Tourists’ psychological responses to terrorism: Framework analysis of fear-arousing walking interviews from the city centre of Munich, Germany

Even though research on tourists’ psychological responses to terrorism is substantial, it largely ignores on-site cognitive assessments and emotional responses. As a result, little is known about how location-related factors influence these responses. The present study intended to address these gaps in the context of urban tourism. Therefore, it aimed to evaluate on-site tourists’ psychological responses to terrorism in a city destination, while the emphasis was on identifying location-related factors that influence them. The study was based on framework analysis of fear-arousing walking interviews with foreign tourists (n = 24) in the city centre of Munich, Germany. Many participants self-reported the absence of fear of terrorism. However, all participants appraised terrorist attacks as possible within the city centre. The perceived probability varied among participants and visited sites. The differences can be explained by 12 identified location-related factors, of which the number of people and antiterrorism measures were key. Other factors, such as the way participants comprehended terrorism, also influenced their psychological responses to it. Theoretical and managerial implications are discussed.


Introduction
Europe has witnessed increased terrorist activity within the last decade (National Consortium for the Study of Terrorism and Responses to Terrorism, 2020;Nesser, 2018). Such threats have had detrimental effects on European tourism (Corbet et al., 2019;Schmude et al., 2020;Vanneste et al., 2017). These effects are in line with other research on tourist responses to terrorism (Krajňák, 2020) and further security threats (e.g., Ivanov et al., 2017;Neumayer, 2004). Nevertheless, a closer look into tourists' psychological responses is required due to the unique nature of the recent terrorist activities. They can be characterised by higher complexities and by a preference for soft targets, like crowds of people in public areas (Nesser et al., 2016). Additionally, the European attacks were mainly perpetrated in cities (e.g., Paris, Nice, Istanbul, Manchester, Brussels, Saint Petersburg, Barcelona, Berlin, and London). Therefore, the recent wave also provides us with an opportunity to examine gaps in research on tourists' risk perceptions and fears of terrorism.
To date, studies researching tourists' psychological responses to terrorism have focused primarily on future travel intentions, including willingness to travel and destination choices (e.g., Adeloye & Brown, 2018;Isaac, 2021;Larsen et al., 2011;Sönmez & Graefe, 1998a). As a result, on-site tourists' psychological responses have been addressed by scholars less frequently. Yet, insight into the minds of tourists while they are visiting destinations is particularly needed, as reflecting on their experiences may be beneficial for proactive and reactive destination and crisis management practices (Tarlow, 2014). Indeed, residents' fears of crime or terrorism are often considered crucial for addressing such threats, while onsite tourists' views are commonly excluded from crises prevention and mitigation even in tourism destinations (Mawby, 2000).
Studying these on-site psychological responses allows for exploration of the influence of destination attributes. According to Karl & Schmude (2017), an understanding of external factors is important for producing a complete picture of future travel intentions affected by tourists' risk perceptions. They add, however, that most previous research has concentrated only on internal factors (i.e., tourists' attributes). Relatedly, research on terrorism-related risk perceptions and associated concepts has predominantly employed questionnaire surveys (e.g., Fuchs et al., 2013;Isaac, 2021;Morakabati & Kapuściński, 2016;Wolff & Larsen, 2017). These studies provide important findings about the effects of terrorism on tourists' risk perceptions and worries, future travel intentions, and the role of internal factors or even information sources. However, a more local view is needed to better understand the role of destination attributes, which this study approaches from a micro-scale perspective. Such detailed picture can be obtained through qualitative research. Surprisingly, only a small number of articles have relied on interviews and other qualitative data (Adeloye & Brown, 2018;Ferreira et al., 2019;Uriely et al., 2007), possibly due to difficulties in interviewing tourists about sensitive topics such as terrorism (Adeloye et al., 2020).
Since the recent terrorist activities in Europe can be characterised by their urban nature, studies on tourists' psychological responses to terrorism must be conducted in cities. However, these spaces have been largely omitted from this strand of literature, despite terrorism being predominantly an urban threat (Marineau et al., 2020). Additionally, the underrepresentation of the local scale is common not only in research on psychological responses, but also in studies examining the effects of terrorism on tourism generally (Krajňák, 2020).
Focusing on cities may be helpful in addressing knowledge gaps concerning the role of antiterrorism measures. Cities have increasingly sought to defend themselves from terrorism by integrating principles of urban resilience and defensive urban planning (Coaffee, 2009). Consequently, access control measures, closed-circuit television (CCTVs), police officers, and other visible security measures are becoming ever more common features of urban landscapes (Dalgaard-Nielsen et al., 2016). Although such measures are traditionally considered valuable and reassuring, they have also been called into question multiple times due to their presumed fear-inducing effects (Aly & Green, 2010;Grosskopf, 2006). Importantly, previous research provides limited knowledge about their influence on tourists' risk perceptions and fears.
The intention of this study was to close such gaps. Therefore, its main aim was to evaluate on-site tourists' psychological responses to urban terrorism, i.e., risk perceptions and fears of terrorism in a city destination. As a case study, the city centre of Munich, Germany, was selected. The emphasis within the study's main aim was on identifying location-related factors, including visible security measures, that influence the psychological responses. To explore the issue on the micro-scale, the study adopted a qualitative approach. Since tourists' responses to terrorism are assumed to differ vastly between domestic and international markets, this study considered only the perspective of international tourists who are described by many as more terrorism-sensitive (Greenbaum & Hultquist, 2006;Schmude et al., 2020;Vanneste et al., 2017;Wolff & Larsen, 2014).

Literature review 2.1.Terrorism, tourists' risk perceptions and related concepts
The term risk perception should be traditionally understood as a cognitive evaluation considering the characteristics of a risk, such as probability and severity (Slovic, 1987;Wolff et al., 2019). According to Loewenstein et al. (2001), adherence to the consequentialist cognitive perspective is insufficient when explaining risk-related decisions and behaviours. Their risk-as-feelings hypothesis postulates that the process of risk-related behaviour is also facilitated by emotions, leading to anomalies in normative decisions. The relationship between emotional reactions to and cognitive evaluations of risky situations works in the opposite direction as well, since cognitive appraisals give rise to emotional responses such as worries and fears. Therefore, this article concerns both cognitive (risk perceptions) and emotional (e.g., fears) psychological responses to terrorism.
Terrorism differs from other events threatening tourism by purposefully attempting to arouse negative psychological responses, most notably fear, and influence audiences' and sometimes tourists' behaviours 1 . Consequently, terrorism can be approached as a fear-arousing communication. The effects of such communications are addressed by fear appeal theories (e.g., Rogers 1983;Witte 1992). Since fear appeals are persuasive messages seeking to arouse fear by emphasising the potential danger to individuals of not adopting the messages' recommendations (Tannenbaum et al., 2015), these theories assume that under a certain degree of fear or threat, and when other conditions are met, people will accept the recommendations and adjust their attitudes and behaviours. A high perceived threat of terrorism, therefore, may result in travel avoidance by potential tourists, or acceptance of other recommended coping mechanisms by actual tourists (e.g., implementing security measures and adjusting micro-scale behaviours).
Media represent one of the most important channels possibly transmitting such terrorism-related messages. Kapuściński & Richards (2016) found news amplifying terrorism risk to cause more tourists' worries than news attenuating the risk. This news effect was, however, stronger for events of political instability involving (violent) protests and clashes. Indeed, resilience to terrorism-related news may exist as media exaggeration can lead to the denial of such messages (Adeloye & Brown, 2018). Conversely, it has been proven that higher interest in and attention to terrorism-related news increase international tourists' risk perceptions and deepen their involvement in trip planning (Seabra et al., 2014). Furthermore, exposure to news videos about terrorism in a foreign destination dampens people's willingness to travel there (Hoffman & Shelby, 2017). Governments' travel advices related to terrorism as well seem to adversely affect tourists' decisions (Buigut et al., 2017).
According to several studies, terrorism belongs among the main threats affecting tourists ' risk perceptions, feelings, and decisions (e.g., Bowen et al., 2014;Ekeberg et al., 2014;Gray & Wilson, 2009;Karl et al., 2020;Sönmez & Graefe, 1998b). Apart from purposefully induced fear, the substantial influence of terrorism can be explained by cognitive biases accompanying decision-making processes. For instance, people tend to overestimate negative prospects with low probabilities but high consequences, like terrorism (Kahneman & Tversky, 1979;Sunstein, 2003). Additionally, when evaluating risks, people are influenced not only by quantitative, but also qualitative characteristics. Since terrorism is a man-made, fatal, uncontrollable, nonvoluntary threat, its strong effect is expected (Kapuściński & Richards, 2016;Schusterschitz et al., 2010). Therefore, the psychological impacts of largescale terrorist incidents may be felt worldwide (Aimable & Rosselló, 2009;McKercher & Hui, 2004). However, global declines in tourist arrivals due to terrorism are rather rare; negative spill-overs to countries in the same region are more frequent (Enders et al., 1992;Neumayer, 2004;Neumayer & Plümper, 2016).
There are also studies revealing relatively low risk perceptions and worries about terrorism among tourists (Brun et al., 2011;Larsen et al., 2009;Rittichainuwat & Chakraborty, 2009;Wolff & Larsen, 2014). The psychological impacts of terrorism are possibly affected by timing of terrorist incidents, with tourists' worries being highest, and intentions to travel lowest, after incidents (Brun et al., 2011;McKercher & Hui, 2004). Contrary to these findings, a terrorist incident does not always change travel desires or overall risk judgements (Larsen et al., 2011), and under some conditions it may even lower tourists' worries and risk perceptions (Wolff & Larsen, 2014). Another study found it is rather likelihood of future attacks than their past occurrence affecting the intent to travel (Walters et al., 2019). Same researchers also discovered the influence of terrorist threats on other aspects of tourists' behaviours such as preference of established hotels over local options, and of group tour formats over independent travel.
Furthermore, tourists' worries about terrorism can be mitigated by the strength of the benefits sought. This mitigating effect is strongest for nature and adventure holidays and weakest for seaside holidays (Morakabati & Kapuściński, 2016), suggesting the notable influence of destination attributes. Additionally, the effect that destination or holiday type has on tourists' risk perceptions seems to be related to previous terrorist incidents. For instance, after terrorist attacks in Madrid and London, taking trips to European cities was perceived as riskier than before the incidents, yet other destination or holiday types remained relatively unaffected, except destinations often associated with terrorism (Larsen et al., 2011). Indeed, destinations frequently affected by terrorism arouse higher risk perceptions and worries (Fischhoff et al., 2004;Wolff & Larsen, 2017). In terrorism-hit destinations, it can be the qualitative characteristics of the incidents, including targets of attacks, perpetrators, location, reaction of security services, possibility of future attacks, and even residents' attitudes, that have a substantial effect (Kapuściński & Richards, 2016). These can be reflected in tourists' psychological responses even on a micro scale as sites with various physical and social aspects within any local destination can be threatened differently by terrorism in tourist's eyes (Chen & Noriega, 2004). In fact, studies addressing the effects of crime on tourists' behaviours recognise that signs of small-scale disorders in public areas, such as rubbish, noise, vandalism, and the presence of intimidating persons, negatively influence tourists' psychological responses to crime (Akkuş & Arslan, 2021).
In general though, actual tourists worry less than potential tourists (Larsen et al., 2009;Wolff et al., 2019). Thus, on-site tourists' risk perceptions and worries remain largely unaffected by terrorism (Uriely et al., 2007;Vanneste et al., 2017;Wolff & Larsen, 2014). This may be a result of various rationalisations actual tourists employ. Uriely et al. (2007) discovered the existence of inward-oriented rationalisations stemming mainly from a well-known cognitive bias where people assume chances of another terrorist attack in the same location to be slim (Wolff & Larsen, 2014). Lack of targets and heightened security presence after an attack represent other examples of this rationalisation type. Actual tourists also project outward-oriented rationalisations, as they sometimes believe a destination is safer than their home areas or the world (Fuchs et al., 2013;Uriely et al., 2007). This is also related to fatalist positions held by some tourists, who state that terrorism can happen anywhere, since it is unpredictable and unpreventable (Ferreira et al., 2019;Rittichainuwat & Chakraborty, 2009;Uriely et al., 2007).

Urban tourism and terrorism
Terrorist incidents related to tourism occur mainly in cities 2 (Pizam & Smith, 2000). Research has already established that their tourism demand is also negatively affected by terrorism. This effect is time-limited (Boger et al., 2005;Greenbaum & Hultquist, 2006;Schmude et al., 2020;Vanneste et al., 2017) and differs among city destinations, for instance, between large and small cities (Greenbaum & Hultquist, 2006) or adult and family destinations (Boger et al., 2005). It further varies within cities, for instance, among accommodation types (Greenbaum & Hultquist, 2006). It also diverges among source markets, for instance, between domestic and international tourists (Schmude et al., 2020;Vanneste et al., 2017), intra-and inter-regional tourists, or business and leisure tourists (Corbet et al., 2019), and may spill over to other cities (Vanneste et al., 2017) or lead to substitution of neighbouring ones (Greenbaum & Hultquist, 2006). Safety and security concerns are also determinants affecting local tourist behaviour within cities (Paül i Agustí, 2020). Regarding terrorism, however, existing research indicates most tourists do not adjust their local behaviours, since their fears remain low. Yet, some are reluctant to visit crowded places and events, sites of previous attacks and those without access control, use public transportation, or explore cities during the night-time (Ferreira et al., 2019;Vanneste et al., 2017).

Antiterrorism measures and tourism
According to Brun et al. (2011) and Larsen et al. (2011), the majority of tourists deem world and popular destinations to be more dangerous as a consequence of the "War on Terror." Hoffman & Shelby (2017), though, found counterterrorism does not necessarily increase feelings of insecurity when communicated appropriately, and makes people more willing to travel to terrorism-affected destinations. The present article is not concerned about the effects of offensive counterterrorism, but defensive antiterrorism measures.
In particular, visible security measures such as access control measures, CCTVs and police officers are under scrutiny. In urban settings, these measures have received a lot of academic attention (Coaffee, 2009;Dalgaard-Nielsen et al., 2016;Grosskopf, 2006). Just as with counterterrorism, disagreements over the ability of visible security measures to reduce terrorism's psychological effects occur among scholars. Some claim visible security measures magnify people's fears instead of providing reassurance (Grosskopf, 2006). According to Dalgaard-Nielsen et al. (2016), this backfiring effect is most frequently assumed rather than tested, and their research confirms the traditional assumption about security measures reinforcing feelings of safety.
Tourism studies have been mostly interested in tourists' reactions to police officers and other security personnel, usually revealing positive attitudes and feelings (Adeloye & Brown, 2018;Chen & Noriega, 2004;Isaac, 2021;Vanneste et al., 2017). In cruise tourism, however, the increased security presence appears generally unpopular among passengers, while the employment of airline-style full-body scanners would be a welcome measure (Bowen et al., 2014). Based on global data, Asongu et al. (2019) did not find significant effects of security personnel on tourist arrivals, though the positive sign was apparent. Ferreira et al. (2019) and Survila et al. (2017) further show the ambivalence of tourists' responses to antiterrorism measures. Most of their respondents agreed with the measures, yet some reported discomfort, believing that soldiers on the streets cause a sense of insecurity and visible security measures disturb rest and relaxation.

Study area
The city of Munich, Germany, was chosen as a case study for the following reasons. Munich has a substantial history of terrorism. Between 1970 and 2019, a total of 52 terrorist incidents occurred. Three of those can be considered high-profile (involving 10 or more fatalities) incidents (National Consortium for the Study of Terrorism and Responses to Terrorism, 2020). These include the kidnapping and killing of Israeli athletes during the 1972 Summer Olympics, the Oktoberfest bombing in 1980, and the latest incident, a 2016 mass shooting in the vicinity of the local shopping mall. Although the last two incidents were perpetrated by neo-Nazi and anti-immigrant extremists 3 , Munich is also one of the German cities threatened by the third phase of jihadi terrorism in Europe (defined by Nesser et al., 2016). During the period of data collection (June 2019), Germany was still on a heightened terror alert level since the Berlin Christmas market killings in 2016 (Federal Ministry of the Interior and Community, 2019).
Munich is among top three destinations in German urban tourism, together with Berlin and Hamburg. Among these cities, Munich has the highest ratio of overnight stays per inhabitants (Namberger et al., 2019). In absolute figures, a total of 18.3 million overnight stays and 8.8 million tourist arrivals were recorded for Munich in 2019 (München Tourismus, 2020). With 3.9 million arrivals from abroad, Munich is the German city that receives the highest proportion of foreign tourists. Americans, followed by Italians and British, are the top three groups of foreign tourists (City of Munich, 2020). Thus, Munich is a popular international tourist destination, making it an appropriate setting in which to study foreign tourists' psychological responses to terrorism.
The present study, however, did not consider the entire city, but focused exclusively on urban tourism precinct of the city centre. City centres, with their spatial concentrations of tourists and residents, as well as "their lack of access control, inbuilt permeability and encouragement of public milieu," represent soft targets that are under increased threat from terrorism since 9/11 (Coaffee, 2009, p. 84). Moreover, apart from being a typically crowded public space, Munich's city centre provided an ideal study setting as numerous tourist attractions and infrastructures are located close to each other, making data collection feasible.

Data collection
Data collection for the present study built on the technique of walking interviews, which are ideal "for exploring issues around people's relationship with space" (Jones et al., 2008, p. 2). They can connect what participants say with where they say it. The positive effect of the environment on participants' speeches during walking interviews was empirically tested by Evans & Jones (2011). During walking interviews, people tend to structure their stories more geographically (focused on specific buildings, environmental features, and their use) and contextually (related to places visited during interviews) than during sedentary interviews.
Walking interviews have been previously utilised in urban tourism research. Commented walks were one of the types employed (Popp, 2012(Popp, , 2018. This method was developed by Thibaud (2001) and our study followed some of its principles ( Figure 1). However, our walking interviews deviated from Popp's studies and principles of commented walks in a few respects. Due to the selected approach and phenomenon under study (i.e., terrorism risk), our walking interviews were more structured and interventional, though still substantially organised by the setting and situations.

Figure 1. Research design phases
The approach of fear-arousing communication, otherwise known as fear appeal, was employed in our walking interviews. The rationale for this approach stems from the understanding of terrorism such a fear-arousing communication (Krajňák, 2020). The approach also aids in overcoming some practical issues. First, it helps to focus the interviews solely on terrorism. Second, it reduces differences among participants' knowledge of terrorism in Munich. Third, it makes an advantage rather than a disadvantage out of biased responses while talking about terrorism.
Ahead of the interview, participants were presented with the fear message, consisting of infographics and newspaper articles about terrorism (worldwide, in Europe, and in Munich), and with visible security measures as general coping strategies. In this phase, in addition to the fear manipulation, participants were assured of the anonymity, confidentiality, and the use of interviews only for research purposes. The focus of interviews on terrorism was introduced, and participants were encouraged to speak freely, to interrupt the interviewer with anything they deemed to be important, and told that there were no right or wrong answers. Participants were also asked for permission to record interviews (Legard et al., 2003). Written informed consent was obtained from all participants.
During the main phase, interviews were recorded using a GoPro camera and clip-on microphone. The video recording turned out to be beneficial, especially for the later analysis, as it captured changing circumstances of interviews. That is, the remote return to the field helped the researchers to assess the effect of unexpected situations and objects on participants' responses. However, it was primarily the use of the predefined route that made the connection of participants' commentaries with locations possible. The route was designed by the researchers along several sites ( Figure 2). Each of them served as a speech object. They included tourist attractions, transportation hubs, and fortified enclaves; all may be considered potential terrorist targets. Participants were given a map with the route for easier orientation and remembrance. On average, the walks lasted 48 (SD=11) minutes. Since conducting walking interviews requires an interviewer to film, walk, talk at the same time, and their changing circumstances open possibilities for unexpected situations, the single interviewer-data collection method was used. By giving the responsibility for conducting interviews to one interviewer (TK), the consistency of data collection was assured (Rosenblatt, 2012). Interviews were collected in accordance with the recommendations provided by the ethics committee of the institution where TK is affiliated.
The combination of the predefined route and changing circumstances resulted in semi-structured indepth interviews. The order of the interviews was mainly determined by the speech objects. A brief description of each of them was provided by the interviewer. After that, questions related to the aim of the present article were asked; similar questions were repeated every time for each site. Those concerned participants' on-site fears and appraisals of terrorism risk, factors influencing these psychological responses, and the role of visible security measures. Whenever participants reported security deficiencies related to terrorism, questions on personal coping strategies followed.
In the phase after the interview, participants were asked to complete a standardised questionnaire. It primarily included questions on personal details. Upon completing the questionnaire, participants were debriefed and given the information about the purpose of the fear-arousing communication employed, which should be separated from their experience of the city.

Figure 2. The route of walking interviews in Munich's city centre
Interviews were undertaken in June 2019. They were scheduled during the daytime and on all days of the week. Because we used the same route for all interviews, the search for diversity was based on two of three variables determined by Thibaud (2013): a previously discussed variety of circumstances; and a variety of points of view. A non-probability sampling technique was used, and a homogenous purposive sample was selected, since we were interested in the psychological responses of foreign tourists only (Ritchie et al., 2003). However, the variety of points of view was assured by recruiting participants, inter alia, from various world regions and countries. Participants were individually contacted and collectively notified using various tourist websites such as CouchSurfing, Airbnb, and specialised Facebook pages. Although the online recruitment method imposes some restrictions on data collection, it was a viable first-choice option for this study due to the sensitive topic of terrorism, tourists as a hard-to-reach population and intrusiveness of walking interviews. Additionally, research shows that online recruitment may lead to increased diversity and inclusion of marginalised groups (Upadhyay & Lipkovich, 2020). In total, we recruited and interviewed 24 participants; data collection ended once data saturation was achieved (Ritchie et al., 2003). The researchers were in consensus to stop recruiting additional participants. Not just confirmatory evidence but also negative cases were sought after to saturate data and achieve rigor (Morse, 2015).

Data analysis
Framework analysis (FA) was used to analyse the collected fear-arousing walking interviews (FAWIs). This method of analysis was ideal for our study, since it is suitable for a combined approach of deductive-inductive research in which some issues to be explored are pre-selected based on previous literature, theories or research questions, but other issues emerge purely from collected data (Gale et al., 2013). Moreover, this method, which was initially developed for applied research, offers a pragmatic approach to data analysis (Parkinson et al., 2016). FA consists of the following five phases (Ritchie & Spencer, 1994).

Familiarisation
The purpose of this phase was to get familiar with the depth and breadth of our interviews. It included three steps, the first of which started during data collection, as notes were made immediately after the interviews. Second, an initial check of the recorded interviews was conducted to ensure all parts were audible for verbatim transcription, and the notes taken after interviews were amended. Third, initial ideas about the data were marked while the transcripts were reviewed.

Identifying a thematic framework
Together with taking initial ideas, open coding was employed to aid with the development of an initial thematic framework (i.e., coding frame). One member of the research team who did not have prior knowledge of the topic of interest (JV) coded two randomly selected transcripts with a highly inductive approach to supplement and control the open coding process of the main coder (TK). This preliminary intercoding process helped with removing any possible preconceptions of the main coder and identifying emergent issues raised by participants. Afterwards, we met to discuss the differences; the initial thematic framework was developed. Each code was identified with a brief description and two to three quotations. The framework was piloted on a randomly selected transcript, showing a relatively strong fit. The model transcript and coding guidelines aimed to standardise the following intercoding process from a formal perspective.
The intercoder reliability of our thematic framework was evaluated in two rounds. TK and JV independently intercoded the same two randomly selected transcripts. Data units within transcripts to be coded were not prespecified in this round, as we did not want to constrain the coding process with any predetermined segmentation. This so-called ad hoc strategy creates complexities disallowing intercoder reliability calculations (O'Connor & Joffe, 2020). Therefore, the first round was supposed to mainly improve the framework, since we were focused on the content of intercoder disagreements rather than the ultimate degree of consistency. The disagreements were discussed, and the framework refined accordingly. For the second round, we followed a recommendation how to prespecify data segmentation by O'Connor & Joffe (2020). TK segmented the transcript and applied relevant codes. JV received a document from TK in which the data segments were visible, but the codes were removed, and used his own judgement to apply codes. After this procedure was followed for the coding of three randomly selected transcripts, single numerical measures of the agreement between coders (i.e., intercoder reliability) were calculated: Krippendorff's Alpha = 0.784; Cohen's Kappa = 0.783; Scott's Pi = 0.783; percent agreement = 79.6%. Values above 0.8 are rated as nearly perfect agreements. Since our FA permitted multiple coding (data segments could overlap), and the framework contained a relatively high number of codes, the obtained values were satisfactory. Nonetheless, the framework was refined slightly based on TK's judgements. For instance, codes with a zero frequency of occurrence during both rounds were discarded by merging them with other existing codes.

Indexing
The final thematic framework contained a total of 58 codes, grouped into 7 categories (Appendix 1). Some of these categories, such as the fifth one, were created a priori. Conversely, the second category and its codes emerged purely from the data. Once the thematic framework was finalised, it was systematically applied to the entire dataset. That is, TK started coding all transcripts using NVivo (March 2020 version).

Charting
NVivo is especially helpful for the charting phase due to its built-in "Framework Matrices" feature. In our approach, each category was assigned a chart. The chart had one row per participant and one column per code. Within the cells, participants' responses were summarised, and illustrative quotations marked. This allowed us to organise the data into a more manageable format for the last phase of FA.

Mapping and interpretation
By reviewing the charts, we were able to identify connections within and between participants and categories. Later, the categories were transformed into themes. This phase also allowed for associations between participants' details and codes to be found. Finally, our FA and reporting drew also on quantitative information, namely frequency counts of the number of transcripts that contained a given code. The following narrative system was used in reporting findings: all (24 transcripts); nearly all (23-20); many (19-14); around half (13-11); some (10-6); a few (5-2); and one (1). For the last two categories, the exact participants with a given code are listed by their IDs in the text.

Findings
Personal and interview details (Table 1) constituted the first chart, which was used for sub-group analyses. Five main themes related to tourists' psychological responses to terrorism in Munich's city centre emerged from the remaining charts.

Themes 4.1.1. Comprehending terrorism
Participants tried to comprehend terrorism to navigate the topic and to rationalise their statements. Nearly all participants recounted terrorist incidents from the past. These included incidents from places of residence, other European cities, or recent major incidents. For instance, a few participants (P11 and P22) mentioned the Easter 2019 attacks in Sri Lanka: Religious places have been a very big target for terrorist attacks, just like in Easter, like several churches were attacked in Sri Lanka. So, you would think all these religious places might be a good target for terrorist attacks because it grabs a lot of attention (P11). Although details about incidents were sometimes based on participants' personal experiences, their background knowledge usually came from media. At the same time, around half of the participants acknowledged the interconnectedness of media and terrorism, stating that media amplify the psychological effects of terrorism or terrorists strive for media attention to achieve their goals.
Indeed, many participants comprehended terrorists as bounded rational actors insofar as terrorist acts follow a purpose. Some participants stressed terrorists' ambitions to create and spread fear within societies. Only a few (P2, P9, P16, and P18) claimed that terrorists may be mentally abnormal. Many still believed terrorists aim to inflict as many casualties as possible: What's the goal of terrorists? They want to kill people. If they want to kill people, they will go to a place where is a lot of people (P6).
Participants further revealed their knowledge about various motives behind terrorism. Two general types of terrorism were associated with the phenomenon in Munich's city centre-religiously-inspired and right-wing. Interestingly, specific motives such as anti-Semitic (P5, P9, P15, P17, and P24) and anticapitalist (P15 and P20) were each mentioned by a few participants. Some participants highlighted jihadism when discussing religiously-inspired terrorism.
Regarding the means of attacks, nearly all participants made it clear that usage of vehicles was a possibility. Other less frequently considered means included firearms, explosives, and suicide attacks. The use of knives (P22), chemical weapons (P21), hostage-taking (P21), drones (P7), and cyberattacks (P10) were each raised by one participant. Some participants believed terrorists can be recognised by authorities because they look or act differently from general population. A few participants (P10 and P18) assumed the opposite, making terrorism less predictable and preventable.
Relatedly, many participants took fatalist positions in regards to terrorism. Such fatalism was expressed by claims that terrorist incidents can happen anywhere and anytime, and can be neither predicted nor prevented: Concerning risk perceptions, Munich's city centre was evaluated to some degree as a risky destination in terms of terrorism. The severity of a terrorist threat was appraised as high, since all participants associated terrorism with fatalities, and especially mass fatalities. However, the perceived severity would be quite similar for all destinations. Thus, possibility and probability are more crucial characteristics to be considered when examining tourists' risk perceptions of terrorism in Munich's city centre. Importantly, all participants judged a terrorist attack to be possible within the city centre. The perceived probability of an attack nonetheless varied among participants and sites (Figure 3). According to nearly all participants, the probability of a terrorist attack was highest in Marienplatz (i.e., the main city square), the main shopping street, and the main railway station. In contrast, the Jewish centre, Augustinerstraße (i.e., the street next to police headquarters), and Frauenkirche (i.e., Munich Cathedral) were rated as sites with the lowest probability: This [Augustinerstraße] is the most "not gonna happen" place (P21).
Yet, participants showed a certain ambivalence when evaluating the possibilities and probabilities of terrorist attacks in religious places, and especially Frauenkirche, with many perceiving it as rather likely: I think it is likely because lately I've seen more attacks happening in religious places, especially in mosques and churches. So, I think there is a big possibility and in the extreme world and place, I think there is a definite concern (P7).

Figure 3. Participants' perceived probability of terrorism across visited sites in Munich's city centre
Many participants tried to compare terrorist activities to other places outside of Munich's city centre. Two main, contradictory ideas emerged. Although a few participants (P13 and P18) stated that terrorist activities in Munich's city centre were either more or less likely compared to other European cities, the perceived probability was prevalently on the same level. However, while comparing the probability to destinations frequently affected by terrorism such as Iraq, Syria, Indonesia, or Israel, and sometimes to places of residence such as Bangladesh, Russia, or the USA, participants described Munich's city centre as less likely to experience terrorism. Around half of participants also made comparisons to other types of events threatening their tourist experiences. They mainly stressed ordinary crimes and various accidents as more threatening than terrorism in the city centre: I might be scared of like small petty theft. So, like someone can mug you or rob you… But like on a big level, I don't see it happening (P18).

Location-related factors influencing on-site psychological responses to urban terrorism
A total of 12 more or less intertwined location-related factors that influence tourists' psychological responses to terrorism in cities were identified. The number of people (F1) and visible security measures (F2) seems to be the most important factors, since they influenced the on-site psychological responses of all participants. The presence of a higher number of people caused higher risk perceptions of terrorism and even evoked negative emotional responses for a few participants (P10, P17, P21, and P22), while a lower number of people decreased risk perceptions and eradicated any fear:

It's less likely to happen… as you can see, there isn't a lot of people here [Jewish centre] right now. It's pretty empty (P11).
For nearly all participants, psychological responses to terrorism were also influenced by religion-related factors (F3); the physical openness and physical suitability of a site for an attack (F4); the presence of foreigners, tourists, and other tourism-related factors (F5); and the symbolism, fame, and popularity of a site (F6). The presence of religious symbols or worshippers on the site increased terrorism-related risk perceptions. For one participant (P17), however, the risk perception was influenced by the site's spirituality in the opposite way. The more symbolic, or famous the site was, the higher the perceived risk. For a few participants (P12, P22, and P23), it even evoked more fear of terrorism. Similarly, salient tourist sites-mainly characterised by a higher number of tourists-made participants perceive and feel more terrorism-related risk and fear, respectively. This was primarily because participants knew foreign tourists may be targeted by terrorists seeking to get international recognition or to harm tourism industry, for instance (see Goldman & Neubauer-Shani, 2017). The physical openness of the site was another important factor in how the risk was perceived and processed emotionally, though it had ambivalent effects. While open sites, either because they offered unrestricted access or simply because they were outside, usually led to negative psychological responses, in several cases their openness had the reverse effect:

Because it's open everywhere. Like you have the two big entrances, and you have trains coming in. It's pretty easy to enter. I mean, pretty easy even for terrorists (P13).
Because we're outside, I feel like even less threatened than in the other places (P15).
Many participants' responses to terrorism were influenced by a site's atmosphere (F7). A pleasant atmosphere, often described as peaceful, relaxed, friendly, and liveable, but also characterised by vigilant people or great lighting, favourably enhanced both cognitive and emotional responses. Conversely, inattentive and hurried people around participants influenced their risk perceptions and feelings adversely.
You feel like you are safe or something. Like nothing can happen here because of the whole vibe this place has to offer (P11).
Around half of participants were further influenced by ongoing or possible events (F8), the presence of economic (F9) and authority targets (F10), or the mass transportation system (F11). Religious ceremonies or other mass gatherings increased the perceived probability of terrorist incidents, as did the presence of economic factors such as shops and high-status people. Some participants also believed police headquarters could be targeted, and a few (P6, P7, P16, P18, and P20) were concerned due to the presence of other authority targets (e.g., the city hall). While visiting the subway and train stations, participants reflected on the presence of the mass transportation system, which increased their perceived risk. All these four factors influenced participants' risk perceptions, but not their emotional responses. Finally, the location of a site (F12) influenced the on-site terrorism-related psychological responses of some participants. When a site, such as Jewish centre, was located somewhat on the edge of the city centre, a few participants (P5, P14, and P22) recognised this as decreasing their risk perceptions or fears. Moreover, it must be stressed that the visited sites are not isolated islands; hence, mutual interactions were occasionally revealed.

The role of and attitudes towards antiterrorism measures
Visible security measures represented the most complex location-related factor. Also, the theme was deductively researched as on-site tourists' psychological responses to urban terrorism were theoretically expected to be influenced by visible security measures. Therefore, an individual theme was devoted to them. The measures, including security personnel, cameras, planters, and bollards, influenced the onsite psychological responses to urban terrorism of all participants. Moreover, they aroused the most emotional responses out of all identified factors. The responses, however, followed both positive and negative directions. Many participants without any doubts admitted visible security measures made them perceive less risk and feel less fear of terrorism, while the lack of such measures increased their risk perceptions and fears: It makes you feel much safer. Much, much safer. So, you know that even if there isn't a policeman, not physically there, you know that someone is still watching (P7).
Still, a few participants (P4, P9, P10, and P12) showed rather mixed responses, with mainly emotional ones being influenced adversely. In other words, the presence of security measures increased their fears, and the lack of such measures had the reverse effect. When assessed cognitively, though, the presence and lack of visible security measures among these few participants were not always risk inducing and reducing, respectively. The discrepancies may be explained by the type of the measure considered.
The measure of police officers and other security personnel was discussed by nearly all participants and influenced psychological responses to terrorism in all these cases. Of all the considered types, it had the strongest influence on participants' emotional responses. Such personnel were the main reason for the above-mentioned negative terrorism-related emotions caused by visible security measures.

The fact that I don't see any armed guards around actually makes me feel safer because I think when you have these people standing by it's a constant reminder of terrorism (P9).
Apart from feeling less safe, participants also revealed certain nervousness and discomfort due to the presence of the personnel. On the other hand, around half of participants' emotional responses were favourably enhanced by security personnel contributing to feelings of safety.
CCTVs facilitated the favourable terrorism-related psychological responses of around half of participants. However, many admitted certain reservations about CCTVs effectiveness due their operational issues or the nature of terrorism. For instance, they thought CCTVs are unable to prevent terrorist attacks when perpetrators are suicide attackers. In addition, a few participants (P10, P16, and P18) were concerned about the effects "Big Brother's" control of CCTVs may have on people.
There are two types of access control measures in Munich's city centre. Security planters were discussed mostly as a measure at Marienplatz, while bollards surrounded and fortified the entire Jewish centre. The former favourably enhanced psychological responses to terrorism of only some participants. The measure was assessed by many as ineffective in preventing terrorism, since the planters were allegedly wrongly placed or in insufficient numbers: I think that they're very badly placed, any car can fit through them. It's stupid, pointless, meaningless, and won't stop anything. Like, look at the width. You could fit a truck through between that. They don't make a difference (P9).
Many participants further stated they were not aware of planters' antiterrorism purpose ahead of the interview, and a few (P4, P7, P11, and P18) reported seeing them as a beautification element in the public place. Bollards were assessed as a more effective access control measure, with many participants having positive psychological responses to terrorism due to them. The bollards, unlike the planters, are mechanical, and since they can be hidden underground when needed, they fortify an entire site, so it is impossible for any vehicle to go around or between them: I think that these bollards are much more effective at stopping threats such as a truck. So, these one's I'm much more comfortable about… it makes me feel safer (P22).

Personal coping with terrorism
Participants' coping mechanisms in dealing with a terrorist threat were identified only sporadically within data. Many participants established they would not change their spatial or temporal behaviours to reduce their fears and risk perceptions of possible terrorist threats. Only a few (P7, P21, P22) admitted their preferences for avoiding environments such as transportation hubs, crowded places, and festivals:

I also don't like to spend time in train stations just to stand around in case there's something happening… maybe a terrorist attack, yeah. So even though I don't feel threatened right now, I wouldn't just spend time there (P22).
Some participants would apply this mechanism if the threats were immediate. Relatedly, the coping mechanism of being vigilant to identify such an immediate threat was also mentioned by some participants. The mental coping mechanism of purposefully ignoring the topic of terrorism was evident among some participants, who sometimes rationalised such moves as not fulfilling terrorist goals:

Most of the time I really avoid thinking about this kind of stuff like that. I don't see the point to think
about it. I will get paranoid, and the thing is, I would serve their objective. So, they want to scare people. Like I don't want to be scared, and I think that's a message that they want to put (P2).

Associations
Our results indicate certain associations between tourists' attributes and the identified themes forming their psychological responses to terrorism. FA revealed only participants from non-European countries, especially Asian countries, reported heightened levels of fear of terrorism at some point. These fears were also reported by participants less experienced with travelling abroad.
Differences in risk perceptions between various participant groups were observed when assessing probabilities across the visited sites. Participants from Asia perceived the relative probability of terrorist attacks as substantially higher in the main railway station compared to participants from Europe and Americas, while Americans perceived the relative probability as higher in Frauenkirche compared to Europeans and Asians. Interestingly, the Jewish centre was considered as the most probable site for terrorist attacks only by a Jewish participant from Israel.
If an attack is going to happen somewhere, here [Jewish centre] is a place with the highest chances that it will happen (P24).
The relative probability at Marienplatz was evaluated as substantially higher by participants who had travelled to the city for leisure purposes than by those who had travelled for business/education, by participants who had come for the first time compared to participants with past experience, by participants who had spent less time in the city compared to those who had stayed for longer periods, and by participants with lower general concerns about terrorism compared to those with higher concerns. The influence of tourists' attributes on their psychological responses to terrorism in Munich's city centre was also supported by participants' commentaries: I'm coming from Israel. So, the odds that something will happen here [Munich's main railway station] are way lower than in Israel. So, I don't feel any unconfidences (P24).
The attitudes to visible security measures were influenced by participants' attributes as well. The backfiring effects appeared only among non-believers, participants who had travelled to the city for leisure purposes, those who had not visited the city in the past, and those who reported rather lower general concerns about terrorism. Additionally, these participants were of British, Irish, Swiss, and Taiwanese nationalities, which seemed to be an important factor, as one participant explained: It makes me feel a bit nervous… I think maybe quite a British attitude because we're really not used to seeing… I don't know if those guys were armed. I think they probably were… But it doesn't make me feel safe and makes me feel the opposite (P4).
CCTVs caused uncomfortable feelings of being controlled only among male participants, non-believers, participants travelling to the city for leisure purposes, and participants visiting the city for the first time.
The employment of behavioural mechanisms to cope with terrorism was noted only in interviews with participants from non-European countries, and especially by Asians. The mechanism of purposefully ignoring the topic of terrorism was present only among male participants.
In terms of associations among main identified themes, the most obvious direct association between location-related factors and on-site psychological responses to urban terrorism was indicated in the previous subsection. Nonetheless, more associations can be identified. To begin, how a participant comprehended terrorism had substantial control over if and how location-related factors influenced their on-site psychological responses to urban terrorism; thus, the participants' comprehension of terrorism indirectly influenced risk perceptions and fears of terrorism. For instance, matching terrorism with mass fatalities was among the main reasons why the number of people was a crucial factor influencing participants' on-site psychological responses to urban terrorism: This is a public place, so many people in every time of day… So, this is the first reason, always people, and terror has no reason you know, the only aim is just to kill people and to spread the violence everywhere (P17).
Recounting previous terrorist incidents also determined how location-related factors influenced the participants' on-site feelings and perceptions. Moreover, the influence of comprehending terrorism on on-site psychological responses to terrorism can be a direct one. The fatalist attitudes to terrorism made some participants assess attacks in the city centre as possible. However, the influence of the fatalism was double-edged. It was a direct cause of the rejection of the idea that people should adjust their behaviours to avoid the threat. A similar logic applied to being aware that spreading fear is a goal of terrorism: Attitudes towards visible security measures were also somewhat influenced by participants' comprehension of terrorism. For instance, seeing terrorists as persons who are hardly recognisable contributed to reservations about CCTVs' effectiveness.
As expected, participants' on-site psychological responses to terrorism had strong influence on personal coping with terrorism. In particular, the admitted changes in local behaviours were restricted to participants who reported heightened fears of terrorism, while the absence of the fear mostly resulted in commentaries that no changes in spatial and temporal behaviours would be adopted. The reverse effect, running from coping to psychological responses, was also observed within the data. According to participants' commentaries, purposeful topic avoidance was employed to eradicate any fears of terrorism.

Discussion of the findings and theoretical contributions
Our research extends the literature on tourists' terrorism-related risk perceptions into the underresearched area of location-related factors influencing tourists' cognitive and affective responses to terrorism. In doing so, it focuses on scarcely examined on-site psychological responses and city destinations, especially those in Europe, which has experienced a unique wave of terrorism within the last decade.
The findings suggest that negative psychological responses to terrorism may occur among actual tourists when visiting European cities. Nevertheless, and in line with previous research (Ferreira et al., 2019;Vanneste et al., 2017), participants' emotional responses, such as fear of terrorism were mostly intact. The simplest explanation for the absence of the fear would be in the perceived low threat. It may be that the fear-appeal message did not have a sufficient effect, and/or that terrorism itself does not have a sufficient effect on many participants in our sample. The cognitive assessments of terrorist threats, however, were far from being low. Participants perceived terrorist incidents as highly lethal, possible, and sometimes even likely. Furthermore, even these instances of perceived high probabilities were often not accompanied by elevated fears, although fears were expected to be sensitive simply to the possibility, rather than the probability, of negative consequences (Loewenstein et al., 2001).
It is also possible that the fear appeal led to the fear-control process, which is a form of defensive processing resulting in the rejection of change of beliefs about a threat and downplaying one's vulnerability to the threat as a coping mechanism with their fears (Witte, 1992). With terrorism being an uncontrollable threat, it is likely participants initiated such process. This explanation for the reported low fears was also supported by participants' commentaries that terrorism is unpredictable and unpreventable, and that they purposefully ignore the topic of terrorism.
Additionally, the relatively low fears can be associated with the sample of participants, which comprised actual, rather adventurous, and young tourists. Negative psychological responses to terrorism are less common among tourists of these types (e.g., Brun et al., 2011;Isaac, 2021;Kapuściński & Richards, 2016;Morakabati & Kapuściński, 2016). However, the present study confirmed cognitive assessments may substantially differ from emotional responses; hence, the absence of fear in perceived high-risk conditions somehow disrupts this explanation. It may remain valid when considering evolutionary preparedness-even if people recognise terrorist threats at cognitive levels, they are likely to react with little fear because evolution has not prepared them for such stimuli (Loewenstein et al., 2001). Moreover, the revealed discrepancy between participants' cognitive and emotional responses to terrorism suggests that felt invulnerability among young tourists is more emotionally than cognitively caused (Loewenstein et al., 2001). Indeed, it was found only participants who reported heightened fears of terrorism were willing to adjust their behaviours to cope with the threat. This finding supports claims that emotional responses are better predictors of behaviours than cognitive evaluations (Wolff et al., 2019).
On-site psychological responses to urban terrorism have been shown to fluctuate due to aspects of the immediate physical and social environment. As a result, the psychological responses seem to vary substantially on a micro-scale-that is, between various sites in city centres. We identified 12 such location-related factors (Figure 4). The number of people was one of the most important ones; an increased presence of people led to more negative psychological responses. This reflects a shift in the trajectory of terrorist attacks away from critical infrastructure assets toward crowded places in the period after 9/11 (McIlhatton et al., 2020). The participants clearly recognised such developments, and associated terrorism with mass civilian casualties. Furthermore, this finding adds another dimension to the issue of overcrowding in the context of (urban) tourism. Apart from having a direct negative impact on tourists' experiences (Popp, 2012;Rasoolimanesh et al., 2019), it may also affect them indirectly through security threats.
It is not only about "body counts," but also about "what body counts" (Kapuściński & Richards, 2016). Our findings revealed that the presence of other tourists increased participants' risk perceptions and even fears of terrorism. In line with Neumayer and Plümper (2016), targets somehow relevant to participants (e.g., Jews for an Israeli participant) amplified these responses.
Previous research found that sites with police stations may be at higher risk of urban terrorism (Marchment & Gill, 2019), and government buildings to be the most avoided leisure sites after 9/11 (Chen & Noriega, 2004). Although the presence of authority targets caused higher risk perceptions among our participants, this factor was often overridden by others. Therefore, a usually empty street next to the police headquarters was evaluated as a place safe from terrorism. Still, the lowest probability of terrorist incidents was perceived for the Jewish centre. This indicates that although religion-related factors influenced the on-site psychological responses to urban terrorism of nearly all participants, they also were not always strong enough to surpass other factors. Marchment & Gill (2019) also discovered that ease of access and escape for terrorists increases the likelihood of an area being targeted. Our participants' psychological responses to urban terrorism were often in line with this finding, since they were adversely influenced by the physical openness of sites. In other cases, the open sites acted more favourably, often through evoking feelings that one was not "squeezed" by the space or that a participant could escape easily if necessary.
Antiterrorism measures are another highly important location-related factor. While including visible security measures, these defensive measures are designed to prevent or deter terrorist attacks (Grosskopf, 2006;McIlhatton et al., 2020). Overall, the measures favourably enhanced risk perceptions and fears of terrorism of most participants. Still, our research supports previous findings about the mixed feelings among tourists (Ferreira et al., 2019;Survila et al., 2017), as a few participants described the visible security measures as fear-inducing. These responses were associated with certain tourists' attributes (e.g., nonreligion). Results also suggest that tourists of nationalities used to less strict law enforcement are more likely to dismiss the idea of protection through visible security measures.
In consensus with previous research on the attitudes of non-tourist populations (Grosskopf, 2006) and of airline passengers (Al-Saad et al., 2019), our study further shows the effects to vary based on the type of the security measure. Similarly to Adeloye & Brown (2018), Chen & Noriega (2004), and Isaac (2021), we found the presence of security personnel mostly contributed to positive feelings and risk perceptions. It was assessed as the most effective measure, but at the same time it had the strongest emotional influence, both positive and negative. Other measures were seen as less effective and aroused fewer emotions. As a result, they were less controversial.

Figure 4. A conceptual model of the influence of location-related factors on on-site tourists' psychological responses to urban terrorism
This study also established that the way tourists comprehend terrorism plays a crucial role in their onsite psychological responses to the threat. Background knowledge about past incidents, understanding of terrorist goals and types, means of attack and other information available to respondents influences whether and how they will perceive and feel terrorist threats. Our participants had considerable knowledge about terrorism, which supports findings of Seabra et al. (2014) that tourists pay attention to it. This interest produces motivation to acquire information in media. Despite that, tourists seem to be aware of the communication role media play in terrorism (Adeloye & Brown, 2018;Isaac, 2021).
The frequent mentions of religiously-inspired and jihadi terrorism in our study correspond to the findings of Kapuściński & Richards (2016) and Woods (2011), according to whom the jihadi type arouses relatively higher risk perceptions and worries. However, we did not find any evidence that it would cause more negative psychological responses than other types of terrorism. Our study further indicates that not only attacks committed by unorganised individuals, as suggested by Wolff & Larsen (2017), but all terrorist incidents, can be seen by tourists as random acts. For that reason, the gambler's fallacy, which in this context would mean a perspective that terrorism will not reoccur in a rarely hit destination, should be even more common cognitive bias affecting tourists' psychological responses. Yet, no participant explicitly raised such ideas during the FAWIs. Perhaps three years after the incident was too long for the responses to be affected this way.
Besides seeing terrorism as unpredictable, our participants also described it as hardly preventable. These fatalist positions were a cause of participants' assessments that terrorist attacks are possible within visited sites. Thus, on a micro-scale, fatalism can contribute to higher risk perceptions. In broader terms, however, it seems that fatalist positions regarding terrorism have positive effects on tourists' risk perceptions and decisions (Fischhoff et al., 2004). Indeed, our study confirmed that these sometimes work as a kind of mental coping mechanism (Ferreira et al., 2019;Rittichainuwat & Chakraborty, 2009;Uriely et al., 2007), so that tourists will continue in their activities within visited destinations, although they may still perceive a certain degree of a terrorist threat.
Our findings also revealed certain associations between participants' attributes and themes forming their psychological responses to urban terrorism. As with any qualitative research, however, these findings represent only assumptions, which should be addressed by future research.
To summarise all presented theoretical contributions, the most important one, which the study provided, was identification of various aspects of the immediate physical and social environment having an influence on on-site tourists' psychological responses to urban terrorism. The number of people and antiterrorism measures appear to be the most important factors. Such a theoretical contribution enriches, from a micro-scale perspective, the previous state of tourism research knowledge regarding the role of destination (external) attributes in terrorism-related psychological responses. In terms of tourists' (internal) attributes, the study clearly explored that the way tourists comprehend terrorism should be recognised as one of the major characteristics having a substantial influence on their on-site terrorism-related perceptions and emotions. The study also confirmed various assumptions and findings of previous studies, for instance, the existence of fatalist positions regarding terrorism and their influence on tourists' behaviours. Furthermore, the validity of the risk-as-feelings hypothesis was confirmed in the context of the tourism-terrorism nexus. Both cognitive assessments and emotional responses facilitate the process of risk-related tourists' behaviours. They may nevertheless substantially differ from each other, and emotional responses to terrorism are better predictors of behaviours than cognitive evaluations. Finally, although the findings are primarily related to the on-site tourists' psychological responses, many of our findings may also apply to the psychological responses when considering future travel intentions.

Managerial implications
Arguably, the best way to reduce terrorism is to take terror out of it (Hoffman & Shelby, 2017). We tried to contribute to this task by asking why tourists in city destinations are afraid of or perceive the threat of terrorism, as well why they do not. To understand these questions, other sub-questions must be addressed. Not only the very important and frequently researched question "Who is afraid?" should be answered, but also the questions "of what, of whom, and through what?" (Warr, 1990). Such comprehensive understanding of tourists' psychological responses to urban terrorism may be valuable for emergency and security professionals, as well for tourism managers and urban planners.
One implication of the present research is that on-site tourists' fears aroused by terrorism may lead to behavioural changes. Although these coping mechanisms, namely crowd avoidance, were infrequent among our participants, city destination management teams should not ignore the need to reduce factors contributing to fears of terrorism, at least those which are in their power. If a destination experiences a heightened (perceived) risk of terrorism, such preparedness may be crucial for its tourism resilience. The present study adds a new argument for the necessity of dispersing tourists within cities, since a higher number of people and tourists at sites leads to more negative psychological responses to terrorism among tourists, and their experiences may therefore be further undermined.
Not all factors increasing on-site fears of terrorism should be reduced. Although the fame or (religious) symbolism of sites can make tourists afraid of terrorism, these attributes are crucial for the tourism industry. In these instances, antiterrorism measures are supposed to restore equilibrium. Indeed, our study provides support for the assumption of visible security measures reinforcing feelings of safety rather than inducing fears (Dalgaard-Nielsen et al., 2016). The positive responses and attitudes should be further interpreted in the context of participants who were willing to accept the measures even without self-reported fears. Previous research found that tourism managers presuppose the positive effects of the antiterrorism measures and reflect them in their strategies to boost tourist confidence (Rittichainuwat & Chakraborty, 2009). However, the present study indicates the existence of possibly small, but for some destinations highly relevant, tourism market whose psychological responses to terrorism are adversely influenced by antiterrorism measures. Hence, destination management teams should be involved in the process of securing cities so that the choice of antiterrorism measures will, to a certain degree, reflect the attitudes of tourists whom the destination wishes to attract.
To successfully build tourism resilience toward terrorism, one must also look-apart from the destination resilience-at the resilience of tourists (Veréb et al., 2020). Such tourist resilience building must take into consideration not only the short-term impacts of terrorism but also its long-term impacts. That is, it should adopt the proactive approach, ideally prior to situations of emergence and recovery. Since the long-term impact of the fear of terrorism is in the ability to distort the personal values of open-mindedness, novelty seeking, and diversity appreciation, fear management in the counter-terrorism domain should aim to neutralise negative, and improve positive, coping mechanisms of the society (Bakker & de Roy van Zuijdewijn, 2022). Rather than recommend avoiding, fear management should recommend, for instance, pulling together. One of the keys to fear management may be to understand how terrorism is comprehended by societies. Our study indicates, for instance, that tourists predominantly associate terrorism with mass killings, and religiously-inspired, random, and unpreventable incidents. Even more importantly, they often understand that terrorism purposefully attempts to arouse negative psychological responses. Working with this knowledge may be a start for comprehensive communication strategies attempting to build resilient tourists. If the communication succeeds, applying positive coping by tourists in terms of adopting positive behaviours and attitudes to minimise their stress may in the end be not only beneficial for the tourism industry but the whole society as tourists may be carriers of the idea to restore normal order.

Limitations and future research
Neither tourism nor terrorism will be eradicated by the COVID-19 pandemic. Regarding terrorism, quite the contrary (Kruglanski et al., 2020;Marone, 2021). It is noteworthy that our data collection took place ahead of the pandemic; therefore, caution is warranted when interpreting the findings, as COVID-19 became the main threat constraining travel in the current times 4 (Perić et al., 2021). Regardless, with the return of mass tourism, the industry must be ready to cope with multiple threats, from which terrorism cannot be omitted. Despite the post-pandemic preference for crowd avoidance (Ivanova et al., 2021), it can be expected that city destinations will be crowded again, and terrorism will hit again. It is crucial to understand its effects on tourism demand, especially psychological ones.
This study has limitations stemming from the sample of participants. Interviews were conducted in English. Accordingly, only tourists able to speak the language were included. Additionally, since around two thirds of the participants were not native English speakers, difficulties to express ideas may have occurred. Focusing on foreign tourists nevertheless largely reduced these biases in the data collection.
Addressing domestic tourists' psychological responses to urban terrorism remains a challenge for future research. Concerning the nationality profile of our sample, rather than to be representative the goal was to cover a great diversity of nationalities in order to analyse tourists' views in the face of terrorism in the global society. The findings of the present study are further restricted to rather adventurous and young tourists. Future studies should incorporate older age groups, for which higher risk perceptions and fears are expected (Brun et al., 2011;Isaac, 2021).
Apart from being international, adventurous, and young, other features of the participants' sample may have further influenced the generalisation ability of the study's findings. The fact that most of the participants were male, their purpose of the visit was mostly leisure, and it was usually their first-time stay in Munich, may have had another substantial influence on the entire picture of their responses in relation to terrorism. In this sense, previous research shows, for instance, female tourists to be more adversely influenced by terrorism than their male counterparts (Brun et al., 2011;Isaac, 2021;Wolff & Larsen, 2017). Our analysis of associations between tourists' attributes and the identified themes forming their psychological responses to terrorism also indicates that the discrepancies in reactions to terrorism may be a result of the mentioned tourist characteristics.
Still, the ultimate goal of qualitative studies is not to generalise but rather "to provide a rich, contextualized understanding of human experience through the intensive study of particular cases" (Polit & Beck, 2010, p. 1452. Such a rich understanding can, however, be used for reasonable extrapolation. Findings that were relevant to all (or many) participants should be indeed considered as broadening theories. Moreover, the generalisability of our findings in the qualitative-research context should be also supported by the possible transferability of this inquiry (Polit & Beck, 2010). Relatedly, a broad and diversified approach in sampling increases the ability of generalisations of theoretical contributions to other settings and different groups of tourists. The diversity arose, for instance, in relation to the days already spent in Munich ahead of the interview. While some participants arrived on the day of the interview, others spent a significant number of days in Munich ahead of the interview enabling them to explore and get more familiar with the city and even the sites visited during the interviews. Since previous literature provides a lot of evidence that familiar environments can dramatically reduce negative psychological responses to terrorism (e.g., Isaac, 2021;Warr, 1990), the possible influence of this attribute on individuals' commentaries must be recognised.
It is also possible that there are other location-related factors this study was unable to explore. For instance, darkness has a strong influence on fears of crime (Warr, 1990). Although increases in threatened feelings during night times were implied by one of our participants, a detailed investigation of the issue would be more feasible during winter months.
Despite having an extensive history of terrorism, Munich cannot be considered a risky destination for several reasons. First, there is its location within the relatively safe and stable region of (Western) Europe (Bayar & Yener, 2019), which also safeguards a low likelihood of reoccurrence of attacks (Walters et al., 2019). Second, Munich has essentially been a single-incident destination during the last decades, which is important considering that frequency deters more tourists than the intensity of terrorist incidents (Pizam & Fleischer, 2002). Third, there were no publicised terrorist incidents during the study's investigative process (Sönmez & Graefe, 1998a). All these factors should have been partly offset by the fear appeal approach to our walking interviews; however, we cannot exclude that in the nonlaboratory environment, such conditions influenced participants' commentaries.
Still, future studies should examine the effect of intense fears (Loewenstein et al., 2001, p. 21). Examining riskier destinations may help in this process. This would, however, encompass a lot of difficulties in the data collection process (Adeloye et al., 2020). In addition, techniques, such as video, can be used for fear arousal. Relatedly, a more quantitative approach to the role of fear appeal in tourists' behaviours should be employed in future studies. The fact that the exact amount of fear aroused in our participants is unknown represents a clear limitation of this study. The use of control groups is necessary to reveal what amount of fear leads to acceptance of coping mechanisms like adjusting micro-scale behaviours.