People have perceived tourism as one of the major driving forces for regional development. Expositions, fairs, and various holiday events represent new forms of tourism. These events not only help cultivate local pride and foster stewardship of the region’s unique natural and cultural assets, but also cast potential economic impact to the hosting areas. For example, the 2010 Shanghai World Expo in China attracted over 73 million visitors and generated billions of dollars in economic impact to the city of Shanghai. The Expo also showcased the city’s soft power which enhanced its image among visitors. The growing importance of event based tourism produces a rich body of scholarly work. Such research work, particularly behavioral studies in event based tourism, can be classified into three broad categories based on the academic fields adopted: (1) the appealing features of touring events; (2) the psychology of tourists; and (3) the cognitive and perceptual factors that influence tourist attitude and behavior.

The first stream of research – the appealing features of touring events – has focused on such features as: (i) the attracting dimensions of a touring event and (ii) the factors influencing destination image formation. The second stream of research – the psychology of tourists – has concentrated on: (i) the demographics of tourists and (ii) motivations [1]. This stream of research, although not always consistent, demonstrates that people with low agreeableness, high loneliness and shyness, low self-esteem, and neuroticism are more inclined to be addicted to online games. The third stream of research – the cognitive and affective factors affecting tourist attitude and behavior – has been mainly concerned with value perceptions and emotions in predicting satisfaction, intentions, and behaviors.

Despite the rising research interests in event based tourism, studies that combine the above three streams of research remain sparse and warrant further research. Hence, the present study aims to fill this research gap by employing the environmental psychology to identify (1) important atmospherics in an exposition, (2) their influence on tourist emotions and image toward the hosting city, and (3) tourists emotions’ impact on city image and extra spending. Additionally, the study investigates if visitors’ gender and geographical location alter the above effects (Figure 1).

Figure 1: Theoretical framework.

The study broadens current understandings of event based tourism through providing insights into how exposition atmospherics alter visitors’ emotions, which in turn enhance their image toward the hosting city and elicit extra spending. Specifically, the findings provide event hosts with useful information to develop marketing strategies and promote image and increase economic return accordingly in planning events. From the tourist perspective, tourists benefit from the results by being more conscious about those cognitive and affective factors shaping their spending behaviors.

Framework

Mehrabian and Russell [2] propose an environmental psychology approach and argue that factors from the surroundings may alter an individual’s essential emotional responses, which fall into three domains, namely, pleasure (happy or sad), arousal (to feel stimulated or uninspired to take action), and dominance (ability to control a situation or be submissive). These emotional responses thus result in an individual’s approach-avoidance behaviors (i.e., either approaching the situation or avoiding the environment altogether). Scholars apply the framework in a variety of service contexts [3-6].

Nonetheless, researchers argue that several limitations impede the framework’s application to consumption-related emotion studies. Machleit and Eroglu [7] demonstrate a potential lack of internal consistency in the traditional Pleasure—Arousal—Dominance (PAD) measure. Prior research also avers that dominance does not significantly relate to behavior [8]. Particularly, the aforementioned conceptualization is deficient in interpreting consumer emotions and might result in the occurrence of neither pleasant nor unpleasant states [9]. Thus, several studies [10-13] utilize two independent dimensions, that is, positive and negative emotions, instead of a pleasure and arousal scheme in examining the relationship between the two types of emotions and behavioral outcomes. The present study concurs with this view in understanding exposition visitors’ emotional states in relation to atmospherics and their behavioral outcomes.

Touring event evaluations

Research identifies a number of evaluative factors in event based tourism that link to behavioral and marketing outcomes. For example, researchers classify four key attributes in assessing festivals, namely, generic features, specific entertainment features, information sources, and comfort amenities. Positive assessment of these elements leads to tourist satisfaction and repeat-visit intentions. Lee et al. [12] discover seven facets of “festivalscape”, a term originated from servicescape to portray festival evaluation. These festivalscape’s dimensions, namely, convenience, staff, information, program content, facility, souvenirs, and food, not only affect patron emotions but also mediate the impact of festivalscapes on patron satisfaction and loyalty. Likewise, Lee, Lee, and Young [12] propose five festival evaluative factors (i.e., informational service, program, souvenirs, food, and convenient facilities) that shape visitors’ value perceptions. In a similar vein of research, the following festival quality dimensions: festival program, informational service, festival product, convenient facility, and natural environment. Evaluation of these quality dimensions influence participants’ functional and emotional value perceptions, thereby result in different levels of satisfaction and behavioral intentions. Review of the literature reveals that the same attribute may have different names, or the same descriptor represents slightly different attributes across studies.

In addition to the above evaluative factors, which tourism scholars largely agree to be critical in touring event evaluation, the study introduces a crucial dimension of touring event evaluation: technology related factors, specific to expositions. Expositions by themselves mean to attract visitors via technologies and innovations. For example, one of the slogans for the 2010 Shanghai World Expo is “Technology makes the World Expo more brilliant”. Demonstrating technology achievements becomes a common theme in recent international expositions.

Festival emotions

As aforementioned, environmental psychologists maintain that factors from the environment affect consumer emotions. Consistent with those environmental psychology studies in services [8,13,14] and research in tourism [12], the study posits that the assessment of the evaluative factors leads to exposition visitors’ emotional responses. In particular, a favorable evaluation provokes visitors’ positive emotions, and soothes their negative emotions. For example, good services arouse visitors’ positive emotions whereas poor services result in negative emotions.

H₁: Visitors’ evaluation of the exposition relates positively to their positive emotions.

H₂: Visitors’ evaluation of the exposition relates negatively to their negative emotions.

City image and extra spending

Tourism researchers recognize the importance of visitor’s affective reaction and maintain that the emotional responses lead to a variety of behaviors such as approach behavior [15], spending levels [16], retail preference and choice [17], loyalty [6], and satisfaction [7]. When visitors experience positive emotions in a touring environment, they are more likely to adopt approach behavior; conversely, negative emotions are more likely to generate avoidance behavior. Among those behavioral outcomes, the exposition host particularly concerns if the hosting city’s image is enhanced and whether the visitors increase spending. Enhanced city image represents the hosting city’s soft power whereas increased spending brings forth the economic benefit.

Destination image refers to an individual’s mental representation knowledge or beliefs, feelings and overall perception of a particular destination. As such, the term “destination image” encompasses many facets that are complex, multiple, relativistic, and dynamic. In tourism research, it shows empirically that those perceptual/cognitive and affective evaluations, namely exposition evaluative factors and emotions in the present context, have a direct influence on the overall image of the destination.

H₃: Visitors’ evaluation of the exposition relates positively to their image of the city.

H₄: Visitors’ positive emotions relate positively to their image of the city.

H₅: Visitors’ negative emotions relate negatively to their image of the city.

While several tourism literature supports the role of destination image in affecting tourist satisfaction, and trip or service quality, few studies explore its direct effect on tourist’s extra spending. Since research affirms the influence of image on behavioral intentions, the study expects that enhanced city image leads to visitor’s extra spending.

H6: Visitors’ image of the city relates positively to their extra spending.

Data Collection: Taipei International Flora Exposition

Data for this study were collected during the Taipei International Flora Exposition, which was held for six months between November 2010 and April 2011 at Taipei, Taiwan. The Exposition is centered on the theme of “Flower, River and New Horizons”, with an exposition ground spread across 91.8 hectares and 14 individual themed pavilions, dedicated to showcase horticultural, science, or environmental protection technology achievements. The city of Taipei has a diverse retail offering and possesses various sociohistorical and religious heritage sites recognizing relevant ancient and current events. Taipei International Flora Exposition attracted nearly nine million tourists from all over the world [18].

Sample

Visitors attending the Taipei International Flora Exposition were randomly selected and approached by four interviewers. The interviewers first outlined the purpose of the study and invited them to participate in the research project once they demonstrated a willingness to join the study. Particularly, the study employed a dual-phase data collection approach. Informants filled out the questionnaire pertaining to atmospherics, emotions, city image, and demographical information in Phase 1 and that of extra spending via mails in Phase 2, which was distributed seven days later. This dual-phase approach allows the study to measure more accurately the true spending behavior. The approach also permits the respondents to answer questions in the appropriate temporal sequence, reducing potential common method bias [19], while providing for defensible temporal association. In order to increase the participating rate, each respondent received a bookstore certificate worth of NT$ 150 (approximately US$ 5.33) at the end of Phase 2. The study matched data from both phases, resulting in 729 usable questionnaire packets-a 28% overall response rate for the participants. Table 1 summarizes the respondents’ profile.

Table 1: Demographic profile.

Measures

The study recorded all responses on 7-point Likert-type scales anchored by 1 (strongly disagree) and 7 (strongly agree), unless otherwise noted. Due to its multidimensional aggregation, festival evaluation was measured based on a number of related research [12,17] generating a preliminary list of 34 items. In addition, the study evaluated positive emotions (5 items) and negative emotions (4 items) based on Beatty and Ferrell [20]. Finally, the study assessed city image and extra spending with 4 items and 3 items, respectively [21,22].

Along with other demographical questions, the aforementioned items were initially prepared in English; they were translated into Chinese by independent translators, and then back translated into English to ensure accuracy and follow appropriate guidelines [23]. The study pre-tested the questionnaire in a pilot study involving 45 festival visitors and 2 experts from related disciplines in consumer psychology and marketing to confirm the clarity of the questions and validity of the instrument. The result of this review suggests necessary modifications to the wording of some questions.

Factor analysis

The study performed factor analyses before testing the above hypotheses. Specifically, the study adopted the varimax procedure to rotate the original principal components analysis solution and pinpointed the underlying factors based on eigen values greater than one. All the items with factor loadings less than 0.4 were eliminated. As Table 2 depicts, three facets of festival evaluation criteria emerged from the factor analysis, namely, service, food, and technology demonstration. The results from Bartlett’s test of sphericity (approximated χ² = 15434.62, df = 351, p < 0.001) and Kaiser–Meyer–Olkin measure of sampling adequacy (KMO value = 0.97) also supported these factor dimensions. All the reliability coefficients for the data exceeded the minimum requirement of 0.70 recommended by Nunnally and Bernstein [24]. These three factors explained 65% of the variance of festival evaluation. Likewise, the analysis of emotions suggested two factors: positive and negative emotions, explaining 80% of the total variance (Table 3). The results from Bartlett’s test of sphericity (approximated χ² = 5331.31, df = 36, p< 0.001) and Kaiser–Meyer–Olkin measure of sampling adequacy (KMO value = 0.91) warranted these factor dimensions. Reliabilities of positive and negative emotions were acceptable with values of 0.92 and 0.94, respectively [24]. Finally, separate factor analyses of city image and extra spending pointed to one factor each, accounting for 80% and 84% of the total variance, respectively. Reliability coefficients were 0.91 and 0.91, respectively, indicating satisfactory reliability (Table 4).

Table 2: Factor analysis results for festival atmospherics.

Table 3: Factor analysis results for emotions, city image, and extra spending.

Table 4: Means, standard deviations, and correlations.

Structural model

The study estimated the structural model (Figure 2) using AMOS (version 7.0). Table 5 demonstrates maximum likelihood estimates for the various parameters. The estimation produced satisfactory model’s fit statistics, thereby providing a good basis for testing the hypothesized paths (χ² = 2051.57, df = 825, p < 0.05, NFI = 0.92, NNFI = 0.95, CFI = 0.95, IFI = 0.95, and RMSEA = 0.05). The proposed festival atmospheric cues explained 76% and 45% of the variance in positive and negative emotions, respectively. Together with emotions, these atmospheric cues explained 76% of the variance in city image. Additionally, all the direct and indirect effects in Fig. 2 yielded the explanatory power of 64% in predicting extra spending.

Figure 2: Structural model results.

Table 5: Standardized parameter estimates.

Hypothesis testing

As Figure 2 shows, eight paths were significant in the structural model. Specifically, hypotheses H1a, H2a, and H3a posited that services affect emotions and city image. As shown in Table 5, the services dimension had a significant effect on positive emotions (γ11 = 0.42, t-value = 7.73, p < 0.001), negative emotions (γ12 = −0.71, t-value = −8.66, p < 0.001), and city image (γ13 = 0.42, t-value = 6.51, p < 0.001). Thus, the results supported H1a, H2a, and H3a. As H1b hypothesized, programs significantly influenced positive emotions (γ21 = 0.40, t-value = 6.76, p < 0.001). In contrast, the results did not support H2b for predicting a negative relationship between programs and negative emotions (γ22 = 0.87, t-value = 1.02, p > 0.05), nor did the results sustain H3b which expected a positive relationship between programs and city image (γ23 = 0.11, t-value = 1.74, p > 0.05). As H1c predicted, technology demonstration had a significant impact on positive emotions (γ31 = 0.16, t-value = 2.81, p < 0.01). H2c, which predicted a negative relationship between technology demonstration and negative emotions, however, was not supported (γ32 = −0.05, t-value = −0.63, p > 0.05). Despite this unexpected result, the data sustained H3c which assumed a positive relationship between technology demonstration and city image (γ33 = 0.13, t-value = 2.34, p < 0.05).

H4 for linking positive emotions and city image was supported (β14 = 0.31, t-value = 5.12, p < 0.001). Unexpectedly, H5 for the relationship between negative emotions and city image was not statistically significant (β24 = 0.05, t-value = 1.73, p > 0.05). Finally, the results supported a positive relationship between city image and extra spending (β45 = 0.74, t-value = 19.95, p < 0.001). In summary, the results supported most hypotheses except H_2b, H_2c, H_3b, and H₅.

Indirect effects

The study followed bootstrap test for mediation to investigate the mediating roles played by emotions and city image. Specifically, if its confidence interval does not include zero, the indirect effect is significant and mediation is established. The study first examined whether festival atmospherics affect city image through visitors’ emotions. As shown in Table 5, the services dimension of festival atmospherics had an indirect positive effect on city image (0.09, 95% C.I. = [0.04, 0.15]). Likewise, programs dimension had a significant positive indirect effect on city image (0.07, 95% C.I. = [0.04, 0.11]), and technology demonstration had a significant positive indirect effect on city image (0.04, 95% C.I. = [0.02, 0.08]). Notably, these indirect effects were via emotion’s positive aspect (services: 0.10, 95% C.I. = [0.06, 0.15]; programs: 0.07, 95% C.I. = [0.04, 0.11]; technology demonstration: 0.04, 95% C.I. = [0.02, 0.08]), rather than through its negative aspect (services: −0.01, 95% C.I. = [−0.04, 0.03]; programs: 0.00, 95% C.I. = [−0.01, 0.01]; technology demonstration: 0.00, 95% C.I. = [−0.01, 0.01]). In regard to city image’s mediating role between festival atmospherics and extra spending, the results revealed that services (0.29, 95% C.I. = [0.23, 0.37]), programs (0.11, 95% C.I. = [0.05, 0.16]), and technology demonstration (0.11, 95% C.I. = [0.05, 0.17]) all had had a significant positive indirect effect on extra spending via city image.

Moderating effects of visitors’ gender and geographical location

Furthermore, the study conducts multi-sample analyses to examine whether the above tested effects in Figure 1 are statistically different for different visitors’ genders and geographical locations. In regard to gender influence, programs dimension had a stronger effect on city image for males than females (Male: 0.31, p < 0.01; Female: 0.02, p > 0.05; χ²_diff = 4.60, df = 1, p < 0.05). In contrast, technology demonstration had a stronger effect on city image for females than males (Male: −0.05, p > 0.05; Female: 0.23, p < 0.01; χ²_diff = 4.70, df = 1, p < 0.05). With respect to visitors’ geographical locations, services had a stronger effect on positive emotions for foreigners than domestics (Domestics: 0.34, p < 0.01; Foreigners: 0.61, p < 0.01; χ²_diff = 4.40, df = 1, p < 0.05). Likewise, both positive and negative emotions exerted stronger effects on city image for foreigners than domestics (PE: Domestics: 0.17, p < 0.01; Foreigners: 0.46, p < 0.001; χ²_diff = 4.30, df = 1, p < 0.05; NE: Domestics: −0.01, p > 0.05; Foreigners: 0.12, p < 0.01; χ²_diff = 4.80, df = 1, p < 0.05). The study further separated domestics into locals and non-locals and found that bad services engender more negative emotions for non-locals than locals (Locals: −0.43, p < 0.001; Non-locals: −0.85, p < 0.001; χ²_diff= 6.90, df = 1, p < 0.05). Poor technology demonstration, however, elicits more negative emotions for locals than non-locals (Locals: −0.50, p < 0.001; Non-locals: 0.17, p > 0.05; χ²_diff = 9.60, df = 1, p < 0.05). Finally, good programs elicit more positive emotions for locals than non-locals (Locals: 0.65, p < 0.001; Non-locals: 0.12, p > 0.05; χ²_diff = 12.80, df = 1, p < 0.05).

The present study examines the consistency of external stimuli and emotions in affecting impulse buying behavior between virtual and physical retail settings. In each setting, the study tests four salient retail store components (i.e., merchandise variety, service quality, atmospherics, and price) for their effects on shopper’s positive and negative emotions, which in turn affect shopper’s impulse buying behavior. The results demonstrate support for the main thesis of the targeted research and provide insights into the predictors of impulse purchases in both contexts. The findings are twofold, that is, the consistency in affecting emotions, and in engendering impulse buying behavior. First, in regard to each stimulus’s impact on emotions, the price dimension produces comparable influence and represents a delighter in both retail settings. Merchandise variety, service quality, and atmospherics, however, cause unparallel effects across retail environments in affecting emotions. Specifically, merchandise variety and service quality play a trivial role in the online context, whereas they represent a performance item, and a basic item, respectively in the offline context. Furthermore, atmospherics exemplify a performance item in online retailing, but signify a delighter in offline retailing. Atmospherics also engender a stronger effect on negative emotions in the online environment than in the traditional shopping environment. Second, the results indicate that both positive and negative emotions are crucial to eliciting consumer’s impulse buying behavior in each retail setting. Particularly, negative emotions have stronger impact on consumer’s impulse buying behavior in the online setting than in the offline setting.

Implications

Theoretically, the study enriches the literature on impulse buying by utilizing the environmental psychology approach as opposed to the more general models based on personality traits and cues [25] or retail store environment variables [4] in exploring buying on impulse. Particularly, the results coincide with the view that shopper’s emotional responses need to be separated into two dimensions (i.e., positive and negative) rather than an aggregate perspective when investigating their roles in impulse purchasing. Additionally, the study advances understanding of the consistency between online and offline retail contexts in how external stimuli affect consumer emotions, and thereby result in impulse purchases. Nonetheless, the surprising results reveal that better services do not lead to positive emotions in offline retailing. Likewise, poor atmospherics not necessarily produce negative emotions in the physical shopping environment. Furthermore, negative emotions generate a significant positive effect on impulse buying in the offline retail setting. The study conducts follow-up interviews with several of the respondents in an attempt to explain the above findings.

The interview results uncover the following possible explanations. First, 77% of the informants from the physical stores hold either bachelors or master’s degrees. Researchers argue that more educated people tend to demand better quality [26]. Consequently, higher service quality does not lead to positive emotions in offline retailing. Second, most offline shoppers (91%) reporting in this study are existing consumers of the store, who have visited the store a number of times. These consumers are generally accustomed to the atmospherics and accept the current setting, making them less sensitive to the environment in engendering negative responses. Third, the study expects both positive and negative influence of negative emotions, producing a marginal overall influence, which is negative, on impulse buying in the offline setting. On the contrary, the result indicates that offline shopper’s negative affective states affect impulse buying positively. Indeed, a number of offline survey participants attribute impulse purchases to their attempt to relieve (or escape) from stress and anxiety. This finding can also be partially explained by the general absence of accompanying people by the shoppers in the study [27]. As the survey results reveal, over 86% of the offline participants come to the store alone. Offline shoppers who visit the store with others are less willing to participate in the study, possibly due to the incurred waiting time for the accompanying people. Although beyond expectation, this inclination toward impulsive buying stemming from shopper’s negative emotions extends and enriches the current understanding of how negative mood drives impulse purchases.

Managerially, practitioners as well as academics that want to better understand online versus offline impulse buying can draw several insights from the present study. First, retailers can develop management priorities by classifying external stimuli into insignificant, basic, performance, and delighter dimensions which influence emotions in different ways. While basic items in any case denote a decisive competitive factor, retailers are encouraged to augment performance items and delighters as they have a greater influence on emotions (Table 3). Second, the discrepancy in expectation between online and offline in-store elements suggests several marketing opportunities for retailers to enhance consumer’s shopping experience in the multi-channel retailing. For example, marketers can integrate various electronic commerce features and improve visual aesthetics in the virtual store [28]. When confronted with the right type of stimuli, online consumers are prone to buying on impulse [29]. Likewise, retail store managers should strive to maintain competitive merchandise selection and service quality for offline consumers. The adoption of mobile shopping assistant (MSA) that facilitates the shopping activity is potentially useful [30]. Third, marketers should be cautious about the result that negative emotions, though still engendering impulse buying, may create negative impacts in other aspects such as customer satisfaction and loyalty. Finally, both online and offline retail consumers can benefit from the findings by understanding how they become susceptible to impulse buying. Compared to offline consumers, online consumers are more disposed to buy on impulse when experiencing negative mood states.

Limitations and Future Research

While the findings help broaden the understanding of consistency between the brick-and-mortar store and the click-and-mortar webstore in engendering impulse buying, certain limitations are of note. First, and perhaps the major limitation of the study is the breadth of the sample. Although the sample includes a broad swath of informants, the present study focuses solely on one retailer’s consumers. Hence, further controlled cross store and cultural research is necessary to establish whether the patterns of effects are globally generalizable [1,21,31,32]. Similarly, further research investigating how the length of patronage and presence of others moderate the proposed relationships is valuable. Second, the study examines the impacts of in-store elements from multiple sections (e.g., appliance, apparel, etc.) in a retail store as antecedents to emotions and impulse buying. Indeed, these in-store elements may cause different levels of cross-sectional influences on the affective responses and buying behavior [33]. Consequently, additional research to detect context-specific effects may be beneficial. Third, the model includes only external variables that affect emotional responses. Although producing favorable results in predictive power for impulse buying behavior over prior studies based on the environmental psychology approach, this assumption is by no means complete and represents another constraint of the present study. Indeed, shopper’s affective states can relate to other individual differences and situational factors as well [34]. Fourth, the concentration on the single and aggregate dependent variable (i.e., impulse buying behavior) constitutes one other limitation. As aforementioned, impulse buying is a complex conceptualization involving different meanings such as pure, reminder, suggestion, and planned impulse buying. Thus, further research efforts can be contributive by investigating various types of impulse buying and/or its broader scope of consequences (e.g., overspending). Apparently, a multitude of research questions deserve further investigation in an attempt to explore what drives consumer’s impulse buying behavior.