The Psychological Determinants of Avoiding Crowded Areas: An International and Political Investigation

Cultural and Political Context of our Research

The rapid spread of COVID-19 was first identified in 2020. By December 31 of the same year, 753 thousand daily increase in cases worldwide (World Health Organization [WHO], 2021). These cases were mainly found in the Americas, but least in the Western Pacific countries. Because of this contrast at the time of data collection, we surveyed citizens from two large cities in the Americas and two in the Western Pacific. Thus, four cities were selected (1) Brasilia, (2) New York, (3) Tokyo, and (4) Taipei.

Brazil declared the COVID-19 pandemic a national public health emergency on February 3, 2020. It quickly became one of the world’s largest centers of COVID-19 spreading (de Souza et al., 2020). Brasilia and its Federal District, under a state of emergency, had 300,393 cumulative cases of COVID-19 by March 3, 2021, even after implementing measures for infection control (Public Security Secretary of the Federal District, 2021). The U.S. also showed a peak of cases by that time with New York reporting roughly 5,599 new cases weekly by January 31^st (New York Times, 2021), after the implementation of a series of control measures since March 2020, such as declaring a state of emergency (Cheng et al., 2020). On the other side of the globe, Japan had a relatively low number of COVID-19 cases and deaths, but it presented the second worst scenario in the Western Pacific (Shimizu et al., 2021). In Tokyo, lockdown measures were not mandatory, but their measures could still be effective to control infection rates based on previous research conducted before our data collection (Yabe et al., 2020). By March 2020, 112,345 cumulative cases were registered in the Tokyo Metropolitan area (Tokyo Metropolitan Government, 2021). As regards Taiwan, this city could be expected to show one of the highest COVID-19 cases, considering its geographic relation to Mainland China (Wang et al., 2020). However, it presented only rare cases due to the rapid adoption of protective measures since January 2020, such as border control (Cheng et al., 2020; Wang et al., 2020). As a result, between February 28 and March 6, 2021, only 19 confirmed cases were registered in the country.

Previous research showed that the success of infection prevention control is likely to depend on the country's cultural background (Borg, 2014). For example, Biddlestone et al. (2020) and Huynh (2020) found that cultural differences play a role in engaging social distancing, whereas, countries with high individualism adopted this measure to lower degrees. Nonetheless, substantial attention to the influence of political views is also needed, since COVID-19 can be a highly politicized matter (Bartusevičius et al., 2021). Rising political polarization, conflicting information from the news, and government representatives can be critical factors to consider in the management of health policies in some countries like Japan, Brazil, and the U.S. (Liff, 2021; Lopes, 2021; Pew Research Center, 2017). For example, at that time, there was a political conflict between the USA and the World Health Organization, which influenced the American response to the COVID-19 pandemic (Kerr et al., 2021). In Japan, political concerns rose due to the hosting of the Olympics during the pandemic, which generated division in domestic opinions (Liff, 2021). Brazil faced increasing political instability in the last years, with discontentment and the rising of diverse political ideologies (Lopes, 2021). On the other hand, domestic political stability seems to be a key issue for the effectiveness of certain countries, like Taiwan, in their fight against COVID (Wang et al., 2020).

Theoretical Model

Data Collection and Questionnaire

The study comprised a cross-national survey with most questions being responded on 5-point Likert-type scales that measured (1) attitudes, (2) injunctive norms, (3) moral norms, (4) trust in authorities, (5) perceived behavioral control, (6) risk perception, (7) behavior intention, and (8) frequency of exposition to COVID-19 information. To measure the first seven constructs we asked participants how much they agreed with the statements described in Table 1 (1 = Strongly disagree; 5 = Totally disagree). Moreover, the frequency of COVID-19 information received was asked from 1 (Never) to 5 (Very often). One question asked how often individuals avoided crowded places when going out. This item was responded to on a seven-point scale, from 1 (Never) to 7 (Every time).

We selected cities based on health, cultural and political differences. Thus, the survey was conducted online in New York, Tokyo, Taipei, and Brasilia between December 31, 2020 and March 3, 2021. The data were collected by private survey companies (Qualtrics and Cross-Market) in New York, Tokyo, and Taipei while using a quota sampling strategy based on equal distribution of age and gender. However, in Brasilia, the researchers resorted to snowball sampling due to the Brazilian national norms of research (i.e., Resolution of the Conselho Nacional de Saúde number 196 of 1996) that do not allow research participants to be paid. The cities selected were located near to where each author lived to avoid biased conclusions over cultural, political, and language barriers.

All procedures in our study followed the ethical standards for studies involving human participants as well as the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The research was approved by the research ethics committee and all participation in this study was consented to and voluntary. The questionnaire, originally written in English, was translated into Portuguese, Mandarin Chinese, and Japanese. Then, it was back-translated. All translators were native to their respective languages and had academic backgrounds. The interested reader may read the translations and back translations in Supplementary Material 1.

For comparing groups with ANOVAs, a minimum sample size of 1096 participants was required taking a power (1 - β) of .8, four groups formulated by the four cultures involved, a low effect size of Cohen's f = .1 and α = .05. A small effect size was stipulated due to a lack of comparative studies. This sample size also satisfied the data necessary for structural equation modelling according to the number of parameters in the model.

Sample Characteristics

The sample consisted of 1196 participants from Taipei (n = 300), Tokyo (n = 300), New York (n = 313), and Brasilia (n = 283). The mean age was 44 years old (SD = 14), but no participant was under 18, and 58.9% were women. Unemployed people were 16.7% while 35.9% were working remotely. Only 30 participants declared staying under quarantine. As regards political orientation, 42% of participants declared being centrists; people not interested in politics amounted to 20.2%.

Importantly, each country considers the political spectrum in different terms. For that reason, in Brasilia we considered the division between left-wing and right-wing, in New York between republicans and liberals, and in Taipei between the Democratic Progressive Party (D.P.P.) and the Kuomintang (K.M.T.)/ the Chinese Nationalist Party, and finally in Tokyo between progressists and conservatives. Further, sociodemographic characteristics can be consulted in Table 2.

Data Analysis

We present our results in two sections: (1) structural model and multi-group analysis, (2) comparison of the mean scores between cities and political views. In the first section, we ascertained the validity of the model comprising all the samples via Structural Equation Modeling (AMOS SPSS v.26). We followed minimum (Bentler, 1992; Byrne, 2013) or superior criteria (Hu & Bentler, 1999) respectively according to these criteria: χ² with p ≤ .05, Root Mean Square Error of Approximation (RMSEA) ≤ .08 or ≤ 0.05, Comparative Fit Index (CFI) ≥ .90 or ≥ .95, NFI ≥ .90 or ≥ .95, and Tucker-Lewis Index (TLI) ≥ .90 or ≥ .95. Because our sample comprises participants from four different cities, we conducted a multi-group analysis. In the second section, we ascertained the differences in the mean scores of each component of the model across the cities and across political views. For this section, we conducted Welch's ANOVAs (Robust Test of Equality of Means) to control for any violation in the homogeneity of variance or differences in the number of participants across groups (Delacre et al., 2019).

Section 1: Structural Model and Multi-Group Analysis

The data fit the model well with χ²(10) = 90.673, p < .001, CFI = .977, NFI = .975, TLI = .898, and RMSEA = .082, 95% CI [.067, .098]. As can be seen in Figure 2, it was possible to explain around 18% of the variance in risk perception, 59% in intention, and 37% in behavior.

The data supported all hypotheses. The reported behavior of avoiding crowded places was predicted by intention (β = .30, p < .001; Hyp. 1), risk perception (β = .11, p < .001; Hyp. 2), moral norm (β = .19, p < .001; Hyp. 3), and perceived behavioral control (β = .20, p < .001; Hyp. 4). Then, risk perception was predicted positively by the frequency of information related to COVID-19 (β = .17, p < .001; Hyp. 5), negatively by trust in authorities and health policies (β = -.13, p < .001; Hyp. 6), and positively by moral norm (β = .18, p < .001; Hyp. 7) and injunctive norm (β = .17, p < .001; Hyp. 8). Moreover, intention was predicted positively by attitude (β = .43, p < .001; Hyp. 9), perceived behavioral control (β = .11, p < .001; Hyp. 10), moral norm (β = .15, p < .001; Hyp. 11), and injunctive norm (β = .24, p < .001; Hyp. 12). In explaining the motivation and behavior to avoid crowded areas, moral norms and risk perception thus contributed over and above the by default components of the theory of planned behavior. Covariates and other details of the model can be consulted in Supplementary Material 2.

For multi-group analysis, a model considering cities (as proxies for cultures) as different groups were analyzed with all parameters to be freely estimated, showed a good fit: χ²(40) = 111.494 (p < .001), CFI = .979, NFI = .969, TLI = .907 and RMSEA = .039, 95% CI [.032, .047] (see Supplementary Material 3 for scores of unconstrained estimates). A constrained model where all estimates were forced to be equal across the cities showed a worse fit with non-invariance in the estimates (Δχ² of 120.334, p < .001; ΔCFI = .024): χ² = 231.828 (p < .001), CFI = .955, NFI = .936, TLI = .893 and RMSEA = .041, 95% CI [.035, .048]. Based on Cheung and Rensvold’s (2002) criterion to decide on invariance (i.e., ΔCFI < .01), we concluded that this model did not show invariance in the structural weights by culture (ΔCFI = .024) or in structural covariants (ΔCFI = .115) and structural residuals (ΔCFI = .121). We thus conducted a step-by-step analysis of the estimates, covariances, and residuals that could be constrained and invariant without significantly harming the model fit across the groups. We were thus able to achieve a significantly better fit with χ²(82) = 179.599 (p < .001), CFI = .972, NFI = .951, TLI = .938, RMSEA (90% CI) = .032 [.025, .038] (see Supplementary Material 4 for more information on parameters before and after adjustment). The latter model accounted for 47% of the behavioral variance in Brasilia, 34% in Tokyo, 35% in New York, and 21% in Taipei. Intention variance was the greatest in Tokyo (69%), while fear was in Taipei (38%). The estimates that would be considered equal across cultures were the paths: 1) from PBC (Δχ² = 5.708, p = .127), moral norms (Δχ² = 4.268, p = .234), and attitude (Δχ² = 6.908, p = .075) to intention; 2) from injunctive norms (Δχ² = 6.920, p = .074) to risk perception; and 3)moral norms (Δχ² = 2.185, p = .535) and risk perception to behavior (Δχ² = 5.981, p = .113) (See Supplementary Material 5 for the list of all constrained parameters).

As for the estimates in the multi-group model, moral norms (p < .001) and risk perception (p < .001) were constrained to predict behavior, while intention and perceived behavioral control varied significantly across cities. Indeed, as can be seen in Table 3, intention was the strongest predictor of behavior in New York, followed by Tokyo, but in Brasília and Taipei, Perceived Behavioral Control was. Intention was predicted with the same effect across cities by moral norms, attitude, and perceived behavioral control, while injunctive norms varied significantly with a stronger effect in Tokyo, and New York, respectively.

As regards risk perception, only injunctive norms were predicted with the same effect across cities. On the contrary, the frequency of COVID information, trust, and moral norms varied significantly. Indeed, the strongest predictors of risk perception were COVID-19 information in Taipei, trust in Brasilia, moral norms in New York and injunctive norms in Tokyo. There were nonsignificant results only in the following paths: from perceived behavioral control to behavior in New York, from Intention to behavior in Taipei, and from trust and information to risk perception in Tokyo. Explained variance in Behavior was stronger in Brasilia (R² = .466), New York (R² = .345), Tokyo (R² = .337), and Taipei (R² = .208), respectively. Explained variance of risk perception and intentions can be consulted in Table 4.

To sum up, in the first section of our results, we could find that several regression weights were similar across the cities, while others were not if we consider preserving the data fit and model`s explanation.

Section 2: Comparison of the Mean Scores Across Cultures and Political Views

Although we could analyze each estimate in the model, comparing the mean scores across cities gave us additional information. Indeed, as can be seen in Table 5, there were significant differences in all measures across cities as indicated by one-way Welch ANOVAs (all ps < .001). Means, standard deviations, F tests, and partial eta squared values can be consulted in Table 5.

Firstly, cities showed a moderate to strong effect on intention ( ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .065) and behavior (η² = .107). For intention, Brasilia scored the highest followed by New York, Tokyo and Taipei. As for reported Behavior, Taipei presented the lowest score, followed by Tokyo, New York and Brasilia. Moreover, the differences in the cities also showed a medium to high effect size on moral norms ( ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .108), attitude ( ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .062), PBC ( ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .087), and low to medium in injunctive norms ( ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .039). Brasilia and New York presented the highest moral norms scores. Brasilia also presented the highest score for injunctive norms, followed by Tokyo and New York.

Brasilia continued to present the highest score for attitude, followed by New York, Tokyo, and Taipei. Perceived Behavioral Control in Tokyo was relatively lower than in Taipei, New York and Brasilia. The differences across cities also showed a moderate effect on the frequency of COVID-19 information (η² = .076), a moderate to strong effect on risk perception (η² = .053), and a strong effect on trust in authorities (η² = .289). The frequency of exposed information on COVID-19 was much higher in Brasilia than in other cities. Trust in authorities was the highest in Taipei, but not much higher than the midpoint of the scale, and the lowest in Brasilia. On the contrary, Brasilia had the highest score in risk perception, followed by New York. Tokyo and Taipei remained relatively equal in their scores on risk perception.

The latter results show that, although components can have a significant effect on influencing intention and behavior, as seen in the first section of our findings, all their mean scores varied substantially across the cities, showing how culture can affect avoiding crowded places and its determinants.

Next, we present the differences in political orientations in each city. As can be seen in Table 6, Taipei showed no differences in the measures across the political dimensions, except for moral norms, and trust in authorities. For moral norms, the greatest difference was between centrists and citizens not interested in politics (Cohen's d = .502). For trust in authorities, the greatest difference was between D.P.P. partisans compared to K.M.T. partisans (d = .735) and those not interested in politics (d = .657).

As reported in Table 7, participants from Tokyo showed significant political differences in six measures, with the largest effect size being on intention and trust. No significant difference was found in behavior, but people who are not interested in politics showed significantly less intention to engage in this behavior than Conservatives, Progressists, and Centrists.

On average individuals in all cities agreed with having avoided crowded areas and producing efforts for that. Conservatives, despite scoring high in injunctive norms, still presented significantly lower scores compared to Progressists (d = .646). Regarding trust in the health policies, these differences were seemingly even higher between the same groups with Conservatives reporting more trust than Progressists (d = .813). The differences between Conservatives and Centrists were also relatively strong (d = .70).

The political differences in New York (see Table 8) were significant in risk perception, intention, behavior, frequency of information, attitude and trust in authorities, with small to medium effect sizes. In behavior, a large difference was found between Liberals and Republicans (d = .491) and Liberals versus citizens not interested in politics (d = .473) respectively. In intention, the largest difference was between Liberals and Republicans (d = .541), while risk perception also presented a relatively large difference between Liberals and Republicans (d = .741). Finally, following the same tendency, trust was higher among Republicans than Liberals (d = .650).

Results for Brasilia can be seen in Table 9. This city showed the largest differences in the scores by political positions with all of them, but PBC and frequency of COVID-19 information, were significant. The most evident differences concerned trust in authorities with it being higher among Right-wing partisans than both Left-Wingers (d = 2.279), and centrists (d = 1.143). Moral norms presented the second greatest difference followed by intention and behavior. Specifically, Right-wing partisans reported fewer feelings of moral obligation than Left-wingers (d = 1.285), centrists (d = .824) and those not interested in politics (d = .777). Right-wingers reported less intention behavior of avoiding crowded places than the other groups, with Cohen's ds varying from 0.679 to 1.218. Right-wingers also perceived less risk than Left-Wingers (d = 1.043).

In sum, our findings indicate that both the behavior of avoiding crowded places in large cities and its determinants can also be influenced by the political views of cities facing the COVID-19 pandemic. Thus, it reveals that the components of the proposed model can vary largely (e.g., in Brasilia), or low (e.g., in Taipei) across political parties depending on the target city or culture of study.

Behavior, Intention, and Attitude

The model`s explained variance of behavior and intention can be considered high compared to previous research (McEachan et al., 2011). Indeed, intention had the most explained variance in behavior, except in Taipei, where authoritiesenforced strictly controlled public health policies (Wang et al., 2020). Furthermore, risk perception and moral norms were relevant determinants to filling the intention-behavior gap in all cities, since behavior is not only directly explained by people`s intention to avoid crowded places. These findings complement other studies focusing on explaining this gap, in the pandemic, by showing that risk perception and moral norms can contribute to explaining behavior of avoiding crowded places along with intention and its determinants (Gibson et al., 2021). This leads our model to be in accordance with the proposals of the generation of extending models of the theory of planned behavior (e.g., TPB), with respect to theoretical coherence and parsimony (Conner & Armitage, 1998; Neto et al., 2020), which follows the critique that the theory of planned behavior considers few behavioral predictors in its original model (Hagger, 2010).

Moderate to large differences in mean scores of behavior and intention across cities hint at the contextual and cultural impact on adherence to protective actions. By looking at the political views, we see significant differences between New York, Brasilia, and Tokyo. Thus, highly politicized health crises can lead people to avoid crowded areas (Cavazza et al., 2021). That is less likely in Taipei, presumably because politicians who self-identify as moderates have increased popularity among voters. Indeed, data from the Taiwan National Security Survey showed that vote choice tends to respect a normal distribution in the political spectrum (Wang, 2019). Thus, policymakers might consider differences in political and cultural views in health public responses in polarized versus non-polarized contexts. Moreover, reducing this polarization and understanding the social psychological processes of communities divided by it is needed (Bartusevičius et al., 2021).

Attitudes, moral and injunctive norms, perceived behavioral control, and risk perception components significantly predicted behavior and/or intention in every city. Similar findings can be seen in the literature (Fischer & Karl, 2022; Pfattheicher et al., 2020; Shanka & Gebremariam Kotecho, 2023). The fact that attitude was the best predictor of intention hints at the importance of attitudinal change through health policies and communication that highlight the positive outcomes of avoiding crowded places (Fujii, 2003). Furthermore, differences in attitudes had a moderate effect size across cities. Attitudes were also significantly different among political parties in Brasilia, New York, and Tokyo, with the three cities mostly polarized between the opposing political parties on how they view avoiding crowded places. Brasilia showed the largest difference in this respect. In summary, highly polarized settings can thus impact attitudes, which might be a product of how political leaders frame and evaluate certain beliefs of their political partisans (Calvillo et al., 2020).

Moral and Social Norms

Moral and social norms explained intention, behavior, and risk perception substantially. In this sense, targeting feelings of moral obligation to the public by raising awareness about people in need, while emphasizing one's ability to do it and indicating their responsibility to become involved can be significant in communications (Schwartz, 1977; Van Bavel et al., 2020). Also, creating means for individuals to be aware of the expectations of their referent people to avoid crowded places is also important in determining intention (Cialdini & Goldstein, 2004; Young & Goldstein, 2021). Moreover, as seen in past literature and our results, these social and moral-based norms can also amplify the perception of risk, thus shaping public response and adherence to measures (Douglas & Wildavsky, 1983; Kahan & Braman, 2003; Kasperson et al., 1988).

Moral norms showed invariance in behavior and intention explanations across countries, in contrast with social norms, which varied significantly across cultures. Thus, the findings show that the effect of moral norms on avoiding crowded places is independent of the effect of social norms, presumably because it is more stable when shaping behavior (Schwartz, 1977). Nevertheless, moral norms presented differences in their mean scores across all cities and political parties in Japan, Taiwan, and Brazil. Thus, cultural and political norms can affect personal feelings of obligation to avoid crowded areas. Furthermore, social norms were significantly different across political parties, especially in Tokyo and Brasilia, but also across the other cities.

These patterns offer evidence that the social-political norms of cities can impact compliance to authority demands. Indeed, previous studies showed that citizens are influenced by opinions from politicians or people closest to them (Cialdini & Trost, 1998). Furthermore, during the COVID-19 pandemic, partisan identification was found to influence social distancing through the control of social norms (Fieldhouse & Cutts, 2021).

Perceived Behavioral Control

The effect of perceived behavioral control on intention was invariant across cities. However, it can explain behavior substantially, especially in Brasilia, Tokyo, and Taipei, which points to its importance in reducing the perceived costs and increasing the perceived controllability of avoiding crowded places in these cities for actual behavioral change. In terms of mean scores, perceived behavioral control had significant differences across cities, but not across political parties. One reason for the latter finding can be attributed to urban planning, which could have facilitated social distancing in crowded areas during the pandemic (e.g., guiding the flow of pedestrians to other routes to alleviate the congestion of citizens in a certain area; Hamidi et al., 2020). This is especially visible in the densest city in the study (i.e., Taipei), which scored the lowest in perceived behavioral control.

Other factors that may be relevant to consider are how the pandemic may have established work and economic constraints in certain cities, which makes social distancing be perceived as difficult for individuals to put into practice (Aschwanden et al., 2021). In this sense, Yabe et al. (2020), for example, found a significant and negative correlation of r = -.696 between taxable income per household in Japan and the number of social contacts during the pandemic, where lower-income households may not perceive flexibility in restricting their mobility in daily lives. Also, in Brazil, the COVID-19 pandemic produced a substantial impact on informal sectors and services, which require activities in dense areas (Ferreira dos Santos et al., 2020).

Risk Perception

Risk perception remained invariant in predicting behavior across cities, with the frequency of COVID-19 information, trust in authorities, and moral and social norms being significant predictors. Risk perception was better explained and had the highest score in New York and Brasilia, which is consistent with the fact that these cities had a higher number of infections compared to Tokyo and Taipei. In this sense, high-risk perceptions may be a product of close contact with the urgent reality and can be amplified by a lack of trust, exposition of risk information, social norms, and feelings of moral obligation to prevent risks (Cvetkovich, 2013; Douglas & Wildavsky, 1983; Kahan & Braman, 2003; Tversky & Kahneman, 1973). This can be taken as evidence that the effect of risk perception in pandemics is not only cognitively formulated by the dynamics of real risk estimations but by a series of social factors (Slovic & Peters, 2006).

Trust in authorities to manage the pandemic was the factor that varied the most across cities and political parties. The smallest difference in trust across political parties and the highest level of trust was found in Taipei, which is consistent with both its health management scenario (Wang et al., 2020) and its internal political stability compared to the other cities (Wang, 2019). On the other hand, the large difference in trust in authorities across the cities may be due to their COVID-19 situation and how effectively the policies may have been conducted. The highest differences across political parties and the highest score of distrust were mostly found in Brasilia, the capital city of a country with high levels of inequality and political instability in the last decade (Barberia et al., 2021; Lopes, 2021).

Moreover, the effect of COVID-19 information frequency on risk perception differed across cities. In this sense, exposure to COVID-19 information regulates people`s perception of infection risk through different sources such as the social/mass media (Stevens et al., 2021). Information frequency also differed significantly across cities in its mean scores, being the highest in Brasilia. Moreover, in New York and Brasilia, there were differences across parties in this component, suggesting potential political boundaries for the spread of key COVID-19 information (Fraser & Aldrich, 2021).

Limitations and Future Agenda

This research presents several limitations. First, we could not conduct a multi-group analysis by political positions accounting for each city, since it would require a much larger sample divided into 16 groups (i.e., four political views for each of the four cities). Second, although some effects in the model can be considered equal across the cultures, replication of data in other cultures and cities is needed before considering estimates as fairly universal. Third, our study is correlational. Besides precluding causal interpretations, our study does not capture nuanced information about each location, which would allow for more accurate interpretations. Conducting a multi-method study is thus important. Fourth, data collection was conducted differently in Brasilia with the use of snowball sampling due to the national regulations, where participants are not allowed to be paid. This may have compromised the equivalence among samples. Finally, despite using diverse samples, they are not representative of the studied cities, much less of the countries and their cultures.

Conclusions

By analyzing the social and cognitive factors of avoiding crowded places during the pandemic, this study offers several significant academic contributions. Also, by verifying the validity of a model based on the theory of planned behavior, risk perception, and moral norms to explain the behavior of avoiding crowded places during the COVID-19 pandemic, we showed that avoiding crowded places can be a planned behavior shaped by social-cognitive determinants. Moreover, the cross-cultural evidence of the study shows that the relationship between people and places can significantly change according to the social context or the political position with which people identify themselves. This provides an additional contribution to the discussion of target-focused policies and to the academic community in building a comprehensive model. Thus, after critical debate and thinking over the findings of this research, and while considering its limitations, these results can be useful to new proposals in health policymaking designed according to the social-political need of each culture (Uzzell, 2015).