Determinants of Political Trust during the Early Months of the COVID-19 Pandemic: Putting Policy Performance into Evidence

Basing on the previous and early months of the COVID-19 pandemic, this article analysis the main determinants of citizens’ trust in the prime minister over that period. Prior research on the political effects of the pandemic has mostly focused on identifying a rally around the flag effect, and little is known about other reasons behind the increase in trust after the outbreak of the pandemic. Based on survey data from February to July 2020 for Spanish citizens, we argue that the reasons for the increased trust in the prime minister following the pandemic outbreak are due more to performance evaluations than to emotional-related factors regarding COVID-19 health issues. We also argue that among performance evaluations, the assessment of policy performance in fighting the COVID-19 crisis is of preeminent relevance in explaining trust in the prime minister during the pandemic period. Findings widely support our argument. By comparing the effects of conventional to emotional-related factors, we extend scholarship on political trust in the context of an exogenous crisis.


Introduction
The first signs of the coronavirus pandemic were detected in December 2019 in Wuhan, China, and were rapidly felt all over the world. Since then, its effects have been devastating from the point of view of human losses as well as its social and economic impact. Research on its effects have been burgeoning, especially studies on citizens' political trust. In a pandemic context, researching political trust is of particular interest as trust is a necessary component of political leadership. Without it, leaders will have difficulty in succeeding (Hetherington, 1998), especially when it comes to looming challenges such as exogeneous shocks (Keele, 2013). This is surely the case in the ongoing COVID-19 pandemic. As broadly demonstrated regarding compliance with government demands and regulations (Levi and Stoker, 2000: 491-493), in the current pandemic, high levels of trust have been shown to be a necessary condition for the implementation of restrictive policies and for public support and compliance with them (Bavel et al., 2020: 466;Devine et al., 2020). As public cooperation is a key element for the government's successful management of the crisis, trust in the political leadership is instrumental to reach this end.
Literature on the effects of external shocks, whether natural disasters, terrorist attacks or a pandemic, has demonstrated that citizens' political trust tends to increase substantially immediately after the emergence of the shock compared to the previous period. The most common reason given for this trend is the rally around-the-flag effect (Dinesen and Jaeger, 2013;Hetherington and Nelson, 2003). In short, this effect consists of a substantial rise in citizens' trust in the political leading figures as a consequence of an external threat, due to emotional reasons (Mueller, 1970; for a literature review, see Baekgaard et al., 2020: 5). For instance, shocks such as the 11 September 2001 terrorist attack in the United States and the 11 March 2004 Madrid terrorist attack led to increased levels of political trust (Dinesen and Jaeger, 2013;Hetherington and Nelson, 2003).
Based on evidence of increased levels of trust in political institutions, research has been highlighting that the COVID-19 pandemic actually led to national rally around the flag effects in many countries (Baekgaard et al., 2020;Esaiasson et al., 2020;Schraff, 2020;Sibley et al., 2020). This research also indicates that citizens rally around governing institutions and leaders for non-cognitive reasons, independently of the policy choices for dealing with the pandemic (Esaiasson et al., 2020;Schraff, 2020). However, this previous research mainly analysed the effect of lockdown on citizens' trust (Baekgaard et al., 2020;Bol et al., 2021;Schraff, 2020;Sibley et al., 2020), disregarding a broader assessment of the phenomenon's causality. That is, whether the substantial increase in political trust at the initial phase of the pandemic was essentially due to emotional-related factors, such as concern or fear of the coronavirus spreading, thus converging with a rally around the flag effect, or whether conventional cognitive factors such as assessment of government performance are still dominant. So far, little is known about the relative contribution of these factors to explain the increased levels of trust during the early pandemic period. Supported by the Spanish case, we intend to contribute to fill this gap by assessing the determinants of political trust in the prime minister in the early months of the pandemic, comparing cognitive with emotional-related factors.
Prior research on the consequences of exogeneous shocks on political trust tended to analyse institutions rather than leaders. Notwithstanding the misunderstandings and misuses underlying the study of the rally around the flag effect (Chowanietz, 2011), its initial conceptualization suggests that trust increases in key political leaders, seen as national symbols, rather than in political institutions (Mueller, 1970). In particular, leadership in the executive branch, as is the case of the prime minister, is especially suitable to study this effect (i.e. to assess emotional-related factors), due to his or her core role in the political system (Helms, 2005: 3-4). Prime ministers are also directly accountable for the executive's performance. They are the most exposed leaders to public scrutiny, usually being the first targets in citizens' evaluation of government performance (being suitable to assess cognitive factors). For these reasons, we focus on the determinants of trust in the prime minister, and not in political institutions, as most previous research.
This study is structured as follows. First, we examine and systematize extant literature on trust and support the hypothesis. Then, we briefly present the Spanish case. This is followed by information on the data and methods used. Finally, we present and discuss the findings.

Literature and Hypothesis
Performance evaluations have been acknowledged as important predictors of political trust since Citrin's (1974) work in the 1970s. Later on, when summarizing dominant explanations in the literature, Citrin and Stoker (2018: 57) state that 'trust declines when governments and institutions fail to meet expected goals or follow prescribed norms'. Recent literature on the impact of the COVID-19 pandemic on trust rejects the prevalence of this explanation during the outburst of the crisis. Stronger emphasis has been placed on the rally around the flag effect. In their search for security, individuals tend to gather around in-group symbols, such as political leaders, essentially due to emotional reasons. The rally effect is thus promoted by feelings of fear and anxiety in the face of an external threat. Empirically, this effect has been identified as the primary cause of the increase in political trust following the emergence of the pandemic (Baekgaard et al., 2020;Esaiasson et al., 2020;Schraff, 2020;Sibley et al., 2020). However, the relative importance of performance evaluation versus emotional-related factors has not yet been duly explored.
Policy performance is at the heart of institutional trust, with most of the attention turned towards the economy. Despite some controversy, economic performance is commonly considered among the most important determinants of political trust (Foster and Frieden, 2017;Hetherington, 1998;Van der Meer, 2018;Van Erkel and Van der Meer, 2016). This especially holds when it comes to perceptions of economic performance (Anderson and Lotempio, 2002;Chanley et al., 2000;Hetherington and Rudolph, 2008). Good economic performance or citizens' evaluation of government performance has a positive impact on trust, while poor performance or evaluation has a negative impact. In this vein, previous scholarship on negative economic shocks (in particular, on the recent Eurozone crisis) has given rise to ample evidence of their detrimental effect on trust in political institutions (e.g. Foster and Frieden, 2017;Torcal, 2014).
Although economic evaluations have dominated the literature on performance effects on trust, some attention has also been devoted to non-economic policy areas albeit to a much lesser extent (Chanley et al., 2000;Hetherington and Rudolph, 2008;Levi and Stoker, 2000;Listhaug and Jakobsen, 2018). Scholars have acknowledged that core attributes of trustworthy political leaders and governments are their ability to design and implement policies, thus demonstrating competence (Levi and Stoker, 2000: 484), promoting wellbeing (Catterberg and Moreno, 2005), as well as of being responsive to citizens' demands (Denters et al., 2007;Torcal, 2014). If people perceive that the government is promoting and/or implementing policies they are satisfied with, then they are likely to trust the government more. Evaluations of government effectiveness have tended to be less relevant in explaining trust than economic policy (Anderson and Lotempio, 2002), except in times of (economic) crisis, when policy responsiveness seems to gain importance (Torcal, 2014).
Although supported by limited evidence, as a few studies have gone beyond the effects of a lockdown, research suggests that in the early pandemic context performance evaluation determinants of trust become insignificant when compared to non-cognitive factors. Citizens seem to rally around for non-cognitive reasons; that is, independently of the strategic choices for dealing with the coronavirus or of typical economic performance predictors (Esaiasson et al., 2020;Schraff, 2020). This means that citizens support governing institutions for emotional reasons, related to anxiety and fear (Schraff, 2020). These results undermine the consolidated knowledge about the relationship between political performance and trust in the government. As put by Devine et al. (2020: 7), in view of these results, it is necessary to distinguish whether this occurs 'because of the affective nature of trust, or simply because the economy is now fundamentally less important'.
We argue that, despite the importance of psychological reasons in increasing trust levels, conventional determinants of trust, such as the assessment of government performance, are not likely to be suspended after the emergence of the pandemic. We support our argument with two main points. First, we claim that policy responses to the crisis are expected to become fundamental from the very first moment. Even limiting the explanatory factors to the lockdown, prior research demonstrated that retrospective evaluation of the lockdown decision was an important factor in explaining increased trust in government at the outset of the pandemic (Bol et al., 2021, although this is not consensual; see Schraff, 2020). This is consistent with what was found concerning other major external shocks. For instance, literature on natural disasters has shown that individuals tend to reward (or blame) government for their efficiency (or incompetence) in the aftermath of disaster events, holding incumbents accountable (e.g. Carlin et al., 2014;Kuipers and Hart, 2014). Therefore, there are good reasons to expect that cognitive responses to the executive's performance are not deferred when external shocks occur, but rather that they promote accountability. It is precisely because there is a threat that requires an effective political response that individuals value government performance. Accordingly, we expect that citizens' assessments of the executive's policy decisions during the outbreak of the pandemic crisis are important in explaining trust.
Second, in addition to having reasons to suppose that accountability remains in place in the context of the COVID-19 pandemic, its importance has not yet been properly assessed. Studies have seldom looked at citizen evaluations of government performance, rather assessing the effect of particular policy measures on citizens' trust, namely a lockdown, downplaying other causal factors (Baekgaard et al., 2020;Bol et al., 2021;Schraff, 2020;Sibley et al., 2020). As a consequence, we know a little about whether the conventional determinants of trust remain with the pandemic. It is likely that, in a major crisis such as this one, as some policy issues become more salient in the public debate, such as healthcare or even the economy, considerations on such issues affect political trust in the executive (Hetherington and Rudolph, 2008). The policy response to the crisis is certainly about the public health emergency, but also about the impact of the crisis on the economy (Podvrsic et al., 2020). Hence, the assessment of performance with regard to the public health crisis and with regard to the economy is likely to affect trust in the executive. We argue that the more positive evaluations of the political performance on these more salient issues, the higher the trust in the prime minister.
Although acknowledging the importance of emotional forces in driving an increase in trust, we claim that citizens do not refrain from holding the government accountable for its policy response during a crisis such as the pandemic (Kuipers and Hart, 2014) and that trust likely reflects this accountability. Based on the idea that, above all, individuals trust governments that act according to their interests and expectations (Citrin and Stoker, 2018), our expectation is that the evaluations of the executive's performance in fighting the pandemic and of the economic performance were main driving forces fuelling the high levels of trust in Spain's prime minister after the pandemic emerged. We thus expect that positive performance evaluations boost trust in the prime minister to a greater extent than emotional-related factors, such as concern or fear regarding the coronavirus. The hypothesis to be tested is therefore as follows: Hypothesis: Citizens' evaluations of the policy response to the pandemic and of the economic performance are expected to have a greater effect on trust in the prime minister during the outbreak of the pandemic than emotional-related factors such as citizens' feelings of concern or fear regarding the COVID-19 virus.

The Spanish Case
Spain detected its first case of COVID-19 disease on 31 January 2020 and local transmission of the virus was identified on 26 February. Since then, the coronavirus spread rapidly throughout the country making it one of the European countries most affected by COVID-19. By the end of March, the country recorded the highest number of new daily cases, corresponding to a peak of the pandemic. In the early months of the pandemic, the high death toll in Spain stood out among European countries (Royo, 2020: 181). As demonstrated in Figure 1, April was the most critical month as new cases remained high and the number of deaths peaked, surpassing a total figure of 16,000. In order to contain the spread of new cases, in mid-March, the executive implemented a set of measures restricting people's movement as well as economic activities, and prioritizing telework (Henríquez et al., 2020). From that day until mid-May, the numbers progressively decreased until stabilizing at daily highs that fluctuated between Source: Data retrieved from https://ourworldindata.org/coronavirus/country/spain. 300 and 900 cases in the months of May, June and July. 1 Only afterwards, the number of cases (and deaths) started raising again. As in other countries, the Spanish economy was harshly affected by the crisis. Gross Domestic Product (GDP) fell by 17.8% in the second quarter of 2020 compared to the first quarter, and unemployment rose to 16.3% in July 2020. 2 The way the government handled the crisis was widely criticized by citizens. From a sample of eight European countries, Spain emerges as the case where more individuals say the country handled the pandemic badly: 47%, while the European average was 27%. 3 This assessment is in line with scholarship diagnosis of a poor government response to the crisis; it was slow, unprepared and uncoordinated, consequently being unable to contain the propagation of the virus (Royo, 2020). It should be noted that this crisis hit Spain at a time of internal political instability, in which two elections took place within a year, resulting in a fragmented parliament and the formation of a minority coalition government (between the Socialist Party, the PSOE, and the far-left Unidas Podemos). The fragility of the executive explains the government's lack of strength and cohesion to take decisive action in the early stages of the pandemic (Royo, 2020: 185).
By studying Spain, we are supported by a paradigmatic case regarding the conditions in which the crisis unfolded, that enables elucidating the role of cognitive versus noncognitive factors in explaining political trust. The severity of the pandemic enhances the prevalence of an emotional response of individuals regarding political trust, as advocated by the rally effect; while critical assessments of government performance suggest the presence of a cognitive response. Unravelling which factors best explain trust in a strong prime minister such as the Spanish (O'Malley, 2007), in a pandemic context, is the central objective of this research.

Data and Methods
We use data from six monthly independent cross-sectional surveys of Spanish citizens to test our hypothesis. These are representative surveys and were conducted by the CIS, Centro de Investigaciones Sociológicas, 4 right before and after the emergence of the COVID-19 pandemic, covering the period between February and July 2020. Besides including February as the pre-pandemic month, our research encompasses the first month of the pandemic in Spain (March, field work between 1 and 13 March), the critical moment of the first wave (April, fieldwork between 30 March and 7 April), and then the months when infections begin to decline and become almost residual (May and June). This time interval is suitable to our research as it encompasses the pre-pandemic phase and different levels of severity of the pandemic phase, covering the first critical period of the crisis in Spain (as seen in Figure 1). This option allows us to measure differences in the determinants of trust at different moments of the pandemic, and their comparison with the pre-pandemic period. By also including July, it enables encompassing citizens' assessment of the first period of severe restrictions on movement and business activity, given that the first state of emergency ended in Spain on 21 June.
Unlike much research on the political effects of the COVID-19 pandemic in its early stage, ours is not panel data. Panel data would allow us to track citizens' pre-and postpandemic levels of trust, but such data are not available for the Spanish case in the period under analysis. However, as our purpose is to assess how much the pandemic affected the determinants of citizens' trust over time, using cross-sectional data such as ours is an adequate option.

Dependent Variable
Our dependent variable is trust in the Spanish prime minister, which was originally ordinal. In order to facilitate the interpretation of results we recoded it as a dummy (1 = 'a lot of trust' plus 'some trust'; 0 = 'little trust' plus 'no trust at all'). 5

Independent Variables
Concern and Fear Regarding the Coronavirus. According to the socio-psychological approach of the rally around the flag effect (Hetherington and Nelson, 2003: 37-38), trust is supposed to increase among those most concerned with and fearful of the health crisis. We measure this emotional-related dimension based on three variables. The first is the extent to which the respondent is concerned about COVID-19 6 (the variable was recoded so that higher values mean higher concern: 1 = 'not at all'; 5 = 'a lot'). The second and third variables refer to the respondents' fear of the COVID-19 health problem based on the identification of the main current problem in Spain and personally affecting the respondent. 7 These were originally semi-open questions, that were recoded into two dummies in which 1 corresponds to answers reporting COVID-19 issues as the most important problem, respectively at a national and personal level, and the remaining issues were coded as 0. In order to assess how much the pandemic context affects the respondents' level of trust in the prime minister, we also included dummy variables for the time interval.
Policy Performance Evaluations. Our measurement of government performance evaluation is based on citizens' assessments of the executive's economic performance and in fighting the pandemic. As people are likely to trust what they perceive to be working well, positive evaluations of economic performance tend to increase political trust (e.g. Hetherington and Rudolph, 2008;Van Erkel and Van der Meer, 2016). Although research has mainly been based on sociotropic perceptions, the same argument is valid for egotropic perceptions. Individual experiences in the labour market, such as being unemployed, less skilled (Foster and Frieden, 2017) or poor (Newton et al., 2018: 47-48) explain institutional distrust. While being a weaker indicator than sociotropic perceptions (Torcal, 2014;Van der Meer, 2018: 603), egotropic economic perceptions are likely to be relevant in the pandemic context, as many saw their personal living conditions strongly affected as a consequence of the lockdown. We investigate both measures of economic performance: on how the respondent evaluates the general economic situation in Spain (sociotropic) 8 and on how he or she evaluates his or her own economic situation (egotropic). 9 In both cases, the variables were recoded so that higher values mean a more positive evaluation (1 = 'very bad'; 5 = 'very good').
To assess the executive's policy performance in fighting the pandemic, we consider a variable measuring the extent to which the respondent considers that the policy measures taken were necessary or not. 10 The variable was recoded so that higher values mean more positive performance evaluations (1 = 'not necessary at all'; 5 = 'very necessary').
Performance evaluations and trust might be reciprocally related. That is, decreasing trust leads to substantially more negative political evaluations, as well as the reverse (Hetherington, 1998). In other words, the evaluation of government performance has an impact on trust and trust in government might as well influence the evaluation of its performance. Although acknowledging the existence of reciprocal effects, this piece of research is exclusively devoted to one of the directions of that relationship; to assess how much citizens' performance assessments impact on political trust.
Control Variables. We control the analysis with a set of variables. First, we control for partisanship. As advocated in the seminal work by Campbell and colleagues, 'identification with a party raises a perceptual screen through which the individual tends to see what is favourable to his partisan orientation' (Campbell et al., 1960: 133). Since Campbell, this argument of partisan selective evaluation has been widely reaffirmed (Belchior, 2019;Tilley et al., 2008). In this vein, the preferred party being in cabinet is among the most important predictors of political trust (Denters et al., 2007;Gershtenson et al., 2006). Inversely, identifying with opposition parties negatively biases perceptions about the government (e.g. Belchior, 2019). Thus, government supporters are supposed to evidence higher trust in the prime minister after the emergence of the COVID-19 crisis than opposition supporters. That is, the centre-left Spanish Socialist Workers' Party (PSOE) supporters are likely to trust the prime minister more than non-supporters. In turn, the main opposition party supporters, the right-wing People's Party (PP) supporters, are likely to trust the prime minister less that non-supporters.
Second, and related to partisanship, ideology is also expected to affect trust. Recent research suggests that ideology might be driving small partisan differences in attitudes and practices related to COVID-19. In particular, right-wing individuals tend to show lower perceptions of COVID-19 severity (Merkley et al., 2020). We measure ideology based on the typical left-right scale (1 = Left; 10 = Right). 11 Finally, we control for education, age and sex. The more educated are expected to exhibit higher levels of trust, although in many cases education does not reach the expected statistical significance (Catterberg and Moreno, 2005;Newton et al., 2018: 47-48; Van Erkel and Van der Meer, 2016). Controlling for age is also important as the COVID-19 pandemic affects those older with much greater severity than the younger. Finally, we also control for sex. Figure 2 presents the evolution of trust in the Spanish prime minister between July 2019 and March 2021, as well as citizens' evaluations of the executive's performance, both regarding the economy and fighting the pandemic (the latter, only from March to July 2020). The time interval covered corresponds to the one for which we have comparable data.

Trust in the Prime Minister and Performance Evaluations in Spain Over Time
With the emergence of the pandemic in Spain, citizens strengthened their trust levels in the prime minister in April by about 10% more than in the previous months. The percentage remained high in the following months after which it started to decline. Similar trends have been identified elsewhere. For instance, the 2021 Edelman (2021) report shows that, among a sample of 11 countries, the increases in trust in the government between January and May 2020 declined in almost all countries by January 2021. Although this trend is consistent with a rally around the flag effect, the fact is that the increase in trust levels was not overwhelming in Spain, as the literature on this effect suggests (Mueller, 1970), and as seen in other cases. For example, the increased trust in the government as a consequence of the pandemic reached 30% both in Australia and New Zealand (Goldfinch et al., 2021). Such differences may be linked to the country's political situation. The Spanish government, in addition to having had a flawed response to the crisis, was not able to promote cooperation with the opposition parties, contrary to what happened in other countries, such as Portugal, Germany, Italy, France and the Netherlands, further penalizing the response to the crisis. Therefore, in Spain, the pandemic has not led to a substantial increase in political trust, but rather seems to have contributed to eroding political institutions (Royo, 2020).
Citizens' positive evaluations of the economy show a surprising increase after the emergence of the pandemic crisis. Despite the fact that the economy is negatively affected by COVID-19 (Podvrsic et al., 2020), the pandemic does not seem to have had a negative impact on the Spaniards assessment of the economy, especially of their personal economic conditions. Individuals are much more optimistic about their personal economic situation after March 2020. This could be due to the reduction of some daily expenses with the lockdown (e.g. travel expenses) and, in the longer run, to the mitigation of the economic impact of the pandemic among the most vulnerable social groups as a consequence of the government's social policies (see, for example, Gobierno de España, 2020). With regard to sociotropic performance evaluations, the fact that the barometer question was slightly modified almost certainly contributed to the abrupt increase in April and May (the dotted part of the line). Instead of merely asking about the general economic situation in Spain, as in all the other months, the April and May barometer asked about the general economic situation in Spain at the margin of  Then, in June and July, the assessment of the national economy is only a little more positive than in the pre-pandemic months.
Concerning the executive's response in fighting the pandemic crisis, we only have data for 4 months from April to July. These are, however, the critical months to be analysed. Figure 2 shows very high initial support for the government's measures in fighting the pandemic. In April, 73% of the respondents considered that the implemented measures were either necessary or very necessary. The percentage drops to 60% in May, and to 54% in the next months. Summing up, positive evaluations of the executive's performance are generally high in the emergence of the pandemic, then declining in the following months (see Table A1 in the online appendix for further details regarding the distribution of the main variables).

Explaining Trust in the Prime Minister in Times of Pandemic
Given the binary nature of the dependent variable, we use logistic regressions to test our hypothesis. To check whether multilevel modelling was necessary, we first ran intercept only models. The calculation of the intra-class correlation coefficient shows that only a small amount of variance is explained at the month level (about 1%), thus discarding the need for multilevel analysis.
A preliminary assessment of the importance of the pandemic crisis in trust in the prime minister shows a statistically significant effect, as measured by the month-dummies variables (Table A2 in the online appendix). Looking at the whole period under analysis (February to July), and taking February as the reference, the effect of the month-dummies is negative in March (note that the March survey fieldwork took place before the lockdown was declared). It rises substantially in April (explaining an increase of about 50 percentage points in the odds of respondents trusting the prime minister compared to February), declining slightly in May (explaining an increase of only about 27 percentage points). Then, although recovering in June, it decreases substantially in July (explaining an increase of about 39 and 15 percentage points, respectively). These results are consistent with a rally around the flag effect. However, by themselves, they do not give a clear indication of the prevalence of such an effect. That is, if the increased trust in the prime minister is motivated by psychological reasons, as the rally effect presupposes, or by other factors. Only by more broadly analysing the determinants of trust can we possibly identify the real nature of this increase in trust. Indeed, when including other determinants, the effect of COVID-19 as measured by the month-dummy variables vanishes. The pattern is repeated when focusing on the first critical stage of the pandemic period in Spain (April to July).
Although this prior analysis gives an important clue about the conditions under which an emotional response may prevail, it suffers from reliability problems due to the shortcomings in assessing the government's economic performance, as reported earlier. For this reason, we rely on a month-by-month analysis in order to test the hypothesis. Table 1 presents this analysis.
Aimed at assessing the effect of citizens' evaluations of the economic performance and of the policy response to the pandemic in its early months, the findings in Table 1 provide broad support for our hypothesis. 12 More than emotional-related factors (such as citizens' concern or fear due to the COVID-19 disease), it is performance evaluation factors (of the economy and of the policy response to the pandemic) -and party ties -that have a greater effect on trust in the Spanish prime minister. 13 Table 1 first looks, in model 1, at the isolated effect of emotional-related factors on trust in the prime minister. Data reveal some effect promoted by fear of the COVID-19 on trust after the pandemic outbreaks. Consistently with a rally effect, considering the health hazard arising from COVID-19 as the most important problem in Spain significantly explains trust over the period of time under analysis. The effect is stronger in April; asserting a health hazard due to COVID-19 increases the odds of trusting the prime  minister by almost 60 percentage points. The odds are smaller in the following months, increasing the likelihood of trusting by about 30-45 percentage points between May and July. Respondents' concern about the COVID-19 is also statistically significant but only in April and in the opposite direction; the more concerned are 14 percentage points less prone to trust the prime minister. Also contrary to the expectations raised by the rally effect, considering the health hazard arising from COVID-19 as the most important personal problem is only significant in June. These emotional-related variables become mostly insignificant when the other determinants are added in the fully specified models, thus aligning with the idea that concern and fear underlying emotional-related responses are not core factors fuelling trust following the pandemic outbreaks. Prior research had already found that people's concern with the disease was not positively associated with trust in government (Dryhurst et al., 2020). Furthermore, emotional responses may not take place after the occurrence of major external crisis, such as natural disasters in which citizens suffer severe losses (e.g. Carlin et al., 2014;Katz and Levin, 2016). This has also been demonstrated to be the case with pandemics, as in Netherlands, in 2009, concerning influenza A (H1N1), as trust declined in the peak of the crisis ( Van der Weerd et al., 2011). Thus, the severity of the pandemic in Spain may contribute to explain why a cognitive response seems to prevail over an emotional one. In addition, citizens' sociotropic economic evaluations as well as of the performance in fighting the pandemic are both consistently significant and in the expected direction; more positive assessments, either of the national economy or of the policy response to the pandemic, are related to higher trust levels. The most robust predictor is citizens' evaluations of the Spanish economy. More positive evaluations increase the chances of the respondent trusting the prime minister by 76% in April and May, 69 in June and 85 in July (the respondent's own economic evaluation does not contribute to explain trust). Although it is not possible to directly compare coefficients before and after the pandemic emerges, the data suggest a strong consistency of the relative importance of economic evaluations. Positive evaluations of the Spanish economy are systematically the second most important factor driving trust (only surpassed by the support for the PSOE). Positive evaluations of the policy measures to fight the pandemic also substantially increase the odds of trusting the prime minister (by 52 percentage points in April, 25 in May, 36 in June and 68 in July). Moreover, the probabilities of trusting the prime minister are very stable over time either for sociotropic economic performance or for policy performance in fighting the pandemic. Trust steadily increases as performance evaluations become more positive (see predicted probabilities in Table 2).
In short, our evidence suggests that citizens' cognitive responses were not suspended during the first phase of the COVID-19 pandemic, but instead, their assessments of the executive's performance remained relevant in this context. Furthermore, our data show that the policy response to the crisis is almost as important as the economic performance. These results challenge previous literature suggesting that performance regarding noneconomic issues is of low relevance in explaining political trust (Chanley et al., 2000;Hetherington and Rudolph, 2008;Listhaug and Jakobsen, 2018). The likely explanation is that performance in non-economic areas takes precedence in the eyes of citizens during crises, such as the COVID-19 pandemic (see, regarding the economic crisis, Torcal, 2014). This appears to be especially the case when concerning crisis-related issues, such as fighting the pandemic, thereby impacting trust.
Regarding controls, voting for the government party, the PSOE, strongly explains trust after the pandemic outbreaks, although its relevance decreased. 14 The odds of a The estimates are from Model 2 in Table 2.
PSOE voter trusting the prime minister is in April four times higher than a PSOE nonvoter, and between two and four times higher between May and July (as found in Australia and New Zealand, where a strong effect on trust was found among voters for the government party; Goldfinch et al., 2021). Consistently, the chances of a PP voter trusting the prime minister are much lower than a PP non-voter (around 80 percentage points lower, except in May, when the percentage decreases to 32). In addition, being a right-winger consistently reduces the chances of trusting the prime minister, either before or after the pandemic emerges (by about 20-36 percentage points). This implies that party bonds and ideology remain effective after the pandemic breakout, thus reiterating that trust in the prime minister is mainly driven by cognitive reasons. Sociodemographic controls are either non-significant or offer a small contribution to explain the dependent variable.

Conclusion
Based on pre-and post-pandemic data in Spain (February to July 2020), this research aimed to analyse the importance of citizens' evaluations of the executive's performance in explaining the increase in trust in the prime minister after the emergence of the pandemic. In particular, we investigated the importance of cognitive factors, such as citizens' assessments of the economic performance and of the policy response to the crisis, compared to psychological factors, such as concern and fear regarding the COVID-19 disease.
Our results lend support to the idea that citizens' assessments of government performance were not suspended with the outbreak of the pandemic. On the contrary, citizens remained able to hold the government accountable for its performance, not only in terms of responding to the pandemic crisis, but also more broadly for economic performance. This evidence challenges previous conclusions that stressed the importance of emotional factors in explaining the increase in political trust, discarding citizens' conventional cognitive responses (Esaiasson et al., 2020;Schraff, 2020). We do not conclude that emotional factors are irrelevant, quite the opposite. They are important when no other determinants are included in the analysis, as happened in most of the early research about the effects of the pandemic outbreak on trust. What our data suggest is that emotional responses lose importance when the evaluations of executive's performance are considered. As in the pre-pandemic period, what explains trust in the Spanish prime minister after the emergence of the pandemic is essentially a cognitive response based on evaluations of the executive performance (as well as on party bonds), rather than an emotional response.
Although results cannot be generalized to other national contexts, our findings open a new perspective on the explanation of political trust in the pandemic context, in particular, on the relative importance of cognitive factors compared to emotional factors. Naturally, these results need comparative research in order to be consolidated.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplementary information
Additional supplementary information may be found with the online version of this article. Context Table A1. Relative Distributions of the Dependent and Independent Variables. Table A2. Explaining Trust in the Spanish Prime Minister in February to July, and in April to July 2020 (DV: 0 = don't trust; 1 = trust). Table A3. Explaining Trust in the Spanish Prime Minister during the Early Months of the COVID-19 Pandemic (DV: 1 = no trust at all; 2 = little trust; 3 = some trust; 4 = lot of trust).

Notes
1. Data from World Health Organization, available at: https://covid19.who.int/region/euro/country/es. 2. Data from Trading Economics, available at: https://tradingeconomics.com/spain/indicators. 3. Data from the Pew Research Centre, available at: https://www.pewresearch.org/fact-tank/2020/11/17/ europeans-approved-of-eus-handling-of-covid-19-this-summer-but-much-has-changed-since/. 4. Available at: http://www.cis.es/cis/opencm/FR/11_barometros/index.jsp 5. The question is 'Does the President of the Government, Pedro Sánchez, personally inspire you with a lot of trust, some trust, little trust or no trust at all?' 6. The question is 'To begin with, could you tell me if the current situation regarding the COVID-19 coronavirus worries you very much, somewhat, a little or not at all?' (the median point in the scale was omitted in the wording of the question). 7. The questions are 'In your opinion, what is the main problem that currently exists in Spain?' and 'And what is the problem that affects you personally the most?' 8. The question in April and May is as follows: 'Referring to the general economic situation in Spain at the margin of Covid-19, how did its quality change? Is it very good, good, bad or very bad?' For the remaining months in the sample, the question is 'Referring to the general economic situation in Spain today, how would you rate it: very good, good, bad or very bad?' (the median point in the scale was omitted in the wording of the questions). 9. The question is 'How would you rate your personal financial situation today: very good, good, bad or very bad?' (the median point in the scale was omitted in the wording of the question). 10. The question is 'Do you consider that the measures that have been adopted in Spain to fight Covid-19 have been: very necessary, necessary, little or not necessary at all?' (the median point in the scale was omitted in the wording of the question). 11. The question is 'When talking about politics, the expressions left and right are normally used. On a scale of 10 points, like a thermometer, ranging from 1 to 10, in which 1 means "the furthest to the left" and 10 "the furthest to the right", where would you place yourself?' 12. For the sake of validation of the results, the analysis was also run using the original ordinal scale measuring trust in the prime minister (see Table A3 in the online appendix). The results are highly consistent across both analyses. We rely on logistic regression instead of ordinal as the former allows for a more direct and intuitive interpretation of results. 13. The tests of collinearity for the variables included in the regression models show that tolerance is always close to one and the variance inflation factor (VIF) slightly above one, thus discarding any collinearity. 14. The mediating effect of partisanship in the relationship between performance evaluations and trust in the prime minister was also tested for each month. The interactions reveal that there is no important mediating effect of partisanship, as coefficients are seldom statistically significant in the expected direction, being most of the times non-significant or of the opposite sign.