Lost in between crises: How do COVID-19 threats influence the motivation to act against climate change and the refugee crisis?

While the COVID-19 pandemic has been found to undermine mental health, it is unclear how it may impact individuals’ motivation to tackle other global crises. There are at least two perspectives on how COVID-19 might psychologically impact how people respond to other global crises. The finite-pool-of-worry hypothesis suggests that worrying about one issue might diminish worry about other issues since individuals have a limited capacity of worry. Conversely, the affect-generalization hypothesis advocates that worry about an issue might generalize to other issues and increase general levels of worry. To test these competing hypotheses, the present research investigated how threats activated by the COVID-19 pandemic might affect individuals’ interest in and motivation to address climate change (Study 1) and the refugee crisis (Study 2) by assessing pro-environmental behavior and prosocial behavior toward refugees, respectively. The results showed that exposure to COVID-19 threats elevated anxiety levels, and trait anxiety, psychological distance, and future orientation moderated this effect. While COVID-19 threats did not influence pro-environmental and prosocial behavior and intentions, exploratory analyses uncovered that being psychologically closer to COVID-19 might predict an increase in pro-environmental and prosocial behavior and intentions, pointing to the affect-generalization hypothesis.

Modern life requires the world to deal with multiple ongoing crises. While crises such as the climate and refugee crises require sustained collective action, the attention is situationally drawn to other crises that are psychologically closer and require relatively more urgent action, as in the case of the COVID-19 pandemic. Indeed, between January and October 2020, at least 25% of front-page online news articles in 11 countries covered COVID-19 (Krawczyk et al., 2021), leaving less space for other global issues, such as climate change and the refugee crisis. While research has shown the detrimental effects of the pandemic on mental health, it is unclear how it may impact individuals' motivation to tackle other global crises. Even if the other crises may be less urgent, or concern the self less than other people, it is important to understand how people deal with information about multiple crises. Accordingly, the present research aims to uncover the possible effects of the activation of threats posed by the COVID-19 pandemic on motivation to act against climate change and the refugee crisis by examining pro-environmental and prosocial behavior.

COVID-19 and other global crises
In March 2020, the COVID-19 outbreak became a global health crisis and had a serious impact on people's lives. With enforced lockdowns and closed borders, many people experienced high levels of uncertainty about the future, loneliness, fear for their health, and concern over the health of loved ones (e.g., Mertens et al., 2020). The pandemic elevated individuals' levels of stress, anxiety, and depression while lowering their life satisfaction (e.g., Wang et al., 2020;Xiong et al., 2020).
Since COVID-19 was perceived as more urgent and psychologically closer than both climate and refugee crises (Van Lange & Huckelba, 2021;Van Lange & Rand, 2021), the pandemic pushed these crises into the background. While COVID-19 increasingly dominated the newspapers across 50 countries in the early months of 2020, the media coverage of climate change substantially decreased over the same period (Pearman et al., 2021). Similarly, the troublesome refugee situation resulting from a failing deal between the European Union and Turkey received little attention from the media during the pandemic (Hutt, 2020;Knoll & Bisong, 2020).
Given the pandemic's influence on people's mental health and the media attention to other issues, COVID-19 might also influence the motivation to take action against other global crises. In the present research, we explore the influence of COVID-19 on acting to mitigate climate change and the refugee crisis by examining pro-environmental and prosocial behavior, respectively. Accordingly, we discuss two models that are grounded in different reasoning leading to two competing hypotheses (see Van Lange et al., 1997 for a similar approach). On the one hand, since individuals have limited cognitive and emotional resources, threats of COVID-19 could decrease motivation to fight other crises. On the other hand, the worry induced by COVID-19 threats could generalize to other crises and derive action against them. In the following section, we discuss these hypotheses and existing evidence in greater detail.

The finite-pool-of-worry hypothesis
Considering that individuals have limited cognitive and emotional resources, COVID-19 might decrease attention to and worry about other crises. Attention is a cognitive resource with limited capacity (Kahneman, 1973). Given the number of everyday life concerns competing for one's attention, such as economic survival and relational problems, more distant issues, such as climate change and the refugee crisis, might capture less of one's attention (Weber, 2006(Weber, , 2010. According to the finite-pool-of-worry hypothesis, worry is also a limited resource, making it difficult for individuals to worry about multiple issues at the same time (Linville & Fischer, 1991). In line with this hypothesis, for example, the percentage of Americans who perceived climate change as a very serious issue decreased from 47% in 2006 to 35% in 2009, a period covering the onset and progression of the global economic crisis (Pew Research Center, 2009). Although this change in climate opinion could also be a result of the shifting political cues (Mildenberger & Leiserowitz, 2017), it suggests that pressing concern over other problems, such as financial or political issues, might cut down attention to and worry about environmental and humanitarian issues.
Preliminary evidence for the finite-pool-of-worry hypothesis showed that greater concern about one risk can reduce concern about other risks. Hansen et al. (2004) were the first, and only, to empirically test this hypothesis by examining how increased concern about climate variability might influence concern about other issues. After being presented with either a forecast of a very dry spring (climate concern condition) or climate information of the past seasons (control condition), farmers reported the extent to which they were worried about the climate and other risks (e.g., political uncertainty). Accordingly, farmers who viewed the pessimistic forecast had a higher risk perception of climate change and lower concern about other risks compared to those in the control condition. Furthermore, in both conditions, farmers who reported greater concern about political risk worried less about the climate. While this study provided early evidence for the finite-pool-of-worry hypothesis by demonstrating that worry about one issue might dampen worry about other issues, it also had a notably small sample size (N = 14). Nevertheless, in light of these findings, we could expect that being preoccupied with COVID-19 undermines concern over climate and the refugee crises and decreases pro-environmental and prosocial behavior and intentions.
Despite the evidence for the finite-pool-of-worry hypothesis, recent studies suggest that COVID-19 might not influence climate change worry and prosociality. For example, a longitudinal panel survey administered in April 2019 and June 2020 provided very little evidence for decreased climate change worry (Evensen et al., 2021). Specifically, while social media data showed that attention to climate change diminished, individuals perceived climate change as a greater threat than the COVID-19 pandemic, suggesting that climate change might have become a permanent worry for many people in recent years. The impact of COVID-19 on prosociality was also recently investigated in an experimental study, in which participants were presented with negative environmental or economic, positive environmental or economic, or neutral outcomes of the pandemic (Fanghella et al., 2021). Accordingly, the researchers measured how negative and positive primes of the pandemic might influence participants' prosocial behavior and expectations about future environmental quality and the economy. Interestingly, while primes influenced participants' expectations, they did not have an impact on prosociality. Thus, there is mixed evidence for the finite-pool-of-worry hypothesis, in that the latter findings suggest that the COVID-19 threats might not influence the motivation to act against other crises.

The affect-generalization hypothesis
A hypothesis that competes with the finite-pool-of-worry hypothesis is the affect-generalization hypothesis, which points to a possible spillover effect of worry. Specifically, worrying about one risk can make an individual more worried in general about all other threats. A classic case in point is a study by Johnson and Tversky (1983) showing that the perceived risk of one threat might generalize to other threats. But there is also recent evidence. A cross-cultural correlational study found that while attention to climate change decreased with increasing attention to and worry about COVID-19, worry about climate change did not decrease (Sisco et al., 2021). Instead, participants who had more negative experiences with COVID-19 worried more about climate change.
Worrying might be effective in mobilizing individuals to prevent or tackle problems. As worry reflects the significance of preventing an unwanted outcome by taking action, it can motivate productive behavior to deal with an issue by focusing one's attention on the issue and its undesirable consequences (Sweeny & Dooley, 2017). In the context of climate worry, for example, worrying about climate change was, directly and indirectly, related to taking and supporting climate action (Bouman et al., 2020).
A spillover effect might also be present in the motivational benefits of worry. For instance, people with greater COVID-19 worry expressed greater support for climate change policies, even when controlling for relevant covariates (e.g., political ideology) (Sisco et al., 2021). Moreover, the fear of COVID-19 increased consumers' environmental concerns, which in turn increased their willingness to make sacrifices to stay at greener hotels (Jian et al., 2020). Similarly, although no relationship was found between COVID-19 risk perception and pro-environmental behavior, risk perception of COVID-19 was positively associated with the intentions to protect the environment (O'Connor & Assaker, 2021). While none of these studies was experimental, their findings suggest that worrying about COVID-19 threats might promote worry about and motivation to act against other crises.

The present research
To test the two competing hypotheses and address the mixed findings in the literature, the present research aimed to shed light on the influence of the threats posed by COVID-19 on the motivation to take action against other crises, such as climate change and the refugee crisis. To this end, we examined how exposure to threats activated by COVID-19 affects anxiety levels (Studies 1 and 2), pro-environmental behavior and intentions (Study 1), prosocial behavior and intentions toward refugees (Study 2), and the interest in reading about social issues (Studies 1 and 2). Accordingly, we predicted that exposure to COVID-19 threats would induce higher levels of anxiety in both studies. In light of the finite-pool-of-worry hypothesis, we expected that the activation of threats posed by COVID-19 would undermine pro-environmental behavior and intentions in Study 1, prosocial behavior and intentions toward refugees in Study 2, and interest in reading about social issues in both studies. Based on the affect-generalization hypothesis, however, we predicted seeing the opposite effect, such that COVID-19 threats would promote pro-environmental behavior and intentions, prosocial behavior and intentions toward refugees, and interest in reading about social issues. Additionally, we examined the possible influence of relevant individual difference variables based on the existing literature for exploratory purposes. We measured trait anxiety in both studies considering that trait anxiety influences the perception of and response to threat (e. g., Byrne & Eysenck, 1995;Eysenck, 1992;Sussman et al., 2016). Furthermore, according to the construal level theory, one's response to a specific event can be related to the psychological distance from that event (Liberman & Trope, 1998;Trope & Liberman, 2003), and psychological distance might play a role in determining attitudes and behavior toward a global crisis (e.g., Maiella et al., 2020). Therefore, we also measured the psychological distance from COVID-19 partly based on individuals' direct (i.e., own) or indirect (i.e., close others) experience.
Many social and environmental problems, including climate change and the refugee crisis, can be considered within the social dilemma framework (Khachatryan et al., 2013;Van Lange et al., 2018) as there is a social conflict (self-versus collective interests) and/or temporal conflict (long-versus short-term interests). Accordingly, biospheric values, future orientation, and social value orientation were theorized and empirically found to promote environmental concern and behavior (e.g., Bouman et al., 2020;Gärling et al., 2003;Milfont et al., 2012;Stern & Dietz, 1994;Van Lange, 2021;Van Lange & Huckelba, 2021;Wang et al., 2021). We, therefore, explored the influence of these variables in Study 1. We measured social value orientation in Study 2 as well since greater prosocial orientation was theorized to lead to more prosocial actions (Van Lange et al., 1997) and was found to increase the willingness to help refugees (Böhm et al., 2018). In line with the theory of planned behavior (Ajzen, 1991), attitudes toward refugees might play a role in one's intentions and behavior of helping refugees. Given that greater threat perception might lead to more negative attitudes toward refugees (Stephan et al., 2005;Yitmen & Verkuyten, 2018), we also examined the influence of both threat perception of and attitudes toward refugees.

Study 1
The present research was approved by the ethical committee (Vaste Commissie Wetenschap en Ethiek [VCWE]) of VU Amsterdam's Faculty of Behavioral and Movement Sciences with the corresponding number VCWE-2021-097, and all data were collected in May 2021. Both studies were preregistered, and preregistrations and the data can be found at Open Science Framework (see https://osf.io/q798x/).

Participants and design
Since there has not been much research on the influence of COVID-19 threats on pro-environmental behavior, we estimated a small effect size. Based on the calculations in G*Power, to achieve an 80% chance of detecting a small effect size (f 2 = 0.02) at an alpha level of 0.05, the required sample size was 395 participants.
A total of 433 participants from the UK were recruited via Prolific (www.prolific.co), a crowdsourcing platform to recruit participants online for research. As for inclusion criteria, we targeted adults (ages 18-65) with no literacy difficulties since the study involved reading and writing. We also included two attention checks in two of the questionnaires, and as preregistered, we excluded the data of participants who failed any of these attention checks (N = 4) as well as participants who took more than 30 min to complete the study (N = 27). The latter exclusion was to ensure that participants were not distracted during the study and that they completed the study without major interruptions.
The final sample size consisted of 402 participants (237 females) between the ages of 18 and 65 (M = 36.29, SD = 11.81). Most participants (61.4%) have completed at least a bachelor's degree, and most (74.4%) had a yearly personal income lower than £40.000. Participants perceived themselves and their families mostly as middle class (47.3%) or working class (44.5%). While the majority (57.2%) did not have any children, the rest of the participants had mostly one (13.7%) or two (17.9%) children.
The present study was an experimental study with a betweensubjects design and was designed to test the influence of COVID-19 threats on anxiety, pro-environmental behavior, environmental behavioral intentions, and interest in social issues. Accordingly, participants were randomly assigned to  or control (N = 203) conditions.

Procedure
The study was prepared on Qualtrics and advertised on Prolific as a study on decision-making. Participants were informed that they would receive £1.40 as compensation for their time and that three randomly chosen participants would receive a bonus of £10. After reading the information letter and providing their consent to participate in the study, participants responded to a set of questionnaires measuring trait anxiety, social value orientation, future orientation, biospheric values, and psychological distance from COVID-19, which were presented in random order. Upon completing these measures, participants were told that they were going to read a short article taken recently from a widely respected source. Participants were then randomly assigned to read a text about either the COVID-19 pandemic or a neutral topic (i.e., the history of doors). After reading the text, participants answered some memory-check questions about the text. Next, participants reported their anxiety levels, after which they responded to manipulation check questions. Afterward, measures of pro-environmental, environmental behavioral intentions, and selection of social issues for reading were presented in random order. Finally, participants responded to some control questions about the text and questions about political orientation and demographics. Participants were then debriefed, thanked for their participation, and dismissed.

Measures and materials
Trait Anxiety. Trait anxiety was measured using the State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983). STAI has two subscales consisting of state and trait anxiety. Based on the purposes of the current study, participants responded only to the 20-item Trait Anxiety Subscale. The reliability of the subscale was excellent (α = 0.95) (see Supplementary Material for the items, scales, and descriptive statistics of all measures).
Social Value Orientation. Social value orientation was measured using the Triple Dominance Measure (Van Lange et al., 1997). Participants were asked to imagine that they have been randomly paired with another person whom they would never meet. Across 9 items, participants were asked to choose between three options that offer points to the self and the other. Respondents were classified as cooperators, individualists, or competitors when they made 6 out of 9 choices consistent with a certain social value orientation. In the present study, 280 participants were cooperators, 86 were individualists, and 13 were competitors.
Future Orientation. Future orientation was measured using the 14item Consideration of Future Consequences Scale (CFC-14; Joireman et al., 2012). The internal consistency of this scale was good (α = 0.88).
Biospheric Values. To measure biospheric values, we used the environmental version of the Schwartz Value Survey (E-SVS; Steg et al., 2014), which assessed hedonistic, egoistic, altruistic, and biospheric values. Only the items measuring biospheric value were included in the analysis, and the internal consistency of these items was excellent (α = 0.90).
Psychological Distance. Participants responded to three items measuring the psychological distance from COVID-19. These three items were averaged into one reliable scale of psychological distance from COVID-19 (α = 0.90).
COVID-19 Threat Manipulation. To manipulate the exposure to COVID-19 threats, we constructed a text reflecting COVID-19 threats and a control text with information about a neutral topic. Accordingly, participants in the experimental condition read about a pessimistic prospect of the COVID-19 pandemic and viewed three related photos. We made up this text by putting together parts of online news articles and deliberately included health-related and economic threats of the pandemic. Participants in the control condition read about the history of doors and viewed three related photos. We constructed this neutral text using parts of blogs written about the history and symbolic meanings of doors. Both texts were at similar lengths (see Supplementary Material for the texts).
After the texts, we included two multiple-choice questions per text to measure whether participants carefully read the texts (see Supplementary Material), which served exploratory purposes. Furthermore, to reinforce the influence of the COVID-19 threats, we asked participants in the COVID-19 condition to briefly write down what worried them the most about the pandemic with a minimum of 130 characters. Participants in the control condition wrote down what they would pay attention to if they were to buy a new door.
Anxiety. Anxiety levels after reading the text were measured using the Visual Analog Scale for Anxiety (VAS-A; Hornblow & Kidson, 1976), which consists of one single item: "How anxious do you feel right now?" Participants rated their anxiety levels from 0 (not at all anxious) to 100 (extremely anxious). In accordance with the general use of VAS-A, the numeric values on the scale were not shown to the participants. This brief measure of anxiety was chosen to sustain the influence of the COVID-19 text for the remaining measures of the dependent variables.
Manipulation Check Questions. To make sure that the COVID-19 manipulation was successful, participants responded to five manipulation check questions.
Pro-environmental Behavior. We measured pro-environmental behavior by asking participants how much of their bonus they wanted to donate to an environmental charity to contribute to climate change mitigation if they were chosen to receive one of the three £10 bonuses. This charity was World Wildlife Fund (WWF); however, the name of the charity was only disclosed in the debriefing to avoid any biases. After completing the data collection, the bonuses were distributed, and according to the answers of the recipients, £15 was donated to WWF. A similar financial donation measure has been used by previous studies and proved to be a valid measure of environmental behavior (e.g., Clements et al., 2015).
Environmental Behavioral Intentions. We measured environmental behavioral intentions using the 4-item scale used by Pan et al. (2018). Accordingly, participants indicated their willingness to engage in four types of environmental behaviors (eco-management, consumerism, persuasion, and civic actions) at that very moment on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). The reliability of this scale was good (α = 0.84).
Interest in Reading About Social Issues. To measure interest in reading about social issues, we asked participants to imagine that they would read a final piece about a topic of their choice. Participants then indicated their first and second preferences of topics from a list comprised of COVID-19, climate change, refugee crisis, fashion, sports, and recommended online programs. The first three topics formed the social issues while the last three formed the entertainment category.
Control Questions. We administered three items to control for the perceived accuracy, informativeness, and convincingness of the text.
Political Orientation. We measured political orientation with two standard items: "On a scale from left to right (where 0 means left and 10 means right), what is your political orientation?" and "On a scale from liberal to conservative (where 0 means liberal and 10 means conservative), how liberal or conservative are you?" Participants responded to both items on an 11-point Likert scale from 0 (left/liberal) to 10 (right/ conservative).
Demographics. At the end of the survey, we asked participants to answer some demographic questions. These involved questions about age, gender, educational attainment, personal income, social class, and the number of children.
All scales and questions mentioned in this section can be found in the Supplementary Material.

Analyses
We conducted confirmatory and exploratory analyses. Before each analysis, we checked and observed that the assumptions were met. First, we created a dummy variable for the presence of COVID-19 threats. To examine the effect of COVID-19 threats on anxiety, pro-environmental behavior, and environmental behavior intentions, we conducted three linear regression analyses. As for the influence of COVID-19 threats on interest in social issues, we conducted one linear regression analysis to examine the number of social issues in participants' top two topic preferences and two binomial logistic regression analyses to examine only the top preferences and whether participants included COVID-19 in their top two preferences.

Manipulation checks
To make sure that the COVID-19 text was more negative than the neutral text, we conducted manipulation checks. We expected that participants who read about COVID-19 would find the text they read more disturbing and threatening than those who read about the history of doors. Also, we predicted that the COVID-19 text would be found to present a more negative future than the text in the control condition. An analysis of variance (ANOVA) revealed that participants who read the COVID-19 text found the text significantly more disturbing (M = 4.53, SD = 1.52) than the participants who read the neutral text (M = 1.36, SD Accordingly, as expected, compared to the text in the control condition, the COVID-19 text was more disturbing and threatening, and it displayed a more negative prospect for the future.

Confirmatory analyses
To test our main hypotheses, we conducted four linear regression analyses with each dependent variable by adding the COVID-19 threat as the independent variable (see Supplementary Material for descriptive statistics and the correlation matrix for all continuous variables).
Next, to examine moderation effects, we conducted five hierarchical multiple regression analyses with separate models for each dependent variable with each individual difference variable (i.e., trait anxiety, biospheric values, social value orientation, future orientation, and psychological distance from COVID-19) and their interaction. In all analyses, we included the same control variables, namely, age, gender, educational attainment, income, social class, number of children, political orientation, and three control variables for the text (i.e., how accurate, informative, and convincing participants reported the text to be). We mean-centered age, two political orientation variables, and control variables for the text, created dummy variables for gender and social class, and included education and personal income as continuous variables. In all of these analyses, control variables were entered in the first step, the COVID-19 dummy variable in the second step, an individual difference variable in the third step, and the interaction term of the individual difference variable in the fourth step. For any significant moderation effects, we conducted simple slope analyses.
Since the addition of the COVID-19 threat variable to the models derived significant results only for one dependent variable, anxiety, we will report the results of the hierarchical regression analyses only for anxiety in the following sections. The summary of all hierarchical regression analyses can be found in the Supplementary Material.
Anxiety. We conducted a linear regression analysis to examine the effect of COVID-19 threats on anxiety. Results are displayed in Table 1. Accordingly, participants who read about the COVID-19 pandemic reported significantly elevated levels of anxiety (M = 47.15, SD = 28.33), compared to those who read about a neutral topic (M = 36.06, SD = 28.77), F(1,400) = 15.15, p < .001; R 2 = 0.04.
Next, we conducted five hierarchical regression analyses to control for demographics and examine the moderating effects of five individual difference variables in separate models. In the first regression, in Step 1, we entered the control variables. Results showed that the model with all control variables was significant, F(11,362) = 2.03, p < .05, and accounted for 5.8% of the variation in anxiety. In Step 2, we entered the dummy variable for COVID-19 threat, which explained an additional 2.4% of the variation in anxiety, and this change in R 2 was significant.
F(1,361) = 9.48, p < .01. In Step 3, we entered centered trait anxiety into the model. This explained an additional 24.8% of the variation in anxiety, and this change in R 2 was significant, F(1,360) = 133.18, p < .001. Finally, in Step 4, we entered the interaction term of trait anxiety, which explained an additional 1.9% of the variation in anxiety. This change in R 2 was significant, F(1,359) = 10.36, p < .01. To further investigate the moderating effect of trait anxiety, we conducted simple slope analyses. Accordingly, COVID-19 threats increased anxiety levels more at low levels (1 SD below mean) of trait anxiety (B = 17.22, SE = 3.37, β = 0.30, p < .001) compared to high levels (1 SD above mean) of trait anxiety (B = 3.55, SE = 3.38, β = 0.06, p = ns) (see Fig. 1). In the four remaining regressions, we repeated Step 1 and Step 2 of the first regression and added other individual difference variables in Step 3 and interaction terms in Step 4. Only the hierarchical regressions with significant results will be reported here (see Supplementary Material for all results).
In one regression, we entered centered future orientation in Step 3, which did not contribute to the model (R 2 = 0.001; F(1,360) = 0.23, p = ns). Introducing the interaction term in Step 4 explained an additional 2.6% of the variance in anxiety, which was a significant change in R 2 , F (1,359) = 10.33, p < .01. To further examine the moderating effect, we conducted simple slope analyses. COVID-19 threats increased anxiety levels more at high levels (1 SD above mean) of future orientation (B = 18.64, SE = 4.01, β = 0.32, p < .001) compared to low levels (1 SD below mean) of future orientation (B = 3.53, SE = 4.01, β = 0.06, p = ns) (see Fig. 2).
In another regression, we entered reverse-coded and centered psychological distance to the model in Step 3. This explained an additional 4% of the variation in anxiety, and this change in R 2 was significant, F (1,360) = 16.42, p < .001. Introducing the interaction term in Step 4 explained an additional 2.8% of the variation in anxiety, and this change was also significant, F(1,359) = 11.70, p < .01. To investigate the moderating effect of psychological distance, we conducted simple slope analyses. Accordingly, at low levels (1 SD below mean) of psychological distance from COVID-19, COVID-19 threats led to more anxiety (B = 21.87, SE = 3.90, β = 0.38, p < .001), compared to at high levels (1 SD above mean) of psychological distance, (B = 2.59, SE = 3.91, β = 0.05, p = ns) (see Fig. 3).
Overall, the results showed that reading about COVID-19 induced higher levels of anxiety compared to reading about a neutral topic. Although individual difference variables explained most of the variance in anxiety after reading the text, the type of text still had an influence on anxiety levels.
Pro-environmental Intentions and Behavior. We conducted two linear regression analyses to examine the effect of COVID-19 threats on pro-environmental intentions and behavior. Results are displayed in Table 2. Accordingly, participants who read about the COVID-19 pandemic did not differ from participants who read about a neutral topic in their environmental behavioral intentions, F(1,400) = 0.72, p = ns; R 2 = 0.002, or in their donations to an environmental organization to help mitigate climate change, F(1,400) = 0.02, p = ns; R 2 = 0.000. In other words, COVID-19 threats did not influence pro-environmental intentions and behavior.
Interest in Social Issues. We conducted two binomial logistic and one linear regression analyses to examine the effect of COVID-19 threats on interest in social issues. First, we conducted a logistic regression by entering the dummy variable of COVID-19 threats as the independent Note. The independent variable in all four models is the COVID-19 threat. a Interest in social issues indicates the number of social issues selected by participants as their top two preferences of topics for reading. b p < .001.  variable and the number of social issues (i.e., COVID-19, climate change, refugee crisis) in participants' top preferences as the dependent variable in the model. Results showed that the exposure to COVID-19 threats did not influence the number of social issues selected by participants as their top preference for reading (see Table 2 for results). When we conducted a linear regression analysis to examine the interest in reading about social issues by taking into account the top two preferences, the participants in experimental and control conditions also did not differ in their interest in reading about social issues (p = .061) (see Table 1 for results). We conducted another binomial logistic regression analysis to investigate the influence of COVID-19 on the likelihood that participants include COVID-19 for reading in their top two preferences. The regression model was statistically significant, χ 2 (1) = 4.46, p < .05.
The model explained 1.5% (Nagelkerke R 2 ) of the variance in selecting COVID-19 for reading and correctly classified 55.2% of cases. Accordingly, participants who read about the COVID-19 pandemic were 1.53 times more likely to prefer to read about COVID-19 than those who read about a neutral topic. We examined possible moderating effects of individual difference variables; however, we did not find any significant effects.

Exploratory analyses
We also conducted some exploratory analyses. First, to examine whether reading the COVID-19 text carefully influenced the results, we conducted the same analyses after excluding the data of participants who failed one or both of the memory check questions asked right after they read the text. The results did not change.
Psychological Distance. Given the moderating effect of psychological distance from COVID-19 on the relationship between COVID-19 threats and anxiety, we conducted three linear regression analyses to further examine the relationship between psychological distance and the dependent variables. The results showed that lower psychological distance from COVID-19 predicted higher levels of anxiety, F(1,400) = 15.19, p < .001; R 2 = 0.04, greater donations to an environmental charity to mitigate climate change, F(1,400) = 14.11, p < .001; R 2 = 0.03, and greater environmental behavioral intentions, F(1,400) = 11.31, p < .01; R 2 = 0.03. Although it is not possible to infer causality, these findings suggest that being psychologically closer to COVID-19 might be related to greater motivation to act against climate change and more pro-environmentally.
As for the interest in reading about social issues, we conducted a binomial logistic regression analysis to examine the relationship between psychological distance from COVID-19 and participants' selection of social issues for reading in their top preferences; however, we did not find any effects. Similarly, when we examined the top two preferences in a linear regression analysis, we did not observe any relationship.
Next, we conducted a binomial logistic regression analysis to examine the relationship between psychological distance and the selection of COVID-19 in the top two preferences. The results showed that lower psychological distance from COVID-19 predicted greater likelihood of COVID-19 being chosen as the top or second preference for reading, χ 2 (1) = 8.66, p < .01, and psychological distance explained 2.8% (Nagelkerke R 2 ) of the variance in interest in COVID-19. Accordingly, being psychologically closer COVID-19 might be associated with a greater interest in COVID-19, although it seems not to be related to interest in other social issues.
Anxiety. We conducted three linear regression and two binomial logistic regression analyses to examine how levels of anxiety after reading about COVID-19 might influence pro-environmental behavior, environmental behavioral intentions, and interest in social issues. While anxiety did not affect environmental intentions and interest in social issues, we found that elevated levels of anxiety predicted greater environmental donations to help mitigate climate change, F(1,400) = 6.75, p < .05; R 2 = 0.02. This suggests that worry resulting from COVID-19 might promote climate action.

Participants and design
As in Study 1, we estimated a small effect size due to insufficient findings on the topic. Based on the calculations in G*Power, to achieve an 80% chance of detecting a small effect size (f 2 = 0.02) at an alpha level of 0.05, the required sample size was 395 participants.
A total of 423 participants from the UK were recruited via Prolific. We used the same inclusion and exclusion criteria as in Study 1, and additionally, we excluded the participants of Study 1 from recruitment. Accordingly, after excluding participants who failed one or more of the attention checks (N = 4) and who spent more than 30 min on the study (N = 19), the final sample size consisted of 400 participants (237 females) between the ages of 18 and 65 (M = 37.67, SD = 12.04). The vast majority of participants were white (86.3%) and did not have a migration background (77.3%). Most participants (52.8%) have completed at least a bachelor's degree, and most (78.5%) had a yearly personal income lower than £40.000. Participants perceived themselves and their families mostly as middle class (42.3%) or working class (46.3%). While the majority (56.5%) did not have any children, the rest of the participants had mostly one (14.2%) or two (21.5%) children.
The present study was an experimental study with a betweensubjects design and was designed to test the influence of COVID-19 threats on anxiety, prosocial behavior toward refugees, intentions to help refugees, and interest in social issues. As in Study 1, participants were randomly assigned to the same COVID-19 threat (N = 201) or control (N = 199) conditions.  Note. In both models, the independent variable is the dummy variable of COVID-19 threat.

Procedure
The study was prepared on Qualtrics and advertised on Prolific as research on decision-making. Participants were informed that they would receive £1.20 as compensation for their time and that three randomly chosen participants would receive a bonus of £10. After reading the information letter and providing their consent to participate in the study, participants responded to measures assessing trait anxiety, social value orientation, and psychological distance from COVID-19, which were identical to those in Study 1. Different from Study 1, participants responded to two different scales measuring attitudes toward refugees and threat perception. All scales were presented in random order. Upon completing these measures, participants were told that they were going to read a short article taken recently from a widely respected source. Participants were then randomly assigned to read a text about either the COVID-19 pandemic or a neutral topic (i.e., history of doors). These texts were identical to the texts in Study 1. After reading the text, participants answered the same memory-check questions about the text. Next, participants reported their anxiety levels, after which they responded to manipulation check questions. Afterward, items to measure prosocial behavior toward refugees, intentions to help refugees, and interest in social issues were presented in random order. Finally, participants responded to some control questions about the text and questions about political orientation and demographics. Participants were then debriefed, thanked, and dismissed.
Since the structure and materials of Study 2 were largely similar to Study 1, we will only provide details of the different measures and materials of Study 2 in the following section.

Measures and materials
Negative Attitudes Toward Refugees. To our knowledge, there has not been a well-validated scale developed to measure attitudes toward refugees. Therefore, we measured participants' general attitudes toward refugees with four items. All items were reverse coded so that higher scores reflected more negative attitudes toward refugees and were averaged into one reliable scale (α = 0.94).
Threat Perception. We measured threat perception toward refugees using modified versions of the realistic and symbolistic threat questionnaires developed by Stephan et al. (1999). All positively phrased items in the questionnaire were reverse-coded so that higher scores represented a greater perceived threat. All items were averaged into one reliable scale (α = 0.94).
Prosocial Behavior Toward Refugees. We measured prosocial behavior toward refugees by asking participants how much of their bonus they would want to donate to a humanitarian charity to contribute to the relief of the refugee crisis if they were chosen to receive one of the three £10 bonuses. As in Study 1, the name of the charity was disclosed in the debriefing to avoid biases. After the data collection, the bonuses were distributed, and according to the answers of the recipients, £11 was donated to United Nations High Commissioner for Refugees (UNHCR).
Intentions to Help Refugees. Intentions to help refugees were measured with the 9-item scale used by Yitmen and Verkuyten (2018). Negatively phrased items were reverse-coded, and all items were averaged into one reliable scale (α = 0.86).
Demographics. In addition to the questions that were asked in Study 1, we also asked participants to report their ethnicity and migration background.

Analyses
We conducted confirmatory and exploratory analyses, making sure that the assumptions were met before each analysis. As in Study 1, we created a dummy variable for the presence of COVID-19 threats to test our hypotheses. To examine the effect of COVID-19 threats on anxiety, prosocial behavior toward refugees, and intentions to help refugees, we conducted three linear regression analyses. We explored the influence of COVID-19 threats on interest in social issues in the same way as in Study 1. Accordingly, we conducted one linear regression analysis to examine the number of social issues in participants' top two topic preferences and two binomial logistic regression analyses to examine only the top preferences and whether participants included COVID-19 in their top two preferences.

Manipulation checks
As in Study 1, we performed manipulation checks and expected the COVID-19 text to be more disturbing and threatening and present a more negative future than the neutral text in the control condition. First, an ANOVA showed that participants in the COVID-19 condition found the text they read significantly more disturbing (

Confirmatory analyses
First, we conducted four linear regression analyses with each dependent variable by adding COVID-19 threat as the independent variable to test our main hypotheses (see Supplementary Material for descriptive statistics and the correlation matrix for all continuous variables).
Next, to examine possible moderation effects, we conducted hierarchical multiple regression analyses with separate models for each dependent variable with each individual difference variable (i.e., trait anxiety, attitudes toward refugees, threat perception, social value orientation, and psychological distance from COVID-19). In all analyses, we included the same control variables as in Study 1 in addition to ethnicity and migration background.
Again, we mean-centered age, two political orientation variables, and three control variables for the text and created dummy variables for gender, social class, ethnicity, and migration background.
We included education and personal income as continuous variables. In all analyses, control variables were entered in the first step, the COVID-19 dummy variable in the second step, an individual difference variable in the third step, and the interaction term of the individual difference variable in the fourth step. For any significant moderation effects, we conducted simple slope analyses.
As in Study 1, the COVID-19 threat variable contributed to the models only for anxiety. Therefore, we will report the results of the hierarchical regression analyses only for anxiety in the following sections. The summary of all hierarchical regression analyses can be found in Supplementary Material.
Anxiety. We conducted a linear regression analysis to examine the effect of COVID-19 threats on anxiety. Results are displayed in Table 3. As in Study 1, participants who read about COVID-19 experienced significantly increased levels of anxiety (M = 47.11, SD = 27.06), compared to those who read about a neutral topic (M = 31.50, SD = 27.70), F(1,398) = 35.07, p < .001; R 2 = 0.08. Next, as in Study 1, we conducted five hierarchical regression analyses to control for demographics and examine moderating effects of individual difference variables. In Step 1, we entered the control variables. Results showed that the model with all control variables was significant, F(13,352) = 1.79, p < .05, and accounted for 6.2% of the variation in anxiety. In Step.
2, we entered the dummy variable for COVID-19 threat, which explained an additional 8.1% of the variation in anxiety and this change in R 2 was significant, F(1,351) = 32.96, p < .001. In Step 3, we entered centered trait anxiety into the model. This explained an additional 29.1% of the variation in anxiety, and this change in R 2 was significant, F(1,350) = 179.65, p < .001. Finally, in Step 4, we entered the interaction term of trait anxiety; however, different than in Study 1, it did not contribute to the model (R 2 = 0.001; F(1,349) = 0.79, p = ns). In the four remaining regressions, we repeated Step 1 and Step 2 of the first regression and entered another individual difference variable in Step 3 and the interaction term in Step 4. We will only report the significant results here (see Supplementary Material for all results).
In one regression, we entered centered threat perception in Step 3, which explained an additional 1.3% of the variation in anxiety, F(1,350) = 5.42, p < .05). Introducing the interaction term in Step 4, however, did not contribute to the model (R 2 = 0.006; F(1,349) = 2.36, p = ns).
In another regression, we entered centered psychological distance to the model in Step 3. This explained an additional 6.6% of the variation in anxiety, and this change in R 2 was significant, F(1,350) = 29.34, p < .001. Introducing the interaction term in Step 4 explained an additional 1.5% of the variation in anxiety, and this change was also significant, F (1,349) = 6.57, p < .05. To further examine the moderating effect of psychological distance from COVID-19, we conducted simple slope analyses. Accordingly, as in Study 1, at low levels (1 SD below mean) of psychological distance from COVID-19, COVID-19 threats led to more anxiety (B = 25.03, SE = 3.69, β = 0.44, p < .001), compared to high levels (1 SD above mean) of psychological distance (B = 9.55, SE = 3.71, β = 0.17, p < .05) (see Fig. 4).
As in Study 1, the results showed that reading about COVID-19 induced higher levels of anxiety compared to reading about a neutral topic. While we replicated the moderating effect of psychological distance, we did not find any moderating effect of trait anxiety in Study 2.
Again, although individual difference variables explained most of the variance in anxiety after reading the text, the type of text still had an influence on state anxiety.
Prosocial Intentions and Behavior Toward Refugees. We conducted two linear regression analyses to examine the effect of COVID-19 threats on prosocial intentions and behavior toward refugees. Results are displayed in Table 3. Accordingly, participants who read about the COVID-19 pandemic did not differ from participants who read about a neutral topic in their intentions to help refugees, F(1,398) = 0.50, p = ns; R 2 = 0.001, and in their donations to a humanitarian organization to help relieve the refugee crisis, F(1,398) = 0.39, p = ns; R 2 = 0.001. In other words, exposure to COVID-19 threats did not influence prosocial intentions and behavior toward refugees.
Interest in Social Issues. As in Study 1, we conducted two binomial logistic and one linear regression analyses to examine the effect of COVID-19 threats on interest in social issues. First, we conducted a logistic regression by entering the dummy variable of COVID-19 threats as the independent variable and the number of social issues (i.e., COVID-19, climate change, refugee crisis) in participants' top preferences as the dependent variable in the model. Results showed that the exposure to COVID-19 threats did not influence the number of social issues selected by participants as their top preference for reading (see Table 4 for results). When we conducted a linear regression analysis to examine the selection of social issues for reading by considering the top two preferences, the participants in experimental and control conditions again did not differ in their interest in reading about social issues (p = .52) (see Table 3). Therefore, participants who read about the COVID-19 pandemic did not choose more or fewer social issues for reading compared to participants in the control condition.
Finally, we conducted another binomial logistic regression analysis to investigate the influence of COVID-19 on the likelihood that participants include COVID-19 for reading in their top two preferences. In contrast to the results of Study 1, this regression model was not statistically significant, χ 2 (1) = 0.038, p = ns (see Table 4). Accordingly, participants who read about the COVID-19 pandemic were no more or less likely to prefer to read about COVID-19 than those who read about a neutral topic. We examined possible moderating effects of trait anxiety, attitudes toward refugees, threat perception, and social orientation value; however, we did not find any significant effects.

Exploratory analyses
As in Study 1, to examine whether carefully reading the COVID-19 text influenced the results, we excluded the data of participants who failed one or both memory check questions, and then we conducted the same analyses. However, the results remained the same.
Psychological distance. As in Study 1, we conducted three linear regression analyses to examine the relationship between psychological distance and the dependent variables. Accordingly, lower psychological distance from COVID-19 predicted higher levels of anxiety, F(1,398) = 26.89, p < .001; R 2 = 0.06, greater donations to a humanitarian charity to help relieve the refugee crisis, F(1,398) = 5.48, p < .05; R 2 = 0.01, and greater intentions to help refugees, F(1,398) = 9.63, p < .01; R 2 = 0.02, which was similar to our findings in Study 1. These findings indicate that being psychologically closer to COVID-19 might be related to increased motivation to help tackle the refugee crisis and prosocial behavior toward refugees.  Note. In both models, the independent variable is the dummy variable of COVID-19 threat.
As for the selection of social issues for reading, we conducted a binomial logistic regression analysis to examine how psychological distance from COVID-19 might be related to the likelihood of reporting a social issue as the top preference for reading. We found that lower psychological distance predicted a greater likelihood of a social issue being reported as the top preference, χ 2 (1) = 15.35, p < .001, and psychological distance explained 5.3% (Nagelkerke R 2 ) of the variance.
Next, to explore the relationship between psychological distance and the number of social issues in participants' top two preferences, we conducted a linear regression analysis. Accordingly, lower psychological distance predicted more social issues being reported by the participants in their top two preferences, F(1,398) = 14.49, p < .001; R 2 = 0.04, indicating that being psychologically closer to COVID-19 might be related to greater interest in social issues.
We then conducted another binomial logistic regression analysis to examine the relationship between psychological distance and the selection of COVID-19 in the top two preferences. The results showed that lower psychological distance from COVID-19 was associated with a greater likelihood of COVID-19 being chosen as the top or second preference for reading, χ 2 (1) = 14.71, p < .001, and psychological distance explained 4.8% (Nagelkerke R 2 ) of the variance.
To understand if the psychological distance from COVID-19 predicted the selection of social issues other than COVID-19, we excluded COVID-19 from the social issue category and conducted a binomial logistic regression and a linear regression analysis to examine the relationship between psychological distance from COVID-19 and the selection of climate change or the refugee crisis in participants' top and top two preferences, respectively. However, we did not find any significant effects. Similar to the findings from Study 1, these results suggest that psychological distance from COVID-19 might be associated with individuals' interest in COVID-19 but not in other social issues.
Anxiety. As in Study 1, we conducted three linear and two binomial logistic regression analyses to how levels of anxiety after reading about COVID-19 might influence prosocial behavior toward and intentions to help refugees and interest in social issues. We did not observe any significant effects.

General discussion
While the impact of COVID-19 on mental health has drawn a great deal of attention, we know very little about how attention to COVID-19 might influence individuals' interest in and motivation to act against other crises. Moreover, existing theories and findings indicate opposite directions. In the present research, we explored how COVID-19 threats might influence the motivation to take action to mitigate climate (Study 1) and refugee (Study 2) crises by examining pro-environmental behavior and prosocial behavior toward refugees, respectively. Also, in both studies, we examined how interest in social issues (i.e., COVID-19, climate change, and refugee crisis) might be affected by exposure to COVID-19 threats. We expected the activation of COVID-19 threats to elevate participants' anxiety levels in both studies. Furthermore, in line with the finite-pool-of-worry hypothesis (Hansen et al., 2004), we expected that threats activated by the COVID-19 pandemic would undermine interest in and motivation to take action against other issues since worry has a limited capacity. Based on the affect-generalization hypothesis, however, we expected worry about one topic to generalize to other issues (Johnson & Tversky, 1983). Our hypothesis about the influence of COVID-19 on anxiety was supported in both studies, and we found moderating effects of trait anxiety, future orientation, and psychological distance from COVID-19. However, we did not find support for either the finite-pool-of-worry or the affect-generalization hypotheses. Accordingly, being exposed to COVID-19 threats did not influence the motivation to act pro-environmentally and prosocially to help tackle climate change and the refugee crisis, respectively. It also did not affect the interest in climate and refugee crises for reading.

Evaluation of hypotheses
The main findings of the present research had noteworthy theoretical implications. Our study did not support the finite-pool-of-worry or affect-generalization hypotheses as we did not find any influence of the exposure to COVID-19 threats on the motivation to help tackle climate change or the refugee crisis. This suggests that these hypotheses might not account for COVID-19 and its influence on climate and refugee crises. Still, our findings were partly in line with the existing literature which found that COVID-19 primes do not influence prosociality (Fanghella et al., 2021) and climate change worry was not affected by the pandemic (Evensen et al., 2021). Also, our findings did not support previous correlational studies that showed a positive association between perceived risk and fear of COVID-19 and environmental intentions (Jian et al., 2020;O'Connor & Assaker, 2021;Sisco et al., 2021).
There can be several reasons why we did not observe the expected results regarding the two competing hypotheses. One reason might be that the two competing hypotheses were valid, but worked against each other, which led to no effect. Another reason is that the activation of COVID-19 threats and the worry it evoked might have been too shortlived to influence later behavior. Since the pandemic was going on at the time of both studies, it was not possible to completely eliminate the COVID-19 threat in the control group, which might have prevented us from observing a difference between the two groups. These possibilities should be explored in future studies by thoroughly testing the two hypotheses using different methods. While further research is necessary, our results in light of the existing literature demonstrate that the activation of COVID-19 threats might not influence the motivation to help tackle climate change and the refugee crisis.
The present research showed that even brief exposure to COVID-19 threats could elevate anxiety levels and that individual differences play a role in this. Specifically, COVID-19 threats induced more anxiety for participants with lower levels of psychological distance from COVID-19 (Studies 1 and 2) and trait anxiety (Study 1) and higher levels of future orientation (Study 1). These were also in line with the existing literature as psychological distance, trait anxiety, and future orientation were previously theorized and empirically found to shape reaction to a social or environmental crisis (e.g., Khachatryan et al., 2013;Milfont et al., 2012;Van Lange & Huckelba, 2021). The moderating effect of trait anxiety reflected that COVID-19 might trigger anxiety even among those low in trait anxiety. Interestingly, however, this moderation in Study 1 was not replicated in Study 2. One explanation for this could be the role of anxiety in climate change versus the refugee crisis. Trait anxiety is linked to concerns about harm to self and close others. Such a concern might be more present in the case of climate change than in the refugee crisis since the latter involves helping others to reduce others' suffering with relatively less impact on oneself and close others. Trait anxiety might therefore be less relevant in the context of the refugee crisis, which might be a reason for the absence of the moderating effect; however, future studies should investigate this further. Taken together, the present findings reflect the detrimental effects of exposure to COVID-19 threats and how some individuals might be more susceptible.
The exploratory analyses in the present research also derived intriguing findings, supporting the affect-generalization hypothesis to some extent. First, we found that lower psychological distance from COVID-19 was significantly associated with higher levels of selfreported anxiety, greater environmental and humanitarian donations, pro-environmental intentions, and intentions to help refugees. While further research is needed to infer causality, these findings suggest that being psychologically closer to COVID-19 (e.g., by having direct or indirect experiences with COVID-19) might predict greater motivation to take action against other crises, such as climate and refugee crises. This points to the affect-generalization hypothesis. Moreover, in Study 1, elevated levels of anxiety after being exposed to COVID-19 threats increased environmental donations to help mitigate climate change. On the contrary to the main findings of the present research, these findings are in line with previous research which demonstrated that greater risk perception and worry about COVID-19 might be associated with climate change worry and action (Jian et al., 2020;O'Connor & Assaker, 2021;Sisco et al., 2021). Taken together, these exploratory findings provide preliminary evidence for the affect-generalization hypothesis by indicating that being close to or worrying about one risk might stimulate action against other risks, at least in the context of COVID-19 and climate change.

Strengths, limitations, and future directions
The present research had remarkable strengths. Given the existing preliminary evidence for the competing finite-pool-of-worry and affectgeneralization hypotheses, this study was novel in terms of testing these hypotheses at the same time. Moreover, as we did not rely only on climate change action but also examined the possible influence of COVID-19 on the motivation to act against the refugee crisis, our results had improved external validity. Taken together, while we did not find support for any of the competing hypotheses and further research is needed to test these hypotheses in-depth, our research was an initial step toward understanding how threats activated by one crisis could undermine or promote action against other crises.
Despite its strengths, the present research also has limitations. Manipulating COVID-19 threats in the times of COVID-19 was challenging. Since the UK, where we conducted the studies, was one of the countries that the pandemic highly affected, the general COVID-19 threat was largely present at the time of the studies. However, since we conducted successful manipulation checks and found support for our hypotheses regarding the influence of COVID-19 threats on anxiety in both studies, we could tackle this limitation to some extent. Nevertheless, different manipulations of COVID-19 threats (e.g., focusing solely on health-related or economic threats) and longer and more intense exposure to these threats might have derived different results, which can be a direction for future studies.
Another future direction could be to use a different approach to measure the motivation to help tackle climate and refugee crises, which might shed more light on how exposure to COVID-19 might influence individuals' motivation to take action for other issues. In the present research, we operationalized the motivation to tackle climate change as pro-environmental behavior and intentions and measured the motivation to help relieve the refugee crisis by assessing prosocial behavior toward refugees and intentions to help them. While this approach was backed up by measuring interest in climate and refugee crises, it could be also useful, for example, to measure individuals' risk perception of and psychological distance from climate change and the refugee crisis upon reading about the COVID-19 pandemic. This could help analyze whether cognitions regarding one crisis might be transferred to other crises and how this might influence taking action.