Abstract
Can losers of once-in-a-generation referenda ever regain faith in democracy? Previous research indicates that elections and referenda can both lower losers’ satisfaction with democracy and increase that of the winners. As long as people win some in addition to losing some, there appears to be little concern that satisfaction will polarize along partisan lines. However, for once-in-a-generation referenda such as Britain’s EU Referendum, there is little chance for losers to chalk up a compensatory win, potentially exacerbating and solidifying the winner-loser gap. This research note extends previous research into the consequences of Britain’s EU Referendum on satisfaction with democracy by analyzing 5 years of panel data across the Brexit period—from the referendum announcement through to the UK’s exit from the EU. Our analyses demonstrate dramatic swings in winners’ and losers’ satisfaction with democracy across those 5 years, often corresponding with Brexit-related events. Most tellingly, after the United Kingdom finally left the EU, there is a notable, though modest, continuation of the winner-loser gap initially triggered by the EU Referendum outcome—suggesting once-in-a-generation referenda can durably fracture democratic satisfaction in ways regular elections do not. These findings have implications for democracies contemplating major constitutional or sovereignty referenda.
Introduction
This research note adds to the small but growing literature on the durational impact of being a “winner” or “loser” in elections and referenda. Focusing on the case of Brexit in the United Kingdom, we extend the temporal scope of analyses of this case study well beyond previous studies. While existing research on the Brexit referendum tracked satisfaction with democracy (Dunn and Thornton, 2022) for 6 months post-referendum, our study tracks satisfaction changes across the entire Brexit implementation cycle—from the referendum announcement through the UK’s actual departure from the EU (April 2016 to May 2021). Specifically, we examine how long the noted changes in satisfaction with democracy caused by the Brexit vote continued, whether the satisfaction of Remain supporters recovered at all, and whether the satisfaction of Leave supporters returned to its previously low levels.
The note adds three contributions to the referendum and democratic satisfaction literatures. First, we provide the first complete longitudinal analysis of a referendum winner-loser cycle, revealing that referendum-induced changes in satisfaction follow a complex temporal pattern with multiple phases. Second, we demonstrate that referendum dynamics differ fundamentally from electoral patterns, with extreme volatility in satisfaction and winners experiencing larger satisfaction declines during implementation delays than losers experienced from initial defeat. Third, we introduce the concept of “implementation uncertainty” as a key theoretical moderator, providing a framework that applies broadly to referendum contexts where implementation requires complex political processes.
Background and Expectations
Electoral competition creates winners and losers through granting one or more political parties control of government and relegating the remaining parties to opposition to question and scrutinize the government. The outcome of such competition impacts people’s satisfaction with the political system itself—not only due to evaluations of the processes of decision-making, but also the more instrumental outcomes that result from those processes (cf., Singh and Mayne, 2023). Electoral winners (those whose vote choice is elected or selected into government) are more satisfied with democracy in their country compared to electoral losers (those whose vote choice does not make it into government) (Halliez and Thornton, 2023; Singh et al., 2012).
This ‘winner-loser gap’ is moderated by perceptions of ‘voice’ in the electoral process. Previous research argues that when electoral losers feel listened to, their satisfaction with democracy is stronger than if they feel ignored or excluded (Merkley et al., 2019; Ridge, 2024). Evidence indicates that the winner-loser gap is smaller: in consensual political systems, where losers are more satisfied with losing and winners less satisfied with winning than in majoritarian systems (Anderson and Guillory, 1997; Bernauer and Vatter, 2012; though see Singh, 2014); in systems with higher quality elections (Dahlberg and Linde, 2016; Nadeau et al., 2023); in systems perceived to more accurately reflect vote outcomes (Ridge, 2024); and when election campaigns sufficiently cover issues important to voters (Merkley et al., 2019). All of this indicates that voice in the electoral process can lessen the instrumental loss resulting from one’s vote choice failing to gain a seat in government.
Voice matters because we want to feel important to those who are important to us (Baumeister et al., 2007; MacCoun, 2005). So, while we may accept that we will, in repeated electoral competitions, lose from time to time, we still wish for our opinions to be influential, rather than just being heard (Esaiasson et al., 2016; Hobolt et al., 2021). Repeated losses are likely to undermine any perception we may have that we are being listened to and therefore raise concerns that the political system is not as fair as we may have once thought (Daniller and Mutz, 2019; Kern et al., 2024). For contentious referenda, unlikely to repeat themselves in the short term, losing shows a pronounced and durational effect on perceptions of fairness (van der Eijk and Rose, 2021), particularly if those referenda create enduring political identities that transcend specific policy preferences (Hobolt et al., 2021).
As electoral losers tend to show decreased satisfaction with democracy, so electoral winners generally show increased satisfaction with democracy (Blais and Gélineau, 2007; Halliez and Thornton, 2023). However, winners vary in the degree to which winning increases their satisfaction. Compared to first-time winners, repeated winners’ satisfaction is conditional on how well their ideological preferences align with those demonstrated by the elected government (Curini et al., 2011), and winners who vote according to their ‘pure’ political preferences are more satisfied with democracy than those who vote tactically (Singh, 2014). Regardless, winners’ increased satisfaction tends to remain at increased levels for years following the election (Loveless, 2021).
This research picks up where previous research on the Brexit referendum (Dunn and Thornton, 2022; van der Eijk and Rose, 2021) leaves off. Whereas both Dunn and Thornton and van der Eijk and Rose examined the time period immediately around the referendum, an ~8-month period for Dunn and Thornton and ~1 year period for van der Eijk and Rose, we extend this previous research to consider the impact of the Brexit vote over the ~5-year period that covers the entire Brexit period, from the start of the referendum campaign in 2016 to the UK’s departure from the EU in 2020. We also account for the potential confounding effects from elections that occur within the timeframe of the data (in 2017 and 2019).
We expect:
H1: Remain voters’ satisfaction with democracy will decrease following the EU referendum (in Wave 9).
H2: Leave voters’ satisfaction with democracy will increase following the EU referendum (in Wave 9).
Following the announcement that the majority of voters preferred to leave the EU, there was a sustained discussion of having a second referendum and uncertainty about whether the United Kingdom could, in actuality, entirely leave the EU (see Table 1). Multiple propositions as to how and when to leave were debated and rejected. Issues such as freedom of movement, whether EU citizens would be able to remain in the United Kingdom, whether the United Kingdom would remain in the EU Court of Human Rights, the border between Ireland and Northern Ireland, among others, held up formally leaving the Union and generated concern among Leave voters and hope among Remain voters that politicians would not (be able to) follow through with Brexit.
Therefore, we further expect:
H3: Following the initial drop in satisfaction with democracy among Remain voters, their satisfaction will steadily increase during the period of Brexit implementation uncertainty (Waves 10–17).
H4: Following the initial boost received to satisfaction with democracy among Leave voters, their satisfaction with democracy will gradually decline during the period of Brexit implementation delays and uncertainty (Waves 10–17).
Brexit Timeline and BESIP Waves.
Boris Johnson led the Conservatives to an electoral victory in the 2019 UK General Election. The Conservative campaign was a well-coordinated promise to voters that Johnson would “Get Brexit Done” (Cooper and Cooper, 2020). The campaign instilled in voters the belief that Johnson would cease with the repeated extensions and exit the EU (Gaskell et al., 2020).
Finally, this leads us to expect that:
H5: Following the 2019 General Election, Remain voters’ satisfaction with democracy will return to its immediate post-referendum level (in Wave 19).
H6: Following the 2019 General Election, Leave voters’ satisfaction with democracy will return to its immediate post-referendum level (in Wave 19).
Data and Measurement
Unlike previous studies which focused on short-term changes in satisfaction immediately following the referendum (e.g., Dunn and Thornton, 2022), this study focuses on potential long-term changes in democratic satisfaction.
We utilize data from Wave 7 through Wave 21 of the British Election Study Internet Panel (BESIP). This data covers the interval between April 2016 and May 2021. Wave 7 represents the first data collection point following the announcement of the passage of the EU Referendum Act. Wave 21 was conducted after the passage of the European Union Act 2020 and the UK’s official departure from the EU at the end of 2020. Waves 12, 14, and 18 did not include a question on satisfaction with democracy and are therefore not included in our analyses. We drop respondents who reside in Scotland due to Scotland’s devolved status, the dominance of the Scottish National Party (SNP), widespread opposition to the Conservative Party, and the possible correspondence between attitudes toward Scottish independence and EU membership; all of which may influence their satisfaction with democracy in the wider United Kingdom.
Sample Construction and Missing Data
Our analyses focus on respondents who provided complete data across all key variables and waves, starting from an initial sample of 30,906 respondents in Wave 7. We exclude: Scottish respondents due to their recent experience with independence referenda (2014 Scottish Independence Referendum); respondents who answered “Don’t know” to the satisfaction with democracy question; and respondents who answered “Don’t know” to the EU referendum vote choice question. Our final sample consists of 1228 individuals who participated in all 12 waves containing the satisfaction question (Waves 7–11, 13, 15–17, 19–21), allowing us to directly observe individual changes in democratic satisfaction among both Leave and Remain supporters.
Analysis and Results
We model voters’ satisfaction with democracy using linear panel regression with fixed effects for individuals and wave-specific coefficients for referendum vote choice. Our first analysis examines the interaction between the EU referendum vote and wave, controlling for general election vote choice/intention to isolate Brexit referendum effects from broader partisan winner-loser dynamics. Our second analysis creates a three-way interaction between EU referendum vote, wave, and general election vote to account for potential moderating effects from the two general elections that occurred during data collection. Rather than presenting the rather lengthy regression tables produced by our analyses, we present substantive graphic representations of these outputs: voters’ predicted satisfaction (Figures 1 and 2) and differences in predicted satisfaction (Figures 3 and 4).
Difference in Satisfaction among Remain and Leave Voters across Waves.
Difference in Satisfaction among Remain and Leave Voters across Waves by Party Voted for in the Most Recent General Election.
Difference in Satisfaction with Democracy between Leave and Remain Supporters in Each Wave.
Difference in Satisfaction with Democracy between Leave and Remain Supporters in Each Wave by Party Voted in the Most Recent Elections.
In Figure 1, we display the changes in satisfaction among Leave and Remain voters from Wave 7 to Wave 21. The democratic satisfaction of Remain voters fluctuates between 0.55 and Wave 8 (SE = 0.01) and 0.42—Wave 15 (SE = 0.01); approximately 13% of the range of the scale (Δ = −0.13, SE = 0.01, p < 0.001). Voters show substantive shifts in their satisfaction between Waves 8 and 9 (Δ = −0.07, SE = 0.01, p < 0.001) and between Waves 13 and 15 (Δ = −0.08, SE = 0.01, p < 0.001). This is consistent with our first hypothesis, that satisfaction among Remain voters would drop in the immediate aftermath of the referendum.
There are gradual upticks in satisfaction with democracy among Remain voters between Waves 9 and 11 (Δ = +0.04, SE = 0.01, p < 0.001), and Waves 17 and 21 (Δ = +0.06, SE = 0.01, p < 0.001). This is consistent with our third hypothesis, noting that satisfaction would somewhat recover in the midst of doubt over whether the UK Government would be able to implement an exit from the EU. However, this is inconsistent with our fifth hypothesis that anticipated a drop in satisfaction between Waves 17 and 19. Contrary to this hypothesis, our analyses indicate a non-significant increase in satisfaction with democracy (Δ = + 0.02, SE = 0.01, p = 0.054). In sum, Remain voters end up notably less satisfied with democracy following the UK’s exit from the EU than they were when the referendum was announced (Δ = −0.06, SE = 0.01, p < 0.001).
The democratic satisfaction of Leave voters exhibited multiple significant fluctuations between consecutive waves. Heading toward the referendum date, potentially anticipating that their preference would be rejected by the public at large, there is a small but significant drop in their satisfaction with democracy (Δ = −0.04, SE = 0.01, p < 0.001). Following the referendum and the revelation of Brexit being the dominant preference of the voting public, the satisfaction of Leave supporters increased substantially (Δ = 0.10, SE = 0.01, p < 0.001). This aligns with our second hypothesis.
Following the announcement of the outcome of the referendum and the onset of the discussion of a second referendum, Leave voters’ satisfaction with democracy dropped back to where it was when the referendum was first announced (Δ = −0.02, SE = 0.01, p = 0.093), only slightly higher than it was leading into the referendum (Δ = 0.02, SE = 0.01, p = 0.048). Leave voters’ satisfaction then fell drastically between Waves 13 and 15 (Δ = −0.13, SE = 0.01, p < 0.001) and again between Waves 16 and 17 (Δ = −0.06, SE = 0.01, p < 0.001) for a total decline in satisfaction of approximately 24% of the range of the scale (Δ = −0.24, SE = 0.01, p < 0.001)—Wave 9 the highest and Wave 17 the lowest. This aligns with our fourth hypothesis, that the satisfaction of Leave voters would decline after an initial peak following the announcement of the referendum results.
Following the Conservative Party’s 2019 General Election win on the promise to “Get Brexit Done,” satisfaction among Leave supporters returned to pre-referendum levels and continued upwards through Wave 21 (Δ[Waves 17 to 21] = 0.22, SE = 0.01, p < 0.001). This somewhat aligns with our sixth hypothesis, which predicts a return to immediate post-referendum levels of satisfaction. However, Leave voters do not quite return to the high point of their satisfaction with democracy (Δ = −0.02, SE = 0.01, p = 0.07), though they do end up slightly more satisfied than they were at the announcement of the referendum (Δ = 0.04, SE = 0.01, p = 0.001). The dramatic decline between Waves 13 and 17 (Δ = −0.19, SE = 0.01, p < 0.001) may exemplify how electoral disappointment can affect satisfaction (cf., Tunç et al., 2023); when winning voters become dissatisfied not with the referendum outcome itself, but with the political system’s apparent inability to implement their democratic choice.
Figure 3 plots the average marginal effect of the referendum on satisfaction with democracy across waves. The average marginal effects indicate the difference between Remain and Leave voters in each wave. When the referendum is announced, Leave voters are, on average, less satisfied with democracy than those who vote to remain in the EU (Wave 7: β = −0.06, SE = 0.01, p < 0.001; Wave 8: β = −0.10, SE = 0.01, p < 0.001). This trend briefly reverses following the referendum, with Leave voters reporting greater satisfaction with democracy than Remain voters (Wave 9: β = 0.06, SE = 0.01, p < 0.001), though any difference between the two groups disappears in the uncertainty between Waves 11 and 13 (−0.03 ≤ β ≤ −0.02, SE ≈ 0.01, 0.317 ≤ p ≤ 0.435) before Leave voters once again express less satisfaction with democracy than Remain voters between Waves 15 and 17 (−0.06 ⩾ β ⩾ −0.13, 0.01 ≤ SE ≤ 0.02, p < 0.001). Upon the UK’.s departure from the EU, Brexit voters were slightly more satisfied with democracy than Remain voters (β = 0.03, SE = 0.02, p = 0.03).
To account for any potential moderating effect that may occur due to the two General Elections that take place within the timeframe of the data collection, we rerun the above analyses, adding party-vote choice into our interaction term. Figures 2 and 4 replicate our above findings for both Conservative and Labor party voters (see Online Appendix C for similar plots for Liberal Democrat and UKIP/Brexit/Reform voters).
Figure 2 yields some interesting results that largely, though not entirely, echo our results reported above. For Conservative voters in Waves 7 and 8, predicted satisfaction with democracy among Remain voters was higher than that for Leave voters (Wave 7: Δ = −0.08, SE = 0.02, p = 0.001; Wave 8: Δ = −0.09, SE = 0.03, p < 0.001). Despite the moderate fluctuations across the waves, satisfaction with democracy among Conservative Remain voters returns, in Wave 21, to the satisfaction level we see initially in Wave 7 (Δ < 0.01, SE = 0.02, p = 0.989). More positively, even though their satisfaction dropped to very low levels in Wave 17 (β = 0.35, SE = 0.01, p < 0.001), Conservative Leave voters end up, in Wave 21, notably more satisfied with democracy than they were when the referendum was announced (Δ = 0.08, SE = 0.16, p < 0.001); effectively as satisfied with democracy as the Remain voters (Δ < −0.01, SE = 0.02, p = 0.999).
The pattern among Labor voters largely follows that of Conservative voters. However, unlike Conservative voters, both Leave (Δ = −0.07, SE = 0.03, p = 0.024) and Remain (Δ = −0.07, SE = 0.02, p = 0.001) Labor voters end up less satisfied with democracy at the conclusion of the time series (Wave 21) than they were at the beginning (Wave 7).
Discussion
Previous research suggests that there is a sustained winner-loser gap following an election: losers feel less satisfied with democracy while winners feel more satisfied. In cross-national analyses, this winner-loser gap has been found to sustain for at least 5 years (Loveless, 2021).
Our analyses indicate a remarkable lack of stability in satisfaction with democracy. We suspect this is due to the lack of certainty and the political divisiveness regarding how the Brexit vote would be operationalized, if at all. However, regardless of this instability, the general pattern indicates that both the reduction in satisfaction for the losers and the increase in satisfaction among the winners sustain over the 5 years of our data; though the satisfaction of the winners does decline modestly when comparing the post-referendum and post-Brexit waves.
Satisfaction with democracy among the losers in the Brexit referendum, the Remain voters, does drop as expected. Repeated failures to attempt to reverse the Brexit vote and the consequent political decision to act on the referendum result resulted in repeated drops in satisfaction among both Remain and Leave voters. However, in each case of decline, there was a quick reversal toward increased satisfaction. Overall, the satisfaction of Remain voters does remain notably lower at the end of our analyses, 5 years later, than at the beginning.
Inversely, there is an increase in satisfaction among Leave voters immediately following the referendum. This was followed by repeated fluctuations in satisfaction due to the following uncertainty. When it did become clear that the UK Government would adhere to the preference of the referendum voters, to pull the United Kingdom out of the EU, and finally did so, satisfaction returned to levels closer to those immediately following the referendum, indicating a sustained positive “winner effect” on satisfaction with democracy.
Unlike previous research evidencing a sustained winner-loser gap over a 5-year period that relied on cross-sectional data, this article uses time-series panel data, allowing us to track changes in individual-level data. The effects of this are twofold: first, our analyses indicate that individual-level satisfaction with democracy is quite unstable in the short term. This may be a contextual effect due to the serious and cross-generational consequences of a vote-choice that was cast along polarizing lines and/or the uncertainty that followed the referenda vote. Similar research from differing contexts will help elucidate whether this pattern is normal instability, or an exception due to a volatile period in UK political history. Second, when combined with previous analyses (e.g., Dunn and Thornton, 2022), we can say with a greater level of confidence that satisfaction with democracy does indeed respond to the political environment and to electoral results in particular.
Our findings suggest that satisfaction effects of contentious referenda are highly dependent on implementation context, reflecting several referendum-specific factors: Brexit represented a rare, generational decision with permanent consequences; the extended implementation period created ongoing uncertainty about whether the outcome would be realized; and Brexit functioned as an identity-defining issue that transcended traditional party loyalties.
Timeframe Limitations and Future Research Directions
Our analysis concludes in May 2021, 4 months after the UK’s formal EU departure, capturing what we term the “implementation phase” of the referendum cycle. While this represents the most comprehensive longitudinal analysis of referendum winner-loser effects to date, spanning the complete period from announcement through policy implementation, it necessarily leaves open questions about longer-term durability.
Theoretically, our 2016–2021 timeframe captures the period most likely to exhibit referendum-specific satisfaction effects—when Brexit remained a dominant political issue and implementation uncertainty was at its peak. The post-2021 period represents a different phase where Brexit-specific effects may fade as political attention shifts to other issues. Whether the satisfaction patterns we observe represent permanent attitude changes or contextually dependent responses remains an important empirical question.
We acknowledge that our final waves (Wave 20: June 2020; Wave 21: May 2021) occurred during and after the initial COVID-19 pandemic wave in the United Kingdom, creating potential confounding between Brexit implementation effects and pandemic-related changes in democratic satisfaction. While our Brexit-related patterns remain consistent across the majority of our time series (Waves 7–19), which predates pandemic effects, we cannot definitively rule out that differential pandemic impacts on Leave versus Remain voters may have influenced our final wave results. The government’s handling of COVID-19, lockdown measures, and their economic consequences could theoretically have affected Brexit identity groups differently, given their distinct relationships with government trust and authority. Future research should examine whether referendum-induced winner-loser gaps persist, converge, or reverse as implementation consequences become apparent and political salience declines.
In sum, our analyses provide additional support to previous research in finding a sustained winner-loser gap in satisfaction with democracy among voters. This is driven by an increase in satisfaction among winners and a decrease in satisfaction among losers and shows evidence of an effect lasting 5 years.
Footnotes
Acknowledgements
We would like to thank the editors for their constructive guidance throughout the review process and for facilitating a smooth and efficient path to publication. We are also grateful to the reviewers for their thorough engagement with our manuscript and for their valuable feedback. The comments we received have substantially improved the clarity and contribution of this research note.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Supplemental material
Additional Supplementary Information may be found with the online version of this article.
Content
Table A1: Predicted satisfaction with democracy among Leave and Remain supporters across waves Table A2: Difference in predicted satisfaction with democracy between Leave and Remain supporters in adjacent waves Table A3: Marginal effect of vote choice on satisfaction with democracy across waves Table A4: Difference in the marginal effect of vote choice on satisfaction with democracy in adjacent waves Table A5: Predicted satisfaction with democracy among Leave and Remain supporters across waves across parties Table A6: Difference in predicted satisfaction with democracy between Leave and Remain supporters in adjacent waves across parties Table A7: Marginal effect of vote choice on satisfaction with democracy across waves Table A8: Difference in the marginal effect of vote choice on satisfaction with democracy in adjacent waves across parties Figure B1: Aggregate satisfaction with democracy among voters in the Brexit referendum Figure C1