Encountered Space and Situated Lay-Knowledge: A Mixed Methods Approach

This research draws on the geographical concept of situated lay-knowledge to highlight how the formation of tourists’ attitudes to travel destinations challenges the theoretical foundation of the theory of planned behavior (TPB). It suggests that situated lay knowledge is dynamic as opposed to static, which is the accepted basis of TPB, and subsequently, proposes a “Situated Lay-Knowledge Travel Behavior Model” (SLKTB). The model was tested in a mixed methods approach where Chinese tourists, who knew little about Portugal, encountered Portuguese culture and cuisine in Macau. The overall results demonstrate that the formation of tourists’ attitudes about travel destinations is not preexisting or static but dynamic and created from their ongoing encounters.


Introduction
This study critically revisits the Theory of Planned Behavior (TPB) in the context of tourism and proposes structural changes to make it relevant to tourist behavior. The TPB, as the dominant attitude-behavioral model with roots in social research (Ajzen 1991(Ajzen , 2005Ajzen and Fishbein 2000), has been widely applied to predict tourist behavior (Huang and Hsu 2009;Y. H. Kim, Kim, and Goh 2011;Ryu and Jang 2006;Shen, Schüttemeyer, and Braun 2009;Lam and Hsu 2006;Sparks 2007). However, tourist behavior differs from the types of behavior that the TPB conventionally addresses, such as quitting smoking, voting, and improving study performance. The associated attitudes are preexisting in the sense that they are accumulated over the course of one's life and are eventually incorporated in one's value system as a form of common sense (Ajzen and Fishbein 2000). As such, the TPB fundamentally addresses behavior where associated attitudes are already established and internalized.
Tourists' attitudes, such as toward travel destinations, however, do not always exist beforehand. They can be vague or even nonexistent until being developed and shaped by relevant encounters (Ajzen and Fishbein 2000). For example, tourists tend to have preexisting attitudes toward New York based on their exposure to American popular culture, and this will affect their intention to visit the city. However, their intention to visit less well known places can only be established once relevant encounters arise to inform their awareness, knowledge, and subsequently, attitudes toward visiting the destinations. It is essential to distinguish the two scenarios in relation to tourists' behavioral intentions. In the former, attitudes are established through previous encounters, whether direct or through exposure to the media or information gleaned from others, but in the latter they are nonexistent until established from relevant encounters. It is the latter that departs from the foundation of the TPB and should be separately examined, which is the primary concern of the present research.
The present study draws on a concept called "situated lay knowledge" (Crouch 2000) to underpin the dynamic nature of tourists' knowledge that moves from nonexistent to subsequently informing the development of attitudes that arise as tourists encounter destinations firsthand. The concept is grounded in geographical study and highlights the fact that tourism is one of the preeminent cases of discourse shaping knowledge of the world, effecting changes in opinions, values, and experiences (Crouch, Aronsson, and Wahlström 944241J TRXXX10.1177/0047287520944241Journal of Travel ResearchJi et al. research-article2020 2001. Tourists are living, active entities (Radley 1995;Scarles 2012) who are constantly figuring and refiguring identities, knowledge, values, and meanings from unexpected encounters and exploration. They can formulate new knowledge through various contexts, improvisational performances, intersubjective engagement, and tourists' poetic and emotional qualities (Crouch, Aronsson, and Wahlström 2001;Scarles 2009). Tourists are increasingly viewed as more fully humanistic figures; their associated meaningmaking capacity is given significance in refiguring the prediscursive, and revising prior beliefs (Crouch and Desforges 2003). As "situated lay knowledge" captures the dynamic nature of attitude development, the present study proposes incorporating it into the TPB as an antecedent to attitudes toward behavior (ATB), subjective norms (SN), and perceived behavioral control (PBC). A revised model called the "Situated Lay-Knowledge Travel Behavior Model" (SLKTB) is developed. The SLKTB also includes "expected emotions," which represents a state that accompanies tourists' situated lay knowledge.
To verify the framework, a mixed methods approach was adopted. A qualitative method was employed, in this case in-depth interviews aided by the Zaltman Metaphor Elicitation Technique (ZMET), to understand the content of situated lay knowledge. A survey followed to test the relationships in the SLKTB. The model was tested in the context of Chinese tourists' experiencing Portuguese heritage in Macau. Considering that they knew little about Portugal beforehand, their developed attitudes could be largely attributed to their encounters in Macau. The study was designed to understand how Chinese tourists' encounters with the Portuguese culture in Macau helped to create their lay knowledge about Portugal (Wong, McKercher, and Li 2016), and how lay knowledge informed other TPB constructs to impact tourists' intentions to visit Portugal. Notably, instead of asking tourists general questions about lay knowledge of Portugal, the angle of dining in fine Portuguese restaurants was employed, on the basis that dining is a concentrated presentation of culture and more importantly the most vernacular cultural expression; hence, it is more engaging and reflexive than high culture to ordinary tourists. Cultural presentation occurs not only in relation to the food consumed but also to decoration and artifacts in the dining space. Restaurant dining can create lay knowledge independently and also provides easy access to any lay knowledge developed at the destination through encounters occurring outside the restaurants (Everett 2008).
The study contributes to improving the efficacy of applying the TPB in tourism research (Ajzen 2011(Ajzen , 2015. It questions the theoretical basis of the factors underlying tourists' behavioral intentions and proposes a theoretical remedy. This study also pioneers the use of a geographical concept in tourist behavior modeling and presents a brand-new multidisciplinary perspective on this subject. Practical implications are also provided.

Literature Review
The TPB was developed from the Theory of Reasoned Action (TRA), by adding perceived behavioral control (PBC). The TPB postulates three conceptually independent determinants of behavioral intention (seen as the best predictor of actual behavior): attitudes toward behavior (ATB), subjective norms (SN), and perceived behavioral control (PBC). The TPB seeks to establish the relative importance of these in determining behavioral intentions . Underpinning ATB, SN, and PBC are salient information or beliefs relevant to the concerned behavior. Belief has been defined as the estimated, subjective probability that an object or action has certain attributes (Fishbein and Ajzen 1975). "Beliefs" about an object/action may be formed by associating with the object's attributes as a result of direct observation or inference processes, or is formed by acquiring information from external sources such as friends, television, newspapers, and books. Beliefs are categorized into behavioral beliefs (BB), normative beliefs (NB), and control beliefs (CB), affecting, correspondingly, ATB, SN, and PBC. The TPB overall is an attitude-based explanation of behavior. "Attitude" is defined as an evaluative response toward an object or behavior. An attitude is stored in one's memory and, once learned, it is triggered by the presence of the relevant objects or behavior (Fazio 1995). However, as will be shown, this attitude-based approach fails to consider some salient features of tourist behavior.

Encountered Space and Lay Knowledge
Despite being one of the most popular predictive behavior models (Fu et al. 2010;Xiao et al. 2011), the TPB is not free from criticism. Critiques often focus on the accuracy of the constructs' measurements, such as whether using PBC or self-efficacy, perceived control or perceived difficulty, selfprediction or behavioral intention is more appropriate, or on the efficacy of SN (Armitage and Conner 2001). A more challenging critique comes from information-processing theory (Hovland, Janis, and Kelley 1953). It is argued that attitudes are static and established prior to decision-making behavior. Once established, they tend to remain relatively unchanged even when new information is available (that can act as persuasive communication) (Hovland, Janis, and Kelley 1953). This challenges the idea that the TPB will work well in studying behaviors based on common-sense principles (Ajzen and Fishbein 2000), such as weight loss, smoking behavior, and political views. These associated attitudes are developed, institutionalized, and embedded in one's everyday life, prior to making behavioral changes. As Perugini and Bagozzi (2001) commented, the TPB is essentially recognized as relevant to "fixed" behavior, irrespective of contextual changes. Responding to the critique, Ajzen and Fishbein (2000) acknowledged the emergent quality of attitudes by suggesting that attitudes "develop in the course of acquiring information about an object, and keep evolving as existing beliefs change and new beliefs are formed" (Ajzen and Fishbein 2000). However, the structure of the TPB does not reflect this changing nature of attitudes. Tourists' attitudes may not be established until firsthand experience of the relevant tourism context informs them and here lies the problem. The TPB has been widely applied in tourism research (Huang and Hsu 2009;Y. H. Kim, Kim, and Goh 2011;Ryu and Jang 2006;Shen, Schüttemeyer, and Braun 2009;Lam and Hsu 2006), despite the changing nature of attitudes, which has been overlooked or, worse, ignored.
Tourists' attitude formation based on their travel experience depends on two variables, their "encountered space" and the subsequent development of "lay knowledge" (Crouch 2000). Encounter is defined as a process in which the subject actively plays an imaginative, reflexive role, not detached but semi-attached, socialized, crowded with contexts (Crouch, Aronsson, and Wahlström 2001). Encountered space offers important resources in the process and the material and metaphorical content are two constituents (Crouch, Aronsson, and Wahlström 2001). The material includes the artifacts of the surrounding world; metaphors are constituted in designed spaces, in the way that wider culture presents and represents leisure, and in imaginative symbolism constructed by individuals (Crouch, Aronsson, and Wahlström 2001). Space can also be viewed from close up and far off and the boundaries between them are blurred (Crouch 2000). The close-up space may be where we sit, shuffle to view an event, mingle among people, share a story, move through. The far-off view may be seen through a window: a space reached only in vision.
Tourist lay knowledge is tourists' own account or comprehension of the encountered space (Crouch 2000). It is "lay" because it reflects what is true and meaningful to tourists themselves, based on their own tourism experience in their encounters. The content of the lay knowledge reflects how people negotiate social and cultural relations, their identities, and the practical contexts in which they conduct their lives. Such flat ontology enables in-depth analysis of performed tourism consumption or production practices, and unravels the embeddedness of tourism practices (Lamers, Duim, and Spaargaren 2017). Ekdale and Tuwei (2016) discovered that the lay knowledge developed by Global North tourists from their firsthand engagement with suffering in the Global South led them to construct a humanitarian self in addition to gaining knowledge of the local life. Ji, Li, and Hsu (2016) identified that the lay knowledge developed by Chinese tourists about Japan from their own trip involved revising their prior stereotypes that were heavily influenced by textbooks and political relations. Li (2000) identified that the lay knowledge of Canadian tourists developed from their package tour to China involved personal growth and development and transformation of a social being by establishing bonds with the "Other." Lay knowledge does not need to be life-changing; even the everyday life and the most mundane activities have potential for creativity, insights, and the unexpected (Edensor 2007). What is essential about lay knowledge is that when any new knowledge is developed, a cognitive change emerges. Such change can be from nonexistence to existence, from little to enriched, from biased toward revised. Tourists' space is often estranged from their familiar home environment; hence their prior knowledge tends to be vague and general, but it is enriched, revised, and rewritten after tourists' handson engagement with the space. Therefore, the development of any lay knowledge effects such change. It develops specific new values and attitudes that subsequently reconstitute prior values and attitudes (Crouch and Desforges 2003). The lay knowledge offers fresh insights on tourists' prediscursive attitudes and opens "potential disruption of the prefigured-the potential distinction between categorized provision and practice" (Crouch, Aronsson, and Wahlström 2001).
Furthermore, tourist lay knowledge is generated by the participation of their agentive and subjective intention in making sense of the countered space (Crouch 2000;Scarles 2012). Tourism involves undertaking activities in an unfamiliar environment (Sirakaya and Woodside 2005). This motivates tourists to be more deliberate in processing available information (Ajzen and Fishbein 2000). Attitudes developed in this way can be more elaborate than those developed largely on superficial cues under lower motivational states (Ajzen and Fishbein 2000). Engagement of the body (or "embodiment") also plays a part in the generation of lay knowledge. Embodiment is a process of experiencing, making sense, and/or knowing through practice as a sensual human subject in the world (Scarles 2009;Ji and King 2018). As Haldrup and Larsen (2006) stated, the construction of space is a heterogeneous process where static and mobile nonhumans, as well as embodied, sensuous humans, play their part. Embodiment features intersubjectivity (social space of being with others), expressivity (body performance and actions used for expressing identities), and poetics (emotions and imagination) (Crouch and Desforges 2003).
Given the above arguments, it is proposed that Hypothesis 1: Tourists' encountered space positively affects the development of their lay knowledge when tourists have little prior knowledge.
Please note that dimensions of encountered space and lay knowledge were deduced by study 1, a qualitative research. As shown later, the dimensions of encountered space include restaurant ambience, restaurant design, social factors, food characteristics, and space outside the restaurant, as shown in Figure 1.
Hypotheses 2 and 3 were informed by existing studies (Goh, Ritchie, and Wang 2017;Hsu and Huang 2012;Meng and Choi 2016;Seow et al. 2017;Quintal, Thomas, and Phau 2015), which have confirmed the antecedent impacts of beliefs on ATB, SN, and PBC. This study postulates that as tourists develop lay knowledge, they can form beliefs about a less-known destination, including the anticipated positive experience of visiting the destination (attitudes toward behavior), gaining support from important others (subjective norms) and confidence to overcome barriers during travel based on their current experience at the destination (perceived behavioral control). More specifically, we propose that:

Hypothesis 2a:
Tourists' lay knowledge about a lessknown destination that is developed from encountered space positively affects their attitude toward visiting the destination. Hypothesis 2b: Tourists' lay knowledge about a lessknown destination that is developed from encountered space positively affects perceived subjective norms related to visiting the destination. Hypothesis 2c: Tourists' lay knowledge about a lessknown destination that is developed from encountered space positively affects perceived behavioral control over visiting the destination. Hypothesis 3a: Tourists' attitudes toward a less-known destination affect their intention to visit the destination. Hypothesis 3b: Subjective norms that tourists receive toward visiting a less-known destination affect their intention to visit the destination. Hypothesis 3c: Tourists' perceived behavioral control over visiting a less-known destination affects their intention to visit the destination. Hypothesis 4: Tourists' lay knowledge about a less-known destination that is developed from their encounters positively affects their behavioral intention.  Kim and Fesenmaier 2015). Despite different research traditions being applied, past researchers have commonly followed the cognitive appraisal approach, with the premise that emotions are conscious, reflective, and developed from appraisal of events that are of personal relevance (Lewis and Haviland-Jones 2000). While cognitive appraisal theorists are keen on identifying antecedents to emotions (such as pleasantness, goal congruence, and perceived likelihood of a particular outcome leading to emotions of joy, positive surprise, and love) (Hosany et al. 2015), what is fundamental to eliciting emotions is the personal relevance of the events or encountered space (Lewis and Haviland-Jones 2000) and the individual's cognitive process brings forward that relevance (J. J. Kim and Fesenmaier 2015). Tourists' lay knowledge is developed as a result of the cognitive process, so it attaches personal relevance to the encountered space, which would otherwise be taken as irrelevant. Emotions should then follow, according to the cognitive appraisal approach, even though it is uncertain what specific emotions would result.

Anticipated Emotions and Tourist Behavior
Existing studies on emotions demonstrate the antecedent role of lay knowledge to tourist emotions even though they were designed for different research purposes. Ji, Li, and Hsu (2016) studied Chinese tourists visiting Japan who eventually developed attachment to the country after establishing their own account of Japanese people, although it significantly deviated from their preconceptions. Tucker (2009) elaborated on the emotions of shame and discomfort felt by white tourists to Turkey after making sense of the locals' way of life and their reaction to tourists. For tourists, lay knowledge is an important antecedent to emotions because it attaches meanings to the encountered space, either positive or negative, and the strength and connotation of the lay knowledge leads to the types of emotions to be stimulated. This perspective advocates the view that "the power for emotional engagement does not lie in the viewed object itself but in the drama and romance of the story of encounter with the object" (Robinson 2012, p. 29).
Apart from being stimulated from the appraisal of realistic events, emotions can also be anticipated based on an imaginal process. As with experienced emotions, the impacts of anticipated emotions on planned behavior have also been widely discussed (Walters, Sparks, and Herington 2012;Perugini and Bagozzi 2001;Goossens 2000;Ajzen 2011Ajzen , 2015. Anticipated emotions help individuals to anticipate and make plans to navigate an uncertain future by providing concrete, vivid images of oneself interacting with a service or product and experiencing the consequences of their use prior to purchase (Goossens 2000). Anticipated emotions can enact participation (Walters, Sparks, and Herington 2012) or mobilize relevant knowledge structures in memory (Goossens 1994). Planned behaviors then are based on achieving prospective positive consequences and avoiding negative ones. Armitage and Conner (2001) delineate the consecutive impacts of anticipated emotions on behavior readiness. They identify that once anticipated emotions are elicited through prospective behavior, they affect individuals' intentions to pursue goals. This then leads to goal achievement, which functions as the basis for a new set of appraisals and accompanying goal-outcome emotions. Isen (2000) explained that positive affect promotes enjoyment and increases the likelihood of engaging in activities that are enjoyable or expected to be fun, rather than activities that would cause harm. Furthermore, Sandberg and Conner (2008) found that the inclusion of anticipated affect, such as anticipated guilt, in the prediction equation accounted for an additional 7% of the variance in behavioral intentions and 1% of the variance in actual behavior.
This study postulates that anticipated emotions are established from tourists' lay knowledge of a destination, as a result of the active cognitive process, when they know very little beforehand. Tourists are then motivated to visit that destination because they anticipate an enjoyable experience that is congruent with, and potentially can enhance, the positive experience that they anticipate.

Hypothesis 5:
Tourists' lay knowledge about a less known destination that is developed from encountered space positively affects their anticipated emotions about visiting the destination.
Hypothesis 6: Anticipated emotions about visiting a lessknown destination positively affect tourists' intention of visiting the destination.

Portuguese Culture in Macau
As a result of 500 years of Portuguese rule, Macau has inherited Portuguese architecture and culture, including Portuguese cuisine. Since its return to mainland China in 1999 and the relaxation of visa entry policy, Macau has become an exotic and popular destination, enabling its major market, Chinese tourists (who account for more than 75% of the overall market), to experience Portugal and gain a glimpse of Europe from the Far East (Wong, McKercher, and Li 2016). Among all forms of cultural expressions, food has been regarded as the most vernacular and can be easily associated with and understood by tourists who are new to the culture (Yang et al. 2020). Restaurants with rich cultural decorations and social amenities provide further cues about a culture. Dining in Portuguese restaurants is one of the most popular tourist activities in Macau ) and therefore provides an important lens to reflect and recall any gained lay knowledge about Portugueseness that has been developed from the dining experience, as well as from outside restaurants during various encounters in Macau.
At the time of this writing, there are about 400 Portuguese restaurants in Macau available for tourists to gain insights into "Portugueseness." Some typical Portuguese dishes are shown in Figure 2A-D and restaurant environments in Figure 2E and F. The context of Chinese tourists was chosen because they tend to have relatively little prior knowledge about Portugal and its cuisine, as there are virtually no Portuguese restaurants in China, unlike other international cuisines such as French and Japanese. This enables the attribution of major attitudinal changes to tourists' encounters in Macau.
This study involves two stages. The first was a qualitative study identifying the contents of lay knowledge and any behavioral intentions developed. These informed the lay knowledge construct in the subsequent survey, which was conducted between May and September in 2015. The second survey, conducted in April and May 2016, tested the SLKTB and generalized the relationships to a wider population.

In-DEPTH INTERVIEW
Study 1 followed Charmaz's (2006) constructivist grounded theory. Grounded theory most simply can be described as a general methodology grounded in data that are gathered and analyzed in a systematic fashion (Corbin and Strauss 2015). Charmaz's development of grounded theory takes the middle ground between positivism and postmodernism; it "assumes the relativism of multiple social realities, recognizes the mutual creation of knowledge by the viewer and the viewed, and aims towards interpretive understanding of subject's meanings" (Charmaz 2003, p. 250). This position assumes that neither data nor theories are "discovered" but are constructed jointly by the researchers and participants through interactions and emerging analyses.
To identify attitudinal changes among Chinese tourists, in-depth interviews were conducted. The interview questions were informed by Zaltman Metaphor Elicitation Technique (ZMET) (Zaltman and Zaltman 2008). It combines the visual method and the laddering technique, effective in eliciting unconscious thoughts, metaphors, and feelings that are difficult to articulate. It has been used in marketing, tourism, and hospitality research to study user experiences. Following ZMET, interviewees were asked to prepare six to eight visual images that best describe what they thought of during dining at Portuguese restaurants. Sample pictures are shown in Figure 3. Typical ZMET steps were applied, including describing the images, missing images, sorting images, multisensory questions, metaphor elicitation, vignette, and collage. The missing images step asks respondents to describe any images that they have not yet found. Sorting images asks the respondents to group the images they provide into meaningful piles, which helps researchers to form initial themes. Construct elicitation is an application of Kelly's Triadic Sorting (KTS) and laddering techniques (Reynolds and Gutman 1988). KTS asks the respondents to identify how any two of the images are similar but different from the third; laddering established a means-end value chain eliciting respondents' personal cognitive structures behind buying decisions and their experiences. Metaphor elaboration seeks to understand respondents' unconscious thoughts by asking how respondents would alter the frame or contents of the images through means such as widening, replacement, or accentuation. Sensory images involve respondents using five senses (sight, hearing, taste, smell, and touch) to metaphorize the quality of their experience. In vignette, respondents are asked to summarize their overall experience in a short story. Collage asks respondents to create a summary image using all the images. Both collage and vignette are insightful in understanding the overall meaning respondents attach to their experience. Interview questions were semistructured and developed from the classic ZMET publications (Zaltman and Zaltman 2008). Details on the administration of ZMET, interview questions, images, and collage are included in the author's published article (Ji and King 2018). Interviewees were purposefully selected according to the following criteria (Patton 2015): (1) that they regarded eating Portuguese food as a way to understand the underlying culture; (2) that they claimed to have little prior knowledge about Portugal and Portuguese cuisine; and (3) that their dining happened within the last three months. Participants were recruited using snowball sampling, as this approach enables the researcher to interview familiar, trusted persons, hence greatly increasing participation (Hennink, Hutter, and Bailey 2020). This is essential to ensure a full application of ZMET. Initial qualified participants were identified from the author's social network, including her friends and friends' relatives who resided in Mainland China, and her graduated students who worked in tour agencies whose customers include Chinese tourists. Potential participants were contacted by virtual meetings one month before their trip to confirm the interview schedule in Macau. Each participant was asked at the interview end to refer the study to their contacts and to make contact with the researcher. To ensure diversity in the sample, the researcher also checked the demographics of the participants and their levels of engagement with Macau and with Portuguese cuisine. In total, thirty-six Mainland Chinese tourists were interviewed in Macau, at which point information saturation was achieved. Each interview lasted around 90 minutes.
Interviews were tape-recorded and sent promptly to an independent professional company for transcription. Each transcription was completed within 2 working days. This enabled simultaneous data analysis and data collection, implementing theoretical sampling-which refers to "the process of data collection for generating theory whereby the analyst jointly collects, codes and analyzes his data and decides what data to collect next and where to find them" (Glaser and Strauss 1964). The transcription subsequently provided a vast amount of unstructured data. While constructivist grounded theory empowers interview respondents, eventual power undoubtedly rests with the researchers as they interpret the account of the encounter. Reflecting on the risks (Finlay and Gough 2003), the first two transcripts were independently coded by two researchers. Coding was described by Charmaz (2006, p. 43) as "the process of naming segments of data with a label that simultaneously categorizes, summarizes and accounts for each piece of data." After each one, the codes were shared; discrepancies were discussed until intercoder agreement was achieved. A codebook with description of the codes was gradually developed and the same codes were applied to all the transcripts to ensure consistency.
The coding procedure started with initial coding (or opening coding), applied line by line, paragraph by paragraph. Following Charmaz (2006), gerunds were used so that the researcher could focus on process and action. These initial codes were treated as provisional and open to modification and refinement to improve their fit with the data through constant back-and-forth comparison between codes and codes, codes and data. This method led to sorting and clustering of initial codes, which in turn resulted in revising existing codes as well as construction of new, more elaborated ones by merging or combining identical or similar initial codes. Focused coding (or selective coding) was then applied to capture the most significant or frequently mentioned initial codes; these are more directed, selective, and conceptual than the initial codes (Charmaz 2006). (The appendix shows examples of open coding and selective coding.) Having the focused codes enabled the researchers to raise them up to tentative conceptual categories, giving these categories conceptual definitions and assessing relationships between them. When progressing to theoretical coding, the researchers analyzed how categories and codes constructed from data might relate to each other as hypotheses to be integrated into a theory. Theoretical coding is aided by a logic process called "abduction" . Abduction requires one's reasoning to constantly move back and forth between the data and preexisting (as well as developing) knowledge or theories and to make comparisons and interpretations in the search for patterns and the best possible explanations (Thornberg, Perhamus, and Charmaz 2014). Meanwhile, data collection and data analysis was paralleled with memo writing, writing up ideas about codes and their relationships as occur to the researcher during the process. Figure 4 summarizes tourists' encounters in Macau, which informed the development of their lay knowledge about Portugal. The encounters took place in two spaces (1) restaurant-related encounters involving food and drink characteristics, social contact or space, amenities and design, and (2) experiences outside the restaurants and in Macau (i.e., "enroute travel experience" and "prior knowledge" about Portugal as shown in Figure 4). These dimensions informed "encountered space" in hypothesis 1, and the feature of the restaurant space is included in the visuals published in Ji and King (2018). The developed lay knowledge about Portugal revolves around the following themes: food culture, people, lifestyle, landscape, and tourists' future wonders. They inform "lay knowledge" in hypothesis 1. Detailed data analysis are published in Ji and King (2018). A synopsis of the research findings is presented below.

Results
To illustrate interviewees' formation of lay knowledge regarding Portuguese people, for example, the traditional Fado background music sung by a melancholy yet refined female voice stimulated interviewees to imagine Portuguese women to be elegant and sophisticated. In another case, through the reactions of Portuguese diners at the restaurant toward Portuguese television football matches, Portuguese people were considered to be passionate.
The Portuguese landscape was perceived to be filled with vineyards or as "a Western countryside." The cultural cues responsible included the wine that tourists observed other patrons drinking. The popularity of poultry as an ingredient led to the perception of Portugal as a humble country connected to the earth, while the prevalence and light flavor of olive oil elicited the natural scent of vegetables that drew one closer to nature and the land. The interviewees' encounters in Macau were recalled when they saw similar cultural representations in the restaurant, including Portuguese churches, architecture, street layout, and color schemes of the buildings. The triangulation enhances tourists' belief of an authentic Portugal. One interviewee claimed, "Houses in Portugal should look similar to those in Macau which have white walls and orange roofs" (Shao Zhang, male).
The encountered space romanticized Portuguese rule in Macau: "Portugal should be a supreme country . . . they navigated to conquer others. Thanks to Bacalhau (dried and salted codfish) helped the Portuguese sailors survive the long harsh sailing to Asia" (Hong Fan, female). Rather than condemning the inhuman Portuguese colonizers, one interviewee preferred to sympathize with the sailors who endured heavy casualties because of the unpredictable weather at sea and the shortage of food.
Ultimately, the knowledge informed by the encountered space formed tourists' intentions to visit Portugal. Several interviewees analogized the overall dining experiences as seeing behind a closed door where countless wonders, mysteries, and the unknown were waiting to be explored. Another said, "Many small European cities are unknown to Chinese tourists. But only in these small places can a tourist see the remarkable difference and picturesque locality. How beautiful it would be if I could make a visit" (Si Wu, female).
The findings illustrate an aesthetic and multisensory process of experiencing, making sense, and knowing of Portugueseness, in this case, through dining in Portuguese restaurants (Everett 2008) as well as sightseeing outside restaurants in Macau. Knowing is drawn from the markers of the cuisine, including ingredients, presentation, cooking philosophy and method, and eating etiquette. The environment within the restaurant supplies further cultural references, including the material and the symbolic. The dining space activates various senses and mobilizes a more physical, emotive corporeal engagement with the food and the culture (Brillat-Savarin 1994), and it is this that generates the multidimensional lay knowledge associated with place (Tuan 1977).
The findings further reinforce the concept that space is a stage of cultural cues that repeatedly mark the boundaries of significance and value (Neumann 1988). Various players on the stage constitute a support network that facilitates, guides, and organizes performances according to normative conventions and industry imperatives (Edensor 2001). While they may serve to reinforce a particular "given culture," tourists can be more ambivalent and contradictory and challenge the fixed order of things (Edensor 2001). The tourist space can become an "ordered disorder" that encourages a "controlled de-control of the emotions" (Featherstone 2007). The result also highlights that space can be distinguished between the immediate surrounding space that people touch and meet (i.e., restaurant); and a distant space that is only reached in vision and sound (i.e., Macau and Portugal) (Crouch 2000). The two spaces are interactive, mediated by points of reference, and they forge a kaleidoscope of events understood in numerous different ways (Crouch 2000).

Study 2
Survey Sample and Procedure Figure 4 was used to inform the development of the questionnaire used in study 2. The quantitative study involved a personally administered survey. Six experienced research assistants were hired to collect the data in Macau. A systematic sampling method with a skip interval of three was used to survey Chinese tourists who had finished dining and walked out of the 25 most popular Portuguese restaurants in Macau. The restaurants were initially identified by Web rankings according to their popularity. The questionnaire was first developed in English that then underwent a twoway blind backtranslation procedure into Mandarin. It was subsequently pilot tested by 12 Mainland Chinese from different backgrounds to increase readability and optimize face validity. To reduce common method bias, different scaling techniques (e.g., Likert-type scale and semantic differential scale) and reverse-coded items were used as the literature recommends (Podsakoff et al. 2003).

Measures
The first part of the questionnaire comprised multiple scales (discussed below), with each scale item measured using a 7-point Likert-type scale anchored from 1 (strongly disagree) to 7 (strongly agree) unless otherwise specified. The second part comprised respondents' sociodemographic information, such as age, gender, prior general knowledge about Portugal and Europe, number of times they had dined in Portuguese restaurants on this trip, and money spent in the restaurants.
A total of 356 completed surveys were obtained, with a response rate of 40%. Half of the respondents were female. Overall, 43.5% were between the age of 26 and 35 years, and 26.1% between the age of 36 and 45 years. With respect to origin, 81.5% were from 24 Chinese provinces and the rest from Taiwan. Most of them (83.1%) had traveled to Macau with at least one companion. The vast majority had never been to Europe (78.4%) or Portugal, and most also claimed to have little prior knowledge about Portugal (90.4%). Most respondents had been to a Portuguese restaurant more than once during their stay, with a mean of 2.70 and a mode of 2.00 times. On average, they spent MOP$245.29 (or Euro 27.40) per person in the restaurant.
Anticipated emotion. Anticipated emotion was measured by a six-item scale informed by Perugini and Bagozzi (2001) and Sparks (2007). Respondents were asked, "How intense do [they] expect to feel about visiting Portugal?" relating to the emotions of excitement, fear, happy, boring, fun, and nervous. The last scale item was later removed because of low reliability. The remaining fiveitem measure had a Cronbach's alpha of 0.84, indicating adequate scale consistency.
Attitudes toward behavior. Attitudes toward behavior was informed by Lam and Hsu (2006) and Huang and Hsu (2009), "I think travelling to Portugal would be" along four 7-point semantic differential scales ranging from satisfactory to unsatisfactory, worthwhile to worthless, unfavourable to favourable, and negative to positive. The scale demonstrated good reliability, with a Cronbach's alpha = .92.
Subjective norms. Subjective norms was measured using a three-item scale adapted from prior studies (Perugini and Bagozzi 2001;Lam and Hsu 2004;, "Most people I know would disapprove/approve of my visit to Portugal," "People who are important to me think I should/shouldn't visit Portugal" and "Most people I know would strongly agree/disagree to visit Portual." The scale demonstrated good reliability, with a Cronbach's alpha = .92. Perceived behavioral control. Perceived behavioral control was assessed by a 3-item scale adopted from (Perugini and Bagozzi 2001) with responses on a 7-point semantic differential scale. The questions concerned tourists' general control over traveling to Portugal based on difficult-easy, extremely unlikely-extremely likely, and no controll-full control. The scale had reasonable scale reliability with Cronbach's alpha = .89.
Travel intention. Travel intention was informed by Lam and Hsu (2004) and Huang and Hsu's (2009) study measuring the likelihood of traveling to Portugal in the future. Example questions include "I intend to visit Portugal in the next two years," "I intend to visit Portugal in the future," "I intend to include Portugal into my travel itinerary in the future if there is a chance." The scale was fairly reliable, with Cronbach's alpha = .87.
Encountered space and lay knowledge. Encountered space was operationalized as the identified spaces responsible for generating tourist lay knowledge about Portugal via dining in Portuguese restaurants. The contents of the two scales were informed by Study 1. The authors followed the scale development procedures of Churchill (1979) and transformed these qualitative results into scale items. First, the domain of encountered space in the restaurant was specified (Crouch 2000;Everett 2008). Second, results from the aforementioned qualitative inquiry were adopted to generate an initial set of items. Third, two experts in food tourism helped to refine the items. A total of 20 items were generated in the process. Fourth, a survey (i.e., study 2) was conducted to further refine the scale items. Fifth, a data set obtained from this survey was split in half to yield two subsamples based on the recommendation of Hair et al. (2006). The first subsample was used for exploratory factor analysis (EFA) using promax oblique rotation and the second was used for confirmatory factor analysis (CFA).
The EFA procedure produced a three-factor solution with 13 items regarding encountered space (i.e., seven items were removed because of low factor loadings and cross-loadings). Restaurant ambience and design was an 8-item factor that assessed the ambient conditions and design of a restaurant. Outside restaurant space was a 3-item factor that assessed the environment surrounding the restaurant. Restaurant social space was a 2-item factor that measured the social interactions between the subject and his or her travel companions. The 3-factor solution explained 69.90% of the variance of the scale (see Tables 1 and 2). The CFA procedure, which assessed the fit of the measurement model, indicated adequate model fit with comparative fit index (CFI) = .98, root mean square error of approximation (RMSEA) = .08, and standardized root mean square residual (SRMR) = .05. Convergent and discriminant validities were demonstrated with each retained item possessing a primary factor loading of .50 or above and a secondary loading of below .30. Average variance extracted (AVE) of each factor ≥.59 and the square of each pair of factors is less than the variance extracted of each factor. Both Cronbach's alpha and composite reliability indicated that all factors were fairly consistent, with values ≥.85. Nomological validity was demonstrated in the structural model presented below.
Tourist lay knowledge was operationalized as the knowledge developed by tourists following their exposure to Portuguese restaurants and other heritage sites during their travel. The scale was developed using the same procedure as encountered space detailed above. That is, we first identified the domain of the construct based on the literature (Crouch 2000;Crouch and Desforges 2003;Li 2000;Ji, Li, and Hsu 2016;Ji and King 2018;  Note: γ = eigenvalue. AVE = average variance extracted. Fit index: Comparative fit index = .98, root mean square error of approximation = .08, and standardized root mean square residual = .05. Total variance explained = 69.90%; Kaiser-Meyer-Olkin = .94; Bartlett's test of sphericity = 1636.86 (91) , p < .001.

2007
). Second, results from study 1 were adopted to generate an initial set of items. Third, two experts in the field of tourism assisted in refinement of the items. An initial set of items were generated in the process. Fourth, a survey (i.e., study 2) was conducted to further refine the scale items. Fifth, a data set obtained from this survey was split in half to produce two subsamples. The first subsample was utilized for exploratory factor analysis (EFA) using promax oblique rotation, while the second was utilized for confirmatory factor analysis (CFA). An initial set of 15 items was created based on the described qualitative inquiry (i.e., study 1) and 1 item was removed based on expert comments. EFA further reduced the scale into 12 items on three factors: (1) the beauty and friendliness of Portugal, (2) the informal and traditional characteristics of Portuguese cuisine and the Portuguese, and (3) Portuguese as a culture of passion (see Table 2). The threefactor solution explained 68.27% of the variance of the scale. The CFA procedure indicated adequate model fit, with CFI = .96, RMSEA = .08, and SRMR = .07. Convergent and discriminant validities were demonstrated, with each retained item possessing a primary factor loading of .50 or above and a secondary loading of below .30. AVE of each factor was ≥.50; and the square of each pair of factors was less than the average variance extracted for each factor. Both Cronbach's alpha and composite reliability indicated that all factors were fairly consistent with values ≥.78. Nomological validity was ensured in the structural model presented below.
Structural equation modeling (SEM) with LISREL 8.80 was conducted to examine the hypothesized relationships. For parsimony, a second-order measurement structure with three first-order factors was employed for both encountered space and lay knowledge (see Tables 1, 2, and 4). The proposed framework was assessed using two nested models for model comparisons as well as for testing the mediating effect of Attitude toward Behavior, Subjective Norms, Perceived Behavioral Control, and Anticipated Emotions. Demographic variables such as gender, age, and place of origin (mainland China vs. Taiwan) were entered into the model as control variables for behavioral intention. The variables Portuguese restaurant dining frequency and prior visit to Portugal were also controlled in a preliminary analysis. As they had no significant impact on behavioral intentions and other independent variables but reduced the model fit, they were not included for further analysis. Table 3 presents descriptive statistics and zero-order interfactor correlations among the variables of interest. Hypothesis 1 postulated a positive effect of encountered space on lay knowledge. Results from Model 1 reveal that the hypothesized relationship was significant (β = .63, p < .001) (see Table 4). The result suggests that a more favourable experience encountered in a Portuguese restaurant improved tourist lay knowledge toward the corresponding origin of the cuisine. Hypothesis 2 proposed a positive effect of lay knowledge toward behavior, subjective norm, and perceived behavioral control. Results from Model 1 indicate that the postulated relationships were significant (β ≥ .66, p < .001), suggesting that tourist lay knowledge has a strong impact on these three criterion measures. Hypothesis 3 replicated the proposed relationships leading from Attitudes Toward Behavior, Subjective Norm, and Perceived Behavioral Control to behavioral intention, as the theory of planned behavior posits. Results from Model 1 confirm these relationships (β ≥ .17, p < .05). Among the three predictors, SN exercised the highest impact on behavioral intention (β = .36, p < .001).

Results
Hypothesis 5 postulated a positive link between lay knowledge and anticipated emotions. Results indicate a positive and significant relationship between the two constructs (β=.85, p < .001), suggesting that tourists developed more favorable emotions toward the destination as they gained knowledge from their dining experiences. Hypothesis 6 proposed a positive impact of anticipated emotions on travel behavioral intention. Results from Model 1 indicate a significant relationship between the two constructs (β ≥ .17, p < .05), supporting the hypothesis.
Hypothesis 4 posited a direct and positive impact of lay knowledge on behavioral intention. In essence, hypotheses 2, 3, 4, 5, and 6 seek to examine the relationship between lay knowledge and behavioral intention through the mediating roles of attitude toward behavior, subjective norm, perceived behavioral control, and anticipated emotions. This study followed the mediation procedure from Baron and Kenny (1986) and tested whether the presence and absence of the three mediators would affect the lay knowledge−behavioral intention relationship. Hence, model 2 was created to illustrate the combined direct effects of lay knowledge and the three mediators on behavioral intention. Results from model 2 reveal that the relationship is negative and significant (β = -.63, p < .05), while results from Table 3 indicate that the relationship between these two constructs should be positive (r = .23, p < .001). The combined evidence suggests multicollinearity is a likely cause of this counterintuitive and inconsistent finding. Hence, hypothesis 4 is not supported. Results also suggest that attitude toward behavior, subjective norm, perceived behavioral control, and anticipated emotions fully mediate the relationship between lay knowledge and behavioral intention. The indirect effect of lay knowledge on behavioral intention was tested and the effect was significant (b=.07, p < .001). A Sobel test further confirmed the significant mediating effect (Sobel Z ≥ 5.67, p < .001). In addition, the indirect effect of encountered space on attitude toward behavior, subjective norm, perceived behavioral control, and anticipated emotions was also significant (b ≥ .05, p < .01; Sobel Z ≥ 2.38, p < .05). Overall, the proposed structural models have adequate fit, with CFI = .95, RMSEA = .08, and SRMR = .08. They predict the criterion variables fairly well, with R 2 ≥.40. Model 1 is a better model than model 2 as it has a better predictive power with a higher R 2 for behavioral intention; and model 1's Akaike information criterion (AIC), which compares the quality of competing models, is smaller than that of model 2 (see Table 4).

Discussion and Conclusion
The TPB has been popularly applied in tourism to predict tourist behavior (e.g. Huang and Hsu 2009;Y. H. Kim, Kim, and Goh 2011;Shen, Schüttemeyer, and Braun 2009). A close review of the existing literature, however, reveals a lack of distinction between the types of behavior the TPB has traditionally addressed and tourist behavior. Researchers applied a static interpretation of "attitudes" to tourists and assumed attitudes were always in place prior to behavioral intentions. In fact, tourists do not always have prior attitudes about a destination until stimulated by a tourism context as shown in this case. Much literature accentuates the fact that tourists are active and capable of making sense of their encounters. They can create lay knowledge that develops their prior limited attitudes, and these affect their behavioral intentions.
Geography provides relevant perspectives to overcome this theoretical deficit, notably the discussion on tourist lay knowledge and encountered place, but tourist behavior research appears unaware of it. Furthermore, although some studies apply the framework to post-event analysis (e.g., Song et al. 2012), in part considering the contextual effects of tourism, they do not sublimate to the deemed theoretical level. The present study adopts the concepts of lay knowledge and anticipated emotions and develops the tourists Situated Lay-Knowledge Travel Behavior Model (SLKTB). Using a rigorous mixed methods research approach, the study first identified the contents of the encountered space and kinds of lay knowledge developed, and second, translated them into the SLKTB, where relationships were tested. The model advances the traditional TPB model in the following respects.
First, in study 1, the qualitative research uncovered the lay knowledge that tourists developed about Portugal even though they had little prior knowledge about it. The lay knowledge informed the tourists' attitudes about Portugal (including Portuguese culture, people, and landscape as well as the Portuguese cuisine) and their intention of visiting Portugal. The lay knowledge developed about Portugal is coherent and systematic. Various signs, cultural artifacts, and human interactions within the restaurant space (i.e., food and drink characteristics, restaurant amenities and design, and social space) as well as the space outside the restaurant (i.e., en route travel experience) are involved. It is the tourists' multisensory and multidimensional engagement (Scarles 2009) with the encountered space and their agentive meaning-making from external cues that have enriched their prior, limited, beliefs about Portugal. The attention on lay knowledge makes the SLKTB model different from the TPB, as it stresses the role of tourists' encountered space and the dynamic nature of attitude formation in predicting subsequent behavioral intentions. The SLKTB revises the TPB in that it is not the tourists' prior beliefs, which can be nonexistent, but their newly developed attitudes based on their actual encounters that impact their behavioral intentions. Furthermore, study 1 confirms that food represents a shift away from the visual repertories of consumption (Franklin and Crang 2001) to multisensory experiences, embodied engagement, and nonrepresentable knowledge generation (Everett 2008). Study 2 highlights that encountered space and lay knowledge contributes mostly to forming anticipated emotions toward behavioral intentions, compared with attitudes toward behavior, subjective norms and perceived behavioral control. Existing studies have established that emotions are significantly and directly impacted by environmental stimuli, such as in the "servicescape" or "service environment" framework (Lee, Lee, and Choi 2010;Song et al. 2012;Wong, Ji, and Liu 2018). This study advances the notion that the anticipated emotions are mediated by lay knowledge. This finding confirms the importance of lay knowledge as it attaches meanings and personal relevance to the encountered space and reveals any felt emotions (J. J. Kim and Fesenmaier 2015).
Second, the study observes that ATB, SN, and PBC are essential in predicting behavioral intention as supported by their full mediating role between lay knowledge and behavioral intentions. In other words, without them, tourist lay knowledge is not developed into expressed behavioral intentions. Consequently, our study suggests that although tourists develop lay knowledge from their actual encounters, their travel intentions are not isolated from ATB, SN, and PBC. Furthermore, the study identifies that SN and PBC, rather than ATB, are the leading factors in predicting tourist behavioral intention. Interestingly, regarded as the weakest predictor in the TPB, the efficacy of SN has long been questioned (Armitage and Conner 2001). Inconsistent findings on SN are also observed in tourism studies. For example, studies by Lam and Hsu (2004) and Ryu and Jang (2006) did not verify SN in predicting tourist revisit intention, whereas Lam and Hsu (2006) and Shen, Schüttemeyer, and Braun (2009) did. This inconsistency may be attributed to the characteristics of research contexts. In our study, Portugal is a difficult place to travel to for the Chinese, because of barriers such as language and lack of cultural associations. Our study seems to suggest that if a destination is not perceived as popular among important reference groups, SN can override other attitudinal factors. As such, depending on the popularity of the destination and difficulties of travel, SN and PBC seem to have more influence than ATB in predicting behavioral intentions. It is also worth mentioning that anticipated emotions play a mediating role between tourists' lay knowledge and behavioral intention, the same as ATB, SN, and PBC; hence, this factor warrants being included in the TPB.
Practically, the study informs destination management, highlighting the importance of tourists' actual travel encounters in forming subsequent behavioral intentions. Ensuring positive onsite experiences is just as important as the marketing effort that focuses on establishing prior-trip imaginings. However, destination-marketing should increasingly pay attention to providing information that can strengthen tourists' SN and PBC, especially where the destination is less known to the target tourists. Furthermore, this study identifies that Portuguese culture in Macau can potentially stimulate Chinese tourists to visit Portugal. Both destinations may consider jointly designing travel itineraries that contain Macau and Portugal. Macau can also serve as a marketing frontier for Portugal, to promote travel information to Chinese tourists. Another practical implication relates to promoting destinations (both the encountered [e.g., Macau] and the projected/resembled [e.g., Portugal] hosts) with cuisines. As the present study demonstrates the spillover effects of gourmet dining on destination image, destination management organizations can take advantage of the local cuisine to induce travel attention to the destination.
This study represents a unique attempt to apply geographical concepts in behavioral models and to measure them in a positivist fashion. The present study calls for wider research consideration of incorporating lay knowledge into existing models that study behavioral intentions as it emphasizes the changes in attitudes that are developed from tourists' actual encounters. As lay knowledge is antecedent to emotions that eventually lead to behavioral intentions, future research can examine what kind of lay knowledge predicts what kinds of emotions, particularly positive emotions. Other innovative methods to study the composition of lay knowledge may be considered. As the study uses the context of tourist-oriented restaurants, replication of the measurement scales is limited to the tourism context. As the predictive power of the constructs in the model may depend on the characteristics of the destination, future studies could test the moderating effect of destination characteristics.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material
Supplemental material for this article is available online.