Are Serbian tourists worried? The effect of psychological factors on tourists’ behavior based on the perceived risk

2.1 Natural hazards risk perception and travel intention

The research studies in tourism have recognized the lack of safety and security as one of the most dominant inhibitors of travel intentions [9, 10]. Thus, the tourism industry is facing a challenge to offer tourists a safe and pleasant visit and minimize different kinds of risks and unsafe situations. The concept of “tourism risk” can be understood as the possibility of experiencing various unfavorable situations or events at a tourist destination [11] without the ability to perceive the exact consequences or negative results after making travel decisions [12]. This causes the state of anxiety, fear and discomfort, which can result in destination avoidance [13, 14, 15]. Consequently, one of the most important streams of general tourism risk related literature is tourists’ risk perception and its influence on their intention to visit or avoid certain travel destination [16]. As noted by Sönmez, Apostolopoulos and Tarlow [13], it is evident that tourism demand decreases as the perception of risks at certain destination increases, which is why research on risk perception and the factors affecting it is so important for the tourism industry. Beside health-related risks, war, terrorism, and political instability as disasters with a greatest negative effect on travel intentions [14, 17, 18], the risk of natural hazards has been also recognized as an important safety treat affecting the choice of travel destination even in a longer period after such event [19, 20, 21, 22, 23]. These hazards that are caused by nature range from eruptions, avalanches and earthquakes to landslides, floods, wildfires, hurricanes and typhoons. Jonas et al. [24] suggest that such risks associated with natural hazards are connected with tourists’ physical risk that influences their decision to visit affected destinations or destinations with a high risk of natural hazards.

In the context of natural hazards, several authors focused on the effects of natural catastrophes on visitors’ risk perception and visit intentions. For instance, Lehto et al. [21] explored the impact of a tsunami on travel intentions and found that such a natural hazard negatively influenced the intention to travel to seaside destinations. Park and Reisinger [25] conducted an analysis based on different types of natural disasters and revealed the negative influence of these on international travel. Interestingly, Walters et al. [26] noticed that a personal connection with the affected area (flood-stricken destination) and repeat visiting, positively influence the intention to visit, while the main factors affecting destination avoidance were the perception of bad weather and insecurity of the place. Based on previously stated, it can be assumed that risk perception is largely related to travel decisions and intentions to travel. However, the perceptions of travel risk and safety, as well as related behavioral intentions can largely differ between different types of tourists. Reisinger and Mavondo [27] claim that marketers need to target tourists with higher risk tolerance and risk acceptance level,which is why the relationship of personality and risk perception is an important topic to study. Thus, in order to fully understand the effect of risk perception and behavioral intention in risky periods, exploring the personal factors affecting travel risk perception and behavioral intentions are especially important. Practically, such information could be applied in making responsive destination management and marketing strategies in the time of crisis. Finally, Thapa et al. [8] analyzed wildfires in the context of Florida (USA) and tourists’ risk perception and reactionary behaviors. They indicate three types of tourists - the most dominant, middle-risk (Conscious Travelers), higher risk (Cautious Travelers) and lower risk (Courageous Travelers) segments. This study has inspired the current research, as the authors used the scale of tourists’ risk perception and reactionary behaviors [8]. This scale, in its essence, is measuring intention to visit based on tourist perception of risk on that particular destination, although Thapa and colleagues [8] mention it only as “travel risk perception scale”. In order to avoid any possible misunderstandings, the authors of this study renamed the scale to “tourist behavior based on the perceived risk”.

2.2 The role of psychological factors in natural hazards risk perception and intention to visit

Many authors acknowledged that tourists perceive and respond to risks differently based on their sociodemographic characteristics [8, 28, 29, 30, 31] and personality traits [3, 27, 32, 33, 34, 35, 36]. Such a notion even influenced one of the earliest tourist typologies connecting psychological traits with risk seeking and risk perception [3, 5]. One of the most well known is typology by Cohen [3]. He differentiates four types of tourists based on personality traits and risk perception and their preferences for either novelty seeking or staying in a safe environment. His study suggests that novelty seeking as a personality trait is tightly related to risk perception, indicating that novelty-seekers will avoid “risky” destinations less. The psychographic typology of Plog [4] (allocentric, psychocentric and midcentric) could be also used to explain the relationship between psychological traits, travel worries and perceived risk in tourism. According to Plog [4], psychocentrics are more anxious and worried; they avoid risk and prefer to travel to familiar and safe destinations. On contrary, allocentrics prefer to travel to unusual places; they are more confident, less worried and anxious, motivated by novelty [37]. Thus, tourists who have “allocentric” traits are expected to perceive less risk and intend more to visit risky destinations.

Correia, Pimpao, and Crouch [38] showed that the risk perception is affected by the age, familiarity with the destination and travel experience but also by novelty seeking. They found that novelty seeking moderates the relationship between mentioned variables and risk perception – novelty seekers are less sensitive to risk. In connection to this, Lepp and Gibson [32] showed that the perception of risk is directly related to preferences for familiarity versus novelty by exploringhowattitudes to specific tourism risks varied across Cohen’s typology [3]. Their study implies that tourism risk factors were perceived to be less risky by novelty-seekers, explaining “what may be a source of fear for the organized mass tourist may be a source of excitement for the drifter ” [32, pp. 617]. Reisinger and Mavondo [27, p. 214-215] also claim “tourist seeking familiarity is likely to perceive an alien environment as riskier than a tourist seeking novelty”. They also show that risk perception produces higher travel anxiety, which negatively affects the intention to travel. In connection to this, an interesting approach was adopted by Larsen et al. [7], who developed a scale to measure tourist worries and related them to the risk perception, risk acceptance and desire to travel. The authors have found that tourist worrynegatively correlates with the desire to travel, and positively with risk perception.

Sensation seeking and its relationship with risk perception have also attracted research attention [32, 33, 34, 35]. Most of these studies show that sensation-seekers, similarly to novelty-seekers, perceive destinations as less risky and show greater intention to visit them. However, a study by Lepp and Gibson [32] showed that sensation seeking does not influence tourist perception of risk but their travel intentions to visit risky destinations. Similarly, Lepp and Gibson [32] have shown that sensation seekers are more likely to choose explorer and drifter roles (with a tendency to travel internationally and choose risky destinations), but found that this was not related to risk perceptions. Pizam et al. [33] also suggested that risky behavior and sensation seeking positively correlate. It is also important to mention that Zuckerman’s [39] concept of sensation seeking consists of four factors (i.e. Thrill and Adventure Seeking, Experience Seeking, Disinhibition, Boredom Susceptibility) while only experience seeking explicitly refers to traveling. Thus, Kapuściński and Richards [36] claim that it is not clear how and to what extent sensation seeking and risk perception are related in terms of tourism, indicating the need for further research of these concepts.

Based on these findings, two hypotheses can be drawn:

Hypothesis 1: Novelty-seeking will positively influence Courageous behavior, but negatively Cautious and Conscious behavior.

Hypothesis 2: Sensation-seeking will positively influence Courageous behavior, but negatively Cautious and Conscious behavior.

In the rather obscure literature focusing on the relationship between personality traits, risk perception and related behavior, the relationship of other psychological traits (such as Big Five Inventory or similar) with risk perception and intention to visit risky destinations was not explored till now. This is why this study intended to analyze how 6 personality traits (Extraversion, Conscientiousness, Openness, Agreeableness, Neuroticism, Honesty-Humility) of Mini-IPIP6 [40] (The International Personality Item Pool) scale, in addition to Novelty seeking and Sensation seeking, are related to risk perception and intention to visit. The aim was to identify the profile of people who would avoid such risky destination and those who would still visit it.

Apart from novelty-seeking and sensation-seeking, other personality traits explored in connection to perceived travel risk were self-confidence [41] and traits such as extroversion, activity, confidence, helpfulness, and venturesomeness [27]. Both studies revealed that personality is associated with the perception of risk and travel anxiety (individuals who are confident, extroverted, and venturesome tolerate high risk and engage more frequently in risky activities). Outside tourism field, it is relevant to mention the study by Cooper, Agocha and Sheldon [42] focusing on personality traits and risky behavior and revealing that neurotic individuals are more ready to engage in risky behaviors as away to cope with aversive mood states, while extraverts were more likely to engage in risky behaviors as a way to enhance positive affective experience [42]. However, this study investigated risky behavioral intentions, rather than the pure perception of risk. Finally, the study by Weller and Tikir [43] explored domain specific (health, recreation, social and ethical) risk-taking and HEXACO personality structure (HEXACO is an acronym for its factors: Honesty-Humility, Emotionality, Extraversion, Agreeableness, Conscientiousness and Openness to Experience). The results of this study are particularly interesting in social (e.g., joining new groups, asserting one’s opinions in a group, etc.), and recreational (e.g. skydiving, whitewater rafting, etc.) domains of risk-taking because of their connection with tourism field. Moreover, health/safety domain (e.g. smoking, binge drinking, failing to wear seatbelt, etc.) is also important in the context of the current study, as natural hazards occurring at certain destination represent a real threat for tourist health and safety. It is reasonable to assume that those who are careless with their health wouldbe less worriedby the possibility of natural hazards occurring at the destination. The main finding of their study suggests that in case of all risk domains, the emotionality dimension is associated with higher risk perceptions and risk-taking. Moreover, openness is positively associated with risk-taking for social and recreational risks,whereas lower honesty/humilitywas associated with greater health/safety and ethical risk-taking but lower risk perceptions. However, honesty/humility did not show significant influence in the recreational and social domain. The authors explained this by the fact that these domains include activities, which are more socially acceptable, while high honesty humility individuals may perceive fewer benefits associated with such activities. This may result in their avoidance of such behaviors. Conscientiousness also showed a significant negative effect on risk perception in all domains. Although extraversion was not found to influence risk-taking in the study by Weller and Tikir [43], some previous studies revealed that there is a positive effect of this personality trait on risk-taking (e.g. [44, 45]. Except for extraversion, agreeableness has not shown significant effects on risk-taking but was positively correlated with risk perception.

Based on these findings, six more hypotheses can be drawn:

Hypothesis 3: Extraversion will positively influence Courageous and Conscious behavior.

Hypothesis 4: Neuroticism will negatively influence Courageous, but positively Cautious behavior.

Hypothesis 5: Conscientiousness will negatively influence Courageous, but positively Cautious behavior.

Hypothesis 6: Agreeableness will positively influence Conscious behavior.

Hypothesis 7: Openness to experience will positively influence Courageous, but negatively Cautious behavior.

Hypothesis 8: Higher Honesty/humility will positively influence Cautious behavior.

As previous studies found that anxiety related to traveling has a negative effect on intention to visit risky destinations [27, 41], the authors assume that tourism worries may mediate the relationship between analyzed personality traits and tourist behavior based on the perceived risk. Thus, the final hypothesis can be proposed:

Hypothesis 9: Tourism worriesmediate the relationship between personality traits and tourist behavior based on the perceived risk.

The formulated hypotheses were tested using structural equation modeling (SEM) done in EQS 6.1.

3.1 Study area

The danger of potential natural disasters can greatly affect the development of the country’s tourism. The current research was conducted for the case study of the Mediterranean country, Greece. As one of the major tourism countries in Europe, Greece is the 8th country in the EU with the highest number of bed-places and overnight stays in tourist accommodation establishments in 2017, according to Eurostat report for 2017 [46], with the total of 21,304,135 tourists in hotels and similar establishments in 2017 [47] (ELSTAT, 2018). However, this country has a very dynamic history of floods and wildfires, which have hit some of the most visited Greek tourism destinations in 2018 – Chalkidiki, Olympic Riviera, Thessaloniki and Athens. Due to the fact that this country is officially the most popular tourist destination for Serbia tourists (Serbian tourists have spent most of the money, 392 million Euros, in Greek summer resorts at theAegean Sea) (Research conducted by Belgrade Tourism Fair, 2018) and having in mind that Serbia is among the top ten countries that visit Greece [48], it was interesting to determine how disasters, which hit this country in 2018, influence tourist behavior based on the perceived risk at those destinations.

According to the EM-DAT (The international disasters database) [49], since the beginning of the 21st century, the top natural disasters in Greece are wildfires, extreme temperatures and floods. Primarily, the reason lies in the geographical position of Greece, its geological composition and the influence of the complex relief [50]. It is situated in the south part of the Balkan Peninsula, specifically, in the southeastern part of Europe.

It is widely known that natural disasters leave the greatest consequences in urban and high-density areas. In Greece, the population is mostly concentrated in coastal regions (Athens, Thessaloniki, Patras) and on the islands (Heraklion), where during the summertime number of inhabitants and tourists increases. Reciprocally to it, disaster risk is also increased.

Mountain ranges, which extend along the central part of the country, and other mountains, affect the air coming from the moisture source in the central Mediterranean Sea [50]. Although Greece is a country with typically Mediterranean climate (mild, wet winters and hot, dry summers) the high altitudes of Pindus mountain range as well as local and seasonal wind have a strong influence on its the pluviometric regime [50]. Namely, in climatic and hydrological terms, Greece can be divided into the western–wet part and the eastern – arid part. The annual rainfall on the western part (side) of Greece is between 1.000 mm and 1.500 mm, and on the eastern part, it is less than 500 mm of precipitation [51]. If we compare the amount of precipitation with the geological composition and structure, the results are unfavorable for Greece. In the geological structure and composition of East Greece dominate rocks and sediments from Paleozoic age [52], mostly not waterproof, but without enough precipitation. This type of terrain explains the fact that the eastern part of Greece is more susceptible to flooding due to extreme precipitation in a short time [53, 54] (Figure 1).

Figure 1

Map of the study area affected by natural hazards (Map created based on [54, 55, 56, 57]

In West Greece, the Pindus mountain range is built of waterproof, Mesozoic limestone [52], and despite the sufficient amount of precipitation in this part of the country, the floods are rare phenomena.

The results of Diakakis et al. [54] showed that November is the month with the richest flood record. Thus, according to EM-DAT [49], areas of Central and Northern Greece suffered the flood in November 2017, where 23 people died and almost 1.000 buildings were damaged. Moreover, the flash floods hit these areas, respectively Attica, Thessaly and Central Macedonia regions in June 2018. According to Copernicus, the European system for Earth monitoring, Avlonas in Attica recorded 156.2 mm of precipitation in 24 hours to 27 June and NeaMoudania in Chalkidiki, Central Macedonia region recorded 110.1 mm [57]. Many settlements with Greek populations, including Serbian tourists, suffered damage (numerous residential buildings and vehicles were flooded, roads and beaches destroyed, vegetation devastated etc.).

In terms of wildfire in Greece, the extremely high temperatures with extended dry period cause these extreme events with destructive consequences in its modern history [58, 59]. Climatic and weather conditions have a strong effect on wildfire occurrence, especially in the southern and eastern part of Greece, but also socioeconomic, LULC (land use–land cover), anthropogenic and human factors [59].

In July 2018, Attica region (near Athens) was struck by a destructive wildfire caused by very hot and dry weather conditions and strong winds (Figure 1). Fifteen fires began on 23 July, and the worst affected area were villages of Mati, Kokkino Limanaki and Neos Voutzas [60, 61]. According to EM-DAT, 126 people have died in the fast-moving fire, which places it in the deadliest fires since 1900. More than 1.500 homes and vehicles have been damaged or destroyed [60].

This study was inspired by the two mentioned major natural hazard events (floods and wildfires in 2018) and their possible effect on Serbian tourists’ behavior based on the perceived risk on those destinations.

3.2 Study sample

The study sample consists of 224 respondents residing in Serbia. There are more female (70.1%) than male (29.9%) respondents in the sample, while the average age of the sample is 29.96 (interval of 18-60 years, Std=8.341). Majority of respondents are highly educated (68.3%). In terms of their travel habits, they travel mainly once or several times a year, mostly individually (67.9%), while the most frequent travel companions are friends and partner. The main sociodemographic characteristics and travel habits of respondents can be seen in Table 1.

It is also important to mention that 58% of the respondents have spent their holiday in Greece in the last few years. Results also show that majority of respondents (94.2%) are familiar with the floods and fires that affected Greece in the summer of 2018. Respondents who are familiar with the fact that Greece was affected by natural hazards were asked to state if these events would affect their future choice of Greece as a tourist destination. The majority of them (64.9%) claim that it will not influence their future choice of destination, 30.8% said they would choose places that have not been affected while only 2.3% said they would not go to Greece because it was affected by natural hazards.

3.3 Research instruments and data collection

The questionnaire consisted of five parts. The first part included sociodemographic characteristics of respondents (gender, age, education) and their travel habits (frequency of traveling abroad, travel companions and organization of their travel).

In the second part, respondents were asked if they have spent their vacation in Greece in the last few years and if they are familiar with the natural hazards that happened in Greece this summer. In connection to this, respondents were asked to state if these events will affect their future choice of Greece as tourist destination (1 - It will not influence their future choice of destination, 2 - I would choose places that haven’t been affected, 3 - I will not go to Greece because it was affected by natural hazards).

The third part of the questionnaire measured personality traits of respondents. The personality of respondents was measured by MINI IPIP-6 (Extraversion, Neuroticism, Agreeableness, Conscientiousness, Openness to Experience, Honesty-Humility), a 24-item scale developed by Međedović and Bulut [40], plus two additional personality traits - Sensation seeking (8- item scale by Hoyle et al. [62]) and Novelty seeking (9 item scale based on McIntosh et al. [63]). In this part of the questionnaire, respondents were asked to express their level of agreement/disagreement (1-I totally disagree, 5 – I totally agree) with the offered statements.

The forth part of the survey measured respondents’ tourism worries on 8-item scale developed by Larsen et al. [7]). Finally, the last part of the questionnaire measured the tourist behavior based on the perceived risk on a 13-item scale developed by Thapa et al. [8]. These statements were also measured using a 5-point Likert scale (1-I totally disagree, 5 – I totally agree).

The respondents were gathered online by using Google Forms. The survey was conducted from July till September 2018. The survey was distributed via Facebook profiles of researchers and their colleagues, with snowball convenience sample technique. Moreover, the survey has been sent via email to more than 1200 customers of travel agencies (two travel agencies have sent emails to their customers with kind request to fill in the survey). The respondents were informed that the survey is anonymous and that their participation is voluntary.

The method applied for data analysis was the structural equation modeling (SEM), specifically, the path model done in EQS 6.1. Pre-testing of the relationships between variables has been done in SPSS 20. in order to assist the authors in creating some initial relations in the path model.

4.1 Descriptive statistics

Research variables (with their mean values and standard deviations) are presented in Table 2. Cronbach’s alpha coefficient for all variables/dimensions is above recommended .7, except Conscientiousness (α = .678), which is still considered acceptable [64, 65]. This means that the instruments used in the study are reliable.

4.2 Exploratory factor analysis (EFA) of tourist behavior based on the perceived risk scale

In their paper about wildfires and tourist behaviors in Florida, Thapa et al. [8] used cluster analysis to extract three types of tourists based on their tendency towards risky behavior. They have isolated three clusters: Cautious, Courageous and Conscious travelers. The current study firstly explored the dimensions of tourist behavior based on the perceived risk, assuming that they will coincide with typology of travelers created by Thapa and collegues [8]. The principal component exploratory factor analysis was performed. Representativeness was good (KMO = .775) and Bartlett’s sphericity test was significant (χ²(66) = 671.66, p < .001),which confirmed that the data is suitable for the analysis. Three dimensions were extracted with 55.12% of variance explained. Varimax rotation was applied since the extracted components were not correlated.

The EFA(Table 3) confirmed the typology fromthe original research, as three factors whose item structure coincides with the Thapa et al. [8] study were extracted: Cautious behavior, Courageous behavior and Conscious behavior. Cautious behavior characterizes tourists who avoid destinations affected by natural hazards and put safety and security first when choosing where to travel. Courageous behavior describes tourists who travel to a certain destination even in case it was affected by natural hazards or in case there is a high risk of It. Conscious behavior characterizes tourists who wouldtravel to a certain destination that has been affected by natural hazards only if they feel safe traveling there.

4.3 Results of the path model

In order to check the hypothesized relationships, the path model was conducted in EQS 6.1 software. To determine the fit of the model, the following indicators were used: Sattora–Bentler χ² (S-Bχ²), which should not be statistically significant and ratio χ2/df, which should be less than 2 [66], or less than 3 [67]; the square root of the mean squared errors of approximation (RMSEA) and standardized square root of the average of the square residuals (SRMR) which should be less than .08; Bentler Comparative Fit Index (CFI), which should be higher than .90 for adequate model fit [68].

The first model (Model 1, see Table 4), which included all hypothesized effects (see Hypothesis 1 to 9), did not show satisfactory fit indices, so it was necessary to make the appropriate changes. Firstly, the Wald test indicated that Sensation-seeking should be excluded from the model, as well as the following relations: direct effect of Novelty-seeking on Cautious and Conscious behavior, direct effects of Extraversion, Neuroticism, Conscientiousness and Openness on Courageous behavior and direct effects of Neuroticism, Honesty-Humility and Openness on Cautious behavior. After running the model, the fit indexes were still not satisfactory (Model 2, see Table 4). In addition, the Wald test suggested that the effects of Novelty-seeking, Agreeableness, Extraversion and Conscientiousness on Tourism worries, as well as the effect of Tourism worries on Conscious and Courageous behavior, should be excluded from the model. The Lagrange Multiplier Index (LM) suggested including correlations between different personality traits (see Figure 2). This model resulted in satisfactory model fit indexes (Model 3, see Table 4 and Figure 2).

Figure 2

Standardized factor loadings for the proposed model

Figure 2 represents the final model with satisfactory model fit. On the left hand side, the personality traits as independent variables are presented. The two-direction arrows connecting different personality traits are indicating significant correlation between them. On the right hand side, three dimensions of the tourist behavior based on the perceived risk are shown as dependents variables. In the middle of the graph, Tourism worries are shown as a mediator between personality traits and three dimensions of the tourist behavior based on the perceived risk. The one-direction arrows are showing regression/influence, with regression coefficients written at the arrow.

The model shows that Tourism worries variable mediates the relationship between certain personality traits such as Honesty-Humility, Neuroticism and Openness to experience on one side and Cautious behavior on another. On the other hand, there is a direct effect of Novelty-seeking on Courageous behavior, Extraversion and Agreeableness on Conscious behavior and Conscientiousness on Cautious behavior.