
offices or homes, and 16% thought it acceptable to publish political figures’ personal

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.3
information online without their consent. These attitudes are not fringe positions: they
represent a substantial minority of the public who condone behaviour that constitutes
intimidation in law and in practice.
2. The gendered and intersectional character of HAI
My research consistently finds that HAI is not experienced evenly across the candidate pool.
It is concentrated amongst those groups who are already underrepresented in Parliament.
This pattern reflects HAI’s function as a mechanism that enforces the existing social
composition of political elites by imposing higher costs on those who deviate from it.
Gender
In 2017, 45% of women candidates reported experiencing HAI, compared with 35% of men.
By 2019, 58% of women reported HAI, compared with 45% of men. The gender gap widened
from 9 percentage points in 2017 to 13 percentage points in 2019. For women candidate s in
2019, experience of HAI whilst campaigning had become the statistical norm rather than the
exception. In 2024, women candidates indicated to feel significantly more unsafe during the
campaign than men candidates.
The psychological impact on women is commensurately severe. In 2019, 48% of women
candidates reported being very concerned because of HAI (compared with 27% of men), 45%
reported being very annoyed (29% of men), and 34% reported being very fearful (14% of men).
Approximately three in every four women experienced some level of fear whilst campaigning.
The Bill’s stated rationale acknowle dges the disproportionate impact on women, but this
acknowledgement has not been translated into specific gendered provisions. I strongly urge
the Committee to address this gap.
Ethnicity
The data on ethnicity reveals an intersectional pattern for concern. In 2017, 37% of minority
ethnic candidates reported experiencing HAI. By 2019, that had risen to 55%, a statistically
significant increase (p≤0.05). The combined effect of gender and ethnicity is especially stark:
63% of minority ethnic women candidates reported experiencing abuse in 2019, compared
with 45% of white women, 38% of ethnic minority men and 34% of white men. Ethnic minority
women are by a considerable margin the group most exposed to HAI in the candidate pool.
Sexual Orientation and Gender Identity
LGBT+ candidates also face disproportionate rates of HAI. Overall, 45% of LGBT+ candidates
reported experiencing abuse in 2019. The intersection of sexual identity and gender is again
significant: 58% of LGBT+ women reported HAI, compared with 39% of LGBT+ men and 46%
of heterosexual cis women.
These patterns collectively demonstrate that HAI functions as a barrier to diverse political
representation. Legislation that treats intimidation primarily as a sentencing matter, without
addressing the structural conditions that concentrate it on already-marginalised groups, will
not alter this dynamic. I recommend that the Bill explicitly require courts to treat the protected
characteristics of the victim as a factor in the operation of the new aggravating provision.

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.4
3. The home address requirement
The Bill’s home address provision is welcome but far too narrow. To understand why, it is
necessary to explain the current legal position in full.
Following a 2008 consultation prompted by security concerns raised in Parliament, candidates
in UK general elections were given the option, from the 2010 election onwards, to withhold
their home address from the ballot paper and the Statement of Persons No minated (SOPN),
replacing it with the name of their constituency. However, this opt-out contained a structural
gap. Every candidate must appoint an election agent — the person legally responsible for
managing their campaign finances. Most candidates from l arger parties appoint a separate
person as their agent. Many smaller party candidates and independents, however, act as their
own election agent to avoid cost and administrative complexity. Under existing rules, election
agents’ addresses must be published in the Notice of Election Agents, and candidates acting
as their own agent had no option to withhold that address even if they had successfully
withheld it from the ballot paper and SOPN. Those candidates were therefore de-anonymised
through the back door of the nominations process.
The Bill closes this specific gap by removing the requirement for candidates who act as their
own election agents to publish their home address in the Notice of Election Agents, allowing
them instead to provide a correspondence address. This is overdue and I welcome it.
Yet my research with Wolfgang Rüdig, based on analysis of all SOPNs across five general
elections (Collignon and Rüdig, 2024), documents address withholding accelerated sharply
across all parties since the murder of Jo Cox in 2016 . Open ended answers confirm that
security for themselves and their families is the primary reason candidates withhold their
address. This is not an administrative preference; it is a safety response.
This trend also carries electoral consequences that compound existing inequalities. My
research demonstrates that the ballot paper address had served as one of the few ways
candidates could signal local connection to voters, and research on voting behaviou r shows
that voters are more likely to support candidates who live close to them. As candidates , and
disproportionately women, withdraw this information to protect themselves, they lose a tool
for communicating localness that party selectorates and some voters value. Security concerns
therefore interact with candidate selection dynamics in ways that place diverse candidates at
further disadvantage.
The Bill’s reform protects candidates who were already trying to protect themselves but were
caught by the election agent loophole. It does nothing for the remaining quarter of candidates
who still publish their home address and who may simply be less secu rity-conscious, less
well-advised, or standing for the first time. The Government should go further by removing
the home address requirement entirely for all candidates across both the ballot paper and all
nomination documents, replacing it with a mandatory correspondence address such as a PO
Box or campaign office address. This would standardise in law what is already the
overwhelming practice, protect those who have not yet taken steps to withhold their address,
and remove the security risk from the nomin ations process entirely. This was a

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.5
recommendation of the Speaker’s Conference and is directly supported by the trajectory of
the data.
4. The upstream deterrence problem
The Bill’s approach to HAI is principally downstream as it strengthens sentencing for offences
committed against candidates, campaigners and election staff. This is valuable. However, my
research demonstrates that HAI also operates powerfully as a deterrent that operates before
the election, shaping who decides to stand in the first place.
In Collignon and Rüdig (2021 ) we show that HAI increases the effective cost of female
representation. Candidates who experience HAI adapt their behaviour: they reduce face -to-
face contact with voters, change their online presence, and in many cases decide not to stand
again. Women are significantly more likely than men to make these adjustments. The practical
effect is that HAI harm individual candidates and shapes the future composition of the
candidate pool in favour of those who are either less exposed to abuse or better resourced to
deal with it.
Research on MPs’ staff adds a further dimension. Data from the Speaker’s Conference inquiry
found that 72% of staff said their experience of witnessing abuse or intimidation whilst
working for an MP made them less likely to stand for election themselves. Women staff were
more likely than men to report this deterrent effect (77% versus 58%). The talent pipeline for
future candidates is thus also being eroded by the current environment.
The Bill does not address these upstream dynamics. Effective legislative reform requires
complementary measures that reduce the baseline level of HAI rather than simply increasing
the punishment once it has occurred. The Committee should consider whether provisions on
party accountability and support for candidates who are targeted would more directly
address the deterrent effect of HAI on candidate diversity.
5. Online HAI and platform accountability
Online harassment has grown faster than any other form of abuse across every election cycle
in our data. Between 2017 and 2019, the share of candidates reporting abuse on social media
rose from 29% to 45%, an increase of 16 percentage points. Abuse by emai l rose from 23% to
35%. The gender differential is large: in 2019, 54% of women and 40% of men reported being
targeted by abuse on social media. The 2024 general election saw a further and alarming
deterioration, about 60 % of general election candidates stated that they had suffered from
online HAI, and 40 % stated that during the campaign someone had distributed misleading
information about them.
The Bill does not contain provisions that directly address the structural conditions on social
media platforms that allow this to flourish. The Electoral Commission has itself stated that
further measures may be needed to ensure candidate safety and has ca lled for a clearer
overarching statutory duty on social media platforms operating in the UK. The Speaker’s
Conference made an equivalent recommendation and specifically called on the Government
to consider mandating Ofcom to produce an elections code of pr actice for social media
platforms, noting the feasibility of including this requirement in the elections bill it had

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.6
committed to bring forward. That recommendation is directly unanswered in the Bill as
introduced. This is a significant omission.
A particular and growing concern is the use of generative AI to produce deepfake content
targeting political candidates, especially women. The Speaker’s Conference heard extremely
distressing evidence about the scale of deepfake pornography targeting femal e MPs and
candidates at the 2024 election. This is a gendered harm of the first order as it is designed to
humiliate, silence and deter women from political life. The existing legislative framework is
inadequate. The Electoral Commission has recommended that section 106 of the
Representation of the People Act 1983, which criminalises false statements about a candidate’s
personal character or conduct, should be amended to make explicit that it covers deepfakes. I
support that recommendation and urge the Committee to address it through this Bill.
The current regulatory framework depends primarily on the willingness of social media
companies to cooperate, which produces significant disparities between platforms in how
online HAI is handled. These companies operate on a transnational scale, which raises serious
questions about whether the safeguarding of national democratic processes should be
entrusted to private entities with no democratic accountability. The legal threshold required
to seek protection from online abuse is, in practice, too high for most candidates to meet. Social
media companies take too long to remove harmful content, and the law requires victims to
document a course of conduct over a sustained period before relief is available. These
structural deficiencies are left unaddressed by the Bill as currently drafted.
Recommendations
1. Strengthen the aggravating factor provision to include protected characteristics.
The new statutory aggravating factor for offences targeting candidates, campaigners and
election staff should explicitly require courts to consider the protected characteristics of the
victim under the Equality Act 2010 as a further aggravating element. Th is is consistent with
how courts already treat hate crime and would reflect the documented evidence that women,
ethnic minority candidates and LGBT+ candidates are disproportionately targeted.
2. Remove the home address requirement entirely for all candidates, replacing it with a
mandatory correspondence address.
The Government should go further by removing the home address requirement from the
ballot paper and all nomination documents for all candidates, replacing it with a mandatory
correspondence address such as a PO Box or campaign office. This was a recommendation of
the Speaker’s Conference and is directly supported by the evidence that three -quarters of
candidates in 2024 chose to withhold their address on safety grounds. The current provision
is far too narrow.
3. Introduce systematic data collection on HAI as a democratic health indicator.
The Bill should include a requirement for the Electoral Commission to collect and publish
systematic data on the experiences of candidates during election campaigns, including
experiences of HAI, in every post-election report. This data is currently collected by academic
researchers and organisations in scattered form rather than systematised and organised by
public institutions. Embedding it in the official reporting framework would create an

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.7
authoritative public record and enable measurement of the effectiveness of protective
measures over time.
4. Place a duty on political parties to report on candidate safety.
The Bill should require registered political parties to report to the Electoral Commission on
the measures they have in place to support candidate safety and to address HAI within their
selection processes. Parties are the key gatekeepers to political office and the frontline bodies
responsible for supporting candidates during campaigns. At present, there is no
accountability mechanism for party-level performance on this issue.
5. Include in the Bill a provision mandating Ofcom to produce an elections code of practice
for social media platforms.
The Speaker’s Conference specifically recommended that the Government consider
mandating Ofcom to produce an elections code of practice for social media platforms and
assess the feasibility of including this requirement in the elections bill. Elections are
identifiable high-risk periods for online abuse. A statutory code of practice would establish
consistent minimum standards across platforms for the removal of illegal content targeting
candidates, require proactive rather than reactive moderation during campaign periods, and
create an enforceable accountability mechanism. Voluntary cooperation has demonstrably
failed. The Government should accept the Speaker’s Conference recommendation and
introduce the necessary provision through this Bill.
6. Amend section 106 of the Representation of the People Act 1983 to explicitly cover
deepfakes and designate electoral deepfakes as a priority offence under the Online Safety
Act.
Whilst the Government has legislated against non-consensual sexual deepfakes more broadly,
these provisions address general intimate image abuse and do not specifically cover the
democratic harm caused when deepfakes are deployed against candidates during elections.
The Speaker's Conference Second Report documented extremely distressing evidence about
the scale of deepfake pornography targeting female MPs and candidates at the 2024 election.
The Bill should amend section 106 of the Representation of the People Act 1983 to make
explicit that it covers deepfake content, closing the ambiguity identified by the Electoral
Commission. The Government should also consider whether section 106 offences should be
added to the list of priority offences under the Online Safety Act , requiring platforms to act
proactively to prevent candidates being targeted by such content.

Dr Sofia Collignon | Evidence to the Representation of the People Bill 2024–26 | p.8
Key references

Peer-reviewed publications
Collignon, S. and Rüdig, W. (2024). 'Who Runs for Office? Understanding Candidate Diversity,
Safety and Localism in the UK General Election 2024'. The Political Quarterly. DOI: 10.1111/1467-
923X.13487
Collignon, S. (2024). 'Addressing Barriers to Women's Representation in Party Candidate Selections'.
The Political Quarterly. DOI: 10.1111/1467-923X.13397
Collignon, S. (2023). 'Harassment and Intimidation of Parliamentary Candidates in the United
Kingdom'. In Bjarnegård, E. and Zetterberg, P. (eds.), Gender and Violence against Political Actors .
Philadelphia, PA: Temple University Press.
Collignon, S., Campbell, R. and Rüdig, W. (2021). 'The Gendered Harassment of Parliamentary
Candidates in the UK'. The Political Quarterly, 93(1), pp.32–38.
Collignon, S. and Rüdig, W. (2021). 'Increasing the Cost of Female Representation? The Gendered
Effects of Harassment, Abuse and Intimidation towards Parliamentary Candidates in the UK'. Journal
of Elections, Public Opinion and Parties, 31(4), pp.429–449.
Collignon, S. and Rüdig, W. (2020). 'Harassment and Intimidation of Parliamentary Candidates in
the United Kingdom'. The Political Quarterly, 91(2), pp.422–429.

Parliamentary and policy submissions
Collignon, S. and Rüdig, W. (2025). Written evidence submitted to the Public Administration and
Constitutional Affairs Committee on the administration, process and conduct of the 2024 General
Election. 26 January 2025.
Collignon, S., Rüdig, W. and Banducci, S. (2025). Written evidence submitted to the Speaker's
Conference on the Security of Candidates, MPs and Elections. Reference SCS0023. 7 February 2025.
Collignon, S. (2025). Written evidence submitted to the Speaker's Conference on the Security of
Candidates, MPs and Elections. Reference SCS0041.
Collignon, S. (2025). Oral evidence given to the Speaker's Conference on the Security of Candidates,
MPs and Elections.
Collignon, S. and Rüdig, W. (2017). 'Inappropriate Behaviour: Experiences of 2017 Parliamentary
Candidates'. Evidence submitted to the Committee on Standards in Public Life, 19 September 2017.

Dr Sofia Collignon | April 2026 | s.collignon@qmul.ac.uk