Little room for motivated reasoning? How partisans respond to government misconduct with well-substantiated evidence

Abstract

How do voters respond to a political scandal? Previous studies emphasize partisan motivated reasoning as a key factor in shaping responses to political scandals but offer mixed evidence. I argue that partisans are less likely to engage in motivated reasoning when two conditions are met: (1) clear evidence of severe misconduct and (2) clear responsibility attribution to an involved party. By analyzing an unexpected video release of the Partygate scandal in Britain during which a nationally representative survey was fielded, I find that even copartisans and independents show a decline in evaluation of government performance in handling COVID-19, feel less favorable toward the involved governing party, and report reduced intentions to vote for the party. In contrast, I find little evidence that supporters of the opposing party show a decline in these evaluations, likely due to prior negative expectations and floor effects. The findings highlight the limitations of partisan motivated reasoning among copartisans and voters’ intention to punish the involved party in the face of unambiguous evidence of government misconduct.

Introduction

Political scandals often negatively affect evaluations of candidates and parties and voting behavior (Funk, 1996; Welch and Hibbing, 1997) and undermine public trust and perceptions of legitimacy (Bowler and Karp, 2004; Van et al., 2020). At the same time, a large body of research discusses the role of partisanship in shaping citizens’ reactions and attitudes after scandals. On the one hand, researchers argue that support for involved politicians and parties among copartisans is less affected than independents and opposing party supporters due to partisan perceptual screens (Anduiza et al., 2013; Solaz et al., 2019). On the other hand, some studies provide limited support for partisan motivated reasoning and show that scandals reduce political trust and party support regardless of partisanship (Aarslew, 2023; Costa et al., 2020).

How do we reconcile these mixed findings? Building upon a theory of partisan motivated reasoning, I argue that a scandal decreases support for the actors involved when citizens find unambiguous evidence of the authenticity and severity of a scandal and perceive that they are responsible for the incident.

This study analyzes the Partygate scandal in Britain that received widespread attention after a video was released that captured a chat between government officials talking about a Christmas party and the violations of COVID-19 restrictions during a lockdown. Using an unexpected event during survey design (UESD), I find a decline in governing party support and evaluations of government performance in managing COVID-19 and lockdowns among both governing party supporters and independents. Additionally, I show that partisans are willing to sanction the Conservative Party government by reducing their vote intention. In contrast, rival party supporters do not show a decrease in government evaluations, consistently showing negative evaluations of the government before and after the scandal.

This study makes multiple contributions. First, while previous studies have provided mixed evidence about the role of partisanship in shaping responses to political scandals, few studies discuss the conditions of partisan motivated reasoning (but see Funck and McCabe, 2022). This study clarifies these conditions. Second, while much of the misconduct literature focuses on voting behavior and vote intention as an outcome (Funck and McCabe, 2022; Welch and Hibbing, 1997), I analyze the effects of political scandals on a wider range of variables, including government performance evaluations and feeling thermometer ratings. Finally, unlike many studies that rely on survey experiments with hypothetical vignettes or conjoint experiments (Klašnja et al., 2021; Anduiza et al., 2013), this study offers real-world evidence that even copartisans are willing to punish political elites involved in a scandal.

Partisan motivated reasoning and responses to scandal

While political scandals hurt candidate and party evaluations (Aarslew, 2023; Funk, 1996), the effect can vary. Some research shows that prior attitudes moderate the effect of political scandals on citizens’ responses, including gender attitudes (Longdon and Banducci, 2023), ideology (Saxton and Barnes, 2022), and political cynicism (Dancey, 2012). One central variable that causes such heterogeneous effects is partisanship (Anduiza et al., 2013; Solaz et al., 2019). According to Kunda (1990), people process information either for accuracy or to achieve motivational goals. For accuracy goals, individuals are motivated to reach an accurate conclusion by broadly collecting and processing information in a less biased manner. In contrast, individuals driven by motivational goals collect and process information to draw a conclusion consistent with their prior beliefs (Kunda, 1990). Since partisanship motivates individuals to positively view their partisan in-group (Green et al., 2004), partisans engage in motivated reasoning and process political information in a biased manner.

However, the extent of partisan motivated reasoning depends on the types of scandal information and information environments (Funck and McCabe, 2022). I argue that partisan motivated reasoning can occur in two distinct forms, biased perceptions of facts and selective blame attribution, and that partisan motivated reasoning is more likely when there is room for different interpretations of a scandal from either of these two perspectives. First, partisans judge whether new information is trustworthy and important. If it is ambiguous or complex, their interpretation of the facts may be influenced by bias, leaving room for subjective perceptions and motivated reasoning. For example, Parker-Stephen (2013) finds that partisan differences in economic evaluations peak when the macroeconomy is ordinary, and as economic conditions get better or worse, different party identifiers agree on the perceptions of economic conditions. Savani and Collignon (2024) find that one cognitive strategy supporters use to avoid blaming politicians involved in a scandal is to disbelieve and reject the allegation. Additionally, partisans may dismiss information that threatens the reputation of their preferred party (Kane and Anson, 2023). In summary, if partisans face information that leaves room for motivated reasoning regarding its authenticity and severity, partisan differences in perceptions increase.

The second is selective blame attribution. Bisgaard (2015, 2019) finds that, even when partisans accurately perceive economic conditions, they still engage in motivated reasoning by selectively attributing blame or credit to the government. These findings suggest that even accurate perceptions of negative facts about one’s partisan in-group may not lead to holding politicians accountable. When there is room to selectively attribute blame or credit to the government, partisans interpret responsibility in a biased manner.

Clarity of responsibility is crucial to attributions of responsibility. As Powell Jr and Whitten (1993, 398) note, clarity of responsibility does not merely result from “individual-level idiosyncrasy” but also from “coherence and control.” The clearer the government’s control over a situation, the more citizens attribute responsibility to government officials. In summary, I argue that partisan motivated reasoning is less likely to occur in the context of government scandals when two conditions are met: (1) clear evidence of the authenticity and severity of a scandal and (2) a situation in which citizens perceive that the government is responsible for the incident.

Case: Partygate scandal in Britain

To test these hypotheses, I analyze the Partygate scandal in Britain. The British government faced a challenge during COVID-19 in controlling the number of infections and deaths. To curb the spreading outbreak of COVID-19, the government implemented multiple lockdowns.

On November 30, 2021, the Daily Mirror released the first news report about a Christmas party held at Downing Street, the prime minister’s official residence, that had occurred in December 2020 during a lockdown (Source A1). ¹ Following the report, other major news outlets, such as the BBC (Source A2), covered the misconduct as well. However, at that point, the news did not receive as much attention as it did afterward (see Appendix B). On December 1, 2021, Johnson claimed that “all guidance was followed completely in No 10” (Source A3).

It was December 7, 2021, when ITV, one of the most traditional media outlets, released a video that showed government officials joking about the Christmas party. They described it as “a cheese and wine” and acknowledged that it was “not socially distanced” (Source A4). The footage clearly shows that the COVID-19 rules were broken and that the government lied about the scandal. The evidence placed responsibility on Johnson as the person accountable for the gathering at his office and for his lie on December 1. The video sparked increased media attention and fueled criticism of the governing party and the prime minister (Appendix B).

Additionally, a day after the video was released, the prime minister, Boris Johnson, apologized “for the impression that it gives” (Source A5) while he repeatedly denied that the party took place. However, a poll taken just after the video footage by Sky News shows that 64% of respondents believed that the Christmas party occurred, 65% believed that the COVID rules were not followed, and 63% said that the prime minister was lying. ² Table 1 summarizes the timeline of the relevant events.

Table 1.

Timeline of the “Partygate” scandal.

Date	Event/Information released
26/11/2021	BESIP Wave 22 Started
30/11/2021	Initial report on Downing Street gatherings during lockdown from the Daily Mirror
01/12/2021	Johnson claimed, “all guidance was followed”
07/12/2021	Leaked video of staff joking about a lockdown party from ITV
07-08/12/2021	Poll from Sky News
08/12/2021	Johnson’s apology in Parliament
15/12/2021	BESIP Wave 22 Ended

The Partygate scandal presents an opportunity to test the expectations outlined above. The initial report by the Daily Mirror can satisfy the condition of responsibility. The article notes that Johnson was “accused of breaking Covid rules by attending parties at Number 10” (Source 1). However, the report offers only ambiguous evidence because all sources are anonymous. Additionally, the government denied the wrongdoing. Such ambiguity leads to little public response to the affair at first due to weakly perceived authenticity. At this stage, Johnson could be considered responsible only insofar as the allegations were true, while the authenticity of the claims remained uncertain.

In contrast, the video release clearly satisfies the two theoretical conditions above. First, the video footage provided clear, unambiguous evidence that there was a party at Downing Street where attendees broke the rules they imposed on the public. The evidence, reported by a reliable media outlet, was so clear that citizens understood what happened regardless of partisanship once the video was released. ³

Second, as the party was held at the official residence and office of the prime minister, it was difficult for Prime Minister Johnson to shift responsibility to other actors and evade blame. In the video, a spokesperson for the prime minister acknowledged that a Christmas gathering had taken place and that COVID-19 rules were broken. The leaked footage provided the first direct visual confirmation that staff had held a party at the prime minister’s office, collapsing the government’s earlier denials and making Johnson’s responsibility clearer. Additionally, Johnson’s apology and the media criticism following the leaked video clarified that he was responsible for the scandal. In summary, the Partygate scandal satisfies the two conditions above and provides an opportunity to test the limits of partisan motivated reasoning.

Methods

To test the effect of well-substantiated government misconduct on public opinion, I use an unexpected event during survey design (UESD) (Muñoz et al., 2020). The UESD enables an estimation of the effect of unexpected events by exploiting quasi-random variation in the timing of survey responses. The exogeneity of the shock arises from the fact that the video release was neither anticipated nor manipulable by political actors. The timing was determined by the media rather than by the government or the public, which makes the exact moment of treatment effectively unrelated to public opinion. Since the event’s timing is exogenous to survey participation, differences before and after the event can be interpreted as treatment effects, based on the assumptions of no anticipation and no other simultaneous shocks. Wave 22 of the British Election Study Internet Panel (BESIP) was administered between November 26 and December 15, which encompasses the initial report on November 30 and the video release on December 7 (Fieldhouse et al., 2026). Thus, UESD allows us to estimate the effect of the scandal by comparing survey responses before and after the event. Scholars have employed UESD to estimate the effect of unexpected scandals (Hegewald and Schraff, 2024).

I focus on three outcomes of interest and six variables: evaluations of government performance in handling COVID-19 and lockdown, favorability toward the prime minister and the governing party, and vote intention for the Conservative and Labour Parties. In addition to attitudes toward the government, I analyze respondents’ vote intention to explore whether partisans are willing to sanction the in-party for a government scandal. Changes in government evaluations may not imply citizens’ willingness to hold the government accountable. Even if voters negatively evaluate politicians involved in a scandal, they may continue to support them when they do not have reasonable alternatives (Agerberg, 2020) or when their evaluations of the governing party are still more positive than the opposing party (Aarslew, 2023). Analyzing not only government evaluations but also vote intention, the present study evaluates whether partisans intend to sanction the copartisan government. Table 2 summarizes outcome variables used in the analyses. I estimate the following equation:

$Y_i=\alpha +\beta \hspace{0.17em}Pos{t}_i+Controls{’}_{}\hspace{0.17em}\gamma +{ϵ}_i$

where Post _i indicates whether a respondent took an interview after the video was leaked. As control variables, I include gender, age, income, education, unemployment, ethnicity, disability, and the 7-day rolling average number of daily new confirmed COVID-19 cases per million in Britain, obtained from Our World in Data (Mathieu et al., 2020). The seven demographic variables account for potential covariate imbalances that could be correlated with response timing. ⁴ I additionally control for the COVID-19 case trend to account for a concurrent increase in infections that may confound the event of interest. Because UESD assumes no simultaneous events affecting the outcome variables, incorporating the COVID-19 trend into the model helps address this concern. ⁵

Table 2.

Outcome variables.

Outcome of interest	Variable	Scale
Government performance on COVID-19	Government performance: COVID-19	1–5
	Government performance: Lockdown	1–5
Favorability	Feeling thermometer: Conservative Party	0–10
	Feeling thermometer: Boris Johnson	0–10
Vote intention	Vote intention: Conservative Party	0–1
(Willingness to sanction)	Vote intention: Labour Party	0–1

Note. The table describes outcome variables used in this paper. See Appendix C for survey instruments of each variable.

I use a 7-day window as a bandwidth before and after the event to estimate the effect of a government scandal. As the first coverage of the scandal was published 7 days before the video was leaked, including samples before the first report may conflate the effect of the first scandal report and the video footage. Since this study argues that well-substantiated evidence of a government scandal leaves little room for motivated reasoning and causes copartisans to hold more negative attitudes towards the in-party government after the incident, I compare respondents who may be exposed to the scandal report per se but not to the video to those who may encounter news about the video. ⁶ Comparing public opinion before and after an unexpected event would not provide an internally valid estimate if respondents in the control and treated groups are exposed to different information besides the video. To ensure the ceteris paribus assumption as much as possible, I use a narrow bandwidth.

Results

Figure 1 shows the daily average of Conservative Party favorability before and after the video among the whole sample and partisan subgroups. The figure indicates that public opinion appears stable immediately after the initial coverage of the scandal. Consistent with this pattern, the results in Appendix M suggest that the initial report had little impact on the outcome variables. This limited response aligns with my argument that voters are less likely to react to a scandal when the available evidence is ambiguous. In contrast, there is a noticeable decrease in the party evaluation following the video release among the full sample, Conservative Party supporters, and independents while Labour supporters do not show a decline in their assessment of the governing party. The overall pattern is consistent with my theoretical expectation that political scandals negatively affect evaluations of involved politicians when clear evidence of the authenticity and severity, and clear responsibility attribution are combined.

Figure 1.

Average score of the Conservative Party favorability across partisan subgroups: (a) Full sample; (b) Conservative supporters; (c) Labour supporters; (d) Independents

Next, I estimate changes in government performance evaluations, party and leader favorability, and vote intention in response to the video release. Figure 2 presents regression coefficients from OLS models that compare respondents surveyed during the 7 days before and the 7 days after the event, controlling for demographics and daily COVID statistics. Figure 2 indicates that the video footage leads to a decline in evaluations of the governing party, as well as evaluations of the government’s handling of COVID-19 and the lockdown, among Conservative supporters and independents.

Figure 2.

Effect of Partygate scandal video footage: (a) Effect on COVID-19 performance evaluations; (b) Effect on party and leader favorability; (c) Effect on vote intention

In contrast, Labour supporters do not show a significant decrease in evaluations of the governing party. ⁷ I offer two potential explanations of the counter-intuitive response among out-party supporters. One is expectation confirmation. Copartisans and independents hold positive expectations of the government and thus show a decrease in their evaluations after a scandal. In contrast, as opposing party supporters do not hold positive expectations of the government, the video just confirms their negative impression and does not lead to a decline in the government evaluations. Another is the floor effect, which occurs when survey respondents show responses at the lower limit of a measurement scale. If opposing party supporters show negative evaluations of the government even before a government scandal, a survey taken after the scandal may not fully capture a further decrease in government evaluations. Consistent with these explanations, Labour supporters hold strongly negative attitudes toward the Conservative Party even before the video release. As Table H1 indicates, in the 7 days before the video’s release, the average feeling thermometer score for the Conservative Party among Labour supporters was 1.45 on a 0–10 scale. Also, Figures H1 and H2 show that more than 50% of Labour supporters rated the party and the leader as 0, or the minimum, before the event.

Finally, I evaluate whether the scandal affects vote intention. Figure 2(c) suggests that voters are willing to punish the governing party for the scandal. Specifically, the percentage of Conservative supporters who would vote for the Conservative Party if there were a general election tomorrow drops by 8.1 percentage points among Conservatives and 5.8 percentage points among independents (Table G5). In addition, Table G6 indicates a significant increase in vote intention for the Labour Party across all partisan groups, by 2.3 percentage points among Conservatives, 5.0 percentage points among Labour supporters, and 5.5 percentage points among independents. This finding is consistent with the argument in Agerberg (2020) that voters are inclined to punish an incumbent party and its politicians for misconduct when they have a reasonable alternative. ⁸ Entirely, these effects are substantively meaningful. In my sample, 29.8% are Conservative supporters, 23.8% are Labour supporters, and 24.3% are independents. If these shifts are directly translated into voting behavior in an election, the governing party would lose roughly 3–4 percentage points of the national vote while the Labour Party would gain around 3 percentage points, based on a simple weighted approximation. In sum, the overall pattern supports my theoretical underpinnings. As the initial report only provides ambiguous evidence, the British public does not strongly react to it at first. However, the video footage, which provided clear evidence of the misconduct and responsibility of the government, caused a decline in governing party evaluations, and even copartisans are willing to sanction the party.

Conclusion

This study investigates how partisans react to government misconduct when presented with well-substantiated evidence. Many studies suggest that partisans engage in motivated reasoning even when they face a copartisan government’s poor performance as long as there is room for different interpretations of the facts (Bisgaard, 2015, 2019). However, if there is little room for the interpretations of the misconduct and the government is clearly responsible, copartisans punish the in-party government.

Analyzing the initial video footage of the Partygate scandal in Britain, I find evidence consistent with the expectations. This study contributes to research on political scandals by providing real-world evidence that motivated reasoning is weakened when evidence is clear and difficult to reinterpret. The results hold across various outcomes, including evaluations of government performance, favorability toward a governing party, and vote intention. The study also broadens the literature of government misconduct by adding evidence from a non-US context. The findings overall suggest that partisan biases, while powerful, do not fully prevent citizens from holding their copartisan government accountable when the misconduct is unambiguous.

This study also has limitations. First, since the survey was fielded around the event and all later panel respondents would have been exposed to the treatment, the duration of the effect cannot be assessed. Second, the generalizability of the findings may be constrained by the unique contextual features. Before the scandal, British citizens were tired of COVID-19 restrictions. These contexts might have further caused anger and strengthened negative responses to the Partygate scandal. However, such an extreme context is itself part of the severity dimension that this study aims to highlight. Future research could analyze other cases that satisfy the same level of severity, authenticity, and clear responsibility and test if the findings of this study are replicable.

Supplemental material

Supplemental material - Little room for motivated reasoning? How partisans respond to government misconduct with well-substantiated evidence

Supplemental material for Little room for motivated reasoning? How partisans respond to government misconduct with well-substantiated evidence by Wataru Onishi in Research & Politics