National Ilan University Graduate Institute of Management Master Thesis



National Ilan University

Graduate Institute of Management Master Thesis

Service value, satisfaction, and switching barriers on behavioural intentions of farm tourists – A Taiwanese experience

Advisor: Ming-Chih Tsai, Ph. D.

Student: Chang-Tse Chen

A Thesis Submitted in Fulfillment of the Requirements for the Degree of Master of Business Administration

July 2007



本研究的目的是建構一個預測遊客在農場消費旅遊服務的行為意向模 型,並以一個台灣農業農場為個案研究,一共有 226 個家庭使用面對面訪 談的方式被收集成為樣本。服務價值、滿意度和知覺移轉障礙這三個變數 的觀察變項,首先會用驗證性因素分析加以測試,接著估計整個結構模式。

結果發現這三個變數都有直接影響行為意向,其中以滿意度最為顯著,接 著是知覺移轉障礙和服務價值,而且,三個變數之間的交互關係,特別是 移轉障礙的干擾效果也被確認可以更好的解釋行為意向,並在檢驗多樣本 分析後發現第一次來到農場旅遊與重遊的遊客在行為意向上沒有顯著不 同,最後我們會根據因素負荷量分析加以討論並提出有效的行銷策略給農 場旅遊做為一個建議。

關鍵字: 行為意向;農場旅遊;結構方程式;知覺移轉障礙 ; 滿意度;




The purpose of this study is to construct a behavioural intention model to predict tourist’s intentions in regards to consuming farm tourism service. A Taiwanese agricultural farm is chosen as a case study. A sample of 226 family groups is collected using face-to-face interviews. Manifests associated with three antecedent factors, including service value, satisfaction and perceived switching barriers, are first tested using confirmatory factor analysis, followed by calibration of a structural equation model. Results suggest that all three variables have a direct effect on tourists’ behavioural intentions. Satisfaction appears to be the best predictor, followed by perceived switching barriers and service value. Interdependences between the three variables, notably the moderating effects of switching barriers, are also confirmed to better interpret tourist intentions. An examination of multi-group analysis indicates that the behavioural intentions of first-time and repeat tourists are not significantly different. Finally, based on a factor loading analysis, a number of effective marketing strategies for farm tourism are discussed and proposed.

Keywords: Behavioural intentions; Farm tourism; Perceived switching barrier;

Structural equation model; Satisfaction; Service value



論文寫作來到了這一個片刻,該是一個階段,首先要感謝宜蘭大學提 供一個良好環境讓我學習與投注心力經營這個系所的每一個人,尤其是處 事細心、鉅細靡遺的 謝鳳嬌阿姨,接著是各位專任與兼任老師的付出。

我的碩士路途得以完成,完全要歸功於我的指導老師 蔡明志老師,

他的細細教誨、誨人不倦的教師性格,不論在是學業或是生活上無一不受 到他的潛移默化,或許再多的形容詞,也無法完整說出他對我的幫助,我 只能說聲謝謝老師。

最後在口試審查階段承蒙 林豐政老師和 蔡明達老師的撥空指教,

不辭辛苦地審閱論文,並給予許多寶貴的意見,讓學生受益良多,使得本 論文能更趨完善,在此一併致上深深的感謝。

研究生 陳昌澤 謹識  宜蘭大學經營管理研究所 




Chapter 1 Introduction ... 1

1.1 Research background and motivations...1

1.2 Research objective...3

1.3 Research flow ...3

Chapter 2 Literature review ... 6

2.1 Service value ...6

2.2 Satisfaction ...10

2.3 Perceived switching barriers...14

2.4 Behavioural intentions...14

2.5 Farm tourism development in Taiwan ...17

Chapter 3 Research methodology ... 23

3.1 Conceptual framework ...23

3.2 Research hypotheses...24

3.2.1 Service value on behavioural intentions...24

3.2.2 Satisfaction on behavioural intentions...24

3.2.3 Perceived switching barriers on behavioural intentions...24

3.2.4 Interactions between independent variables...25

3.2.5 Control variable ...26

3.3 Measurement ...26

3.3.1 Service value ...26

3.3.2 Satisfaction ...27


3.3.3 Perceived switching barrier ...27

3.4.4 Behavioural intention ...28

3.4 Data analysis method...29

Chapter 4 Data collection and result analysis ... 33

4.1 Data collection...33

4.2 Confirmatory factor analysis ...34

4.3 Structural regression analysis ...38

4.4 Multi-group analysis...40

Chapter 5 Conclusions ... 43

5.1 Findings and managerial implications...43

5.2 Limitations and future research ...47

References ... 48

Appendix A. Measurement items for questionnaire ... 61



Table 2.1 Farm tourism elements...19

Table 2.2 Accommodation and activities provided by 61 large farms ...21

Table 3.1 Procedure for data analysis in the current study...29

Table 4.1 Traveler profile of respondents ...34

Table 4.2 Results of convergent validity tests ...36

Table 4.3 Scale reliability and CFA analysis results...36

Table 4.4 Results of discriminant validity analysis...38

Table 4.5 Invariance tests across first-time and repeat tourists...42



Figure 1.1 Research Flow... 5 Figure 2.1 A means-end model relating price, quality and value………..10 Figure 2.2 Evolution of farm tourism in Taiwan………...18 Figure 2.3 Distribution of large leisure farms in Taiwan and the study

case... 22 Figure 3.1 Overall model of tourist behavioural intentions ... 23 Figure 3.2 Confirmatory factor analysis ……... 31 Figure 4.1 Results of path analysis (Standardized parameter



Chapter 1 Introduction

1.1 Research background and motivations

Agriculture has experienced considerable changes over the past several years (Evans and Ilbery, 1989, 1992; Frater, 1982; Ilbery et al., 1998). An inability to generate sufficient revenue has, in many cases, led farmers to diversify from an agricultural base (Davies and Gilbert, 1992; Fleischer and Pizam, 1997; Rickard, 1983). Farm tourism is one recognizable activity that may assist in counteracting economic decline (Walford, 2001). In particular, in developing nations where people rapidly become rich but living conditions are broadly congested (e.g. far-east Asia region), farm tourism might be more likely to succeed. However, many farmers are still isolated due to a lack of knowledge, expertise, and training in the tourism field, as a farm tourism enterprise requires completely different skills as compared to an agricultural enterprise (Alexander and McKenna, 1998; Chang, 2003; Davies and Gilbert, 1992; OECD, 1994).

Farm tourism entrepreneurs may fail to understand market needs or fail to respond to the service expectations of their customers. Such failures, in addition to the aforementioned declining agricultural income, could have devastating moral and economic effects for farmers (Reichel et al., 2000).

A complete understanding of tourist behavioural intentions (BI) is basic yet necessary information for the successful implementation of farm tourism marketing practices. Customer BI is viewed as an important indicator of whether customers will remain with or defect from their service providers (Engel et al.,


1995; Kumar et al., 1999; Zeithaml et al., 1996). However, BI research on farm tourism thus far has been rare, and past studies have not developed or included any analytical models.

In general tourism research, a large number of studies have shown that service value and satisfaction are positively associated with tourist behavioural intentions (e.g. Petrick and Backman, 2002; Tam, 2000). However, the effects of these two factors and their interaction have not been examined in a farm tourism setting. Furthermore, competition is becoming more and more significant in farm tourism, which is immersed in a dynamic business environment in which tourists have changed and farms are competing with one another in increasingly aggressive ways (Chang, 2003; Dernoi, 1983; Evans et al., 1992; Reichel et al., 2000). Effects of perceived switching barriers on customer retention as well as the moderating role on service value and satisfaction have received increasing attention (Jones et al., 2000; Yang and Peterson, 2004). But these issues were not considered in regards to previous tourism research. No study to date has investigated these constructs in a single framework.

In an empirical analysis, we applied the developed model to large-scale leisure farms with better resource capabilities (rather than small family-run farms), because they are more likely to have the necessary resources required to afford and practice various forms of marketing. In Taiwan, there are a total of 1,102 leisure farms, of which 61 are categorized as large-scale farms. This designation is based on their accommodation facilities and range of leisure activities. These 61 large-scale farms have accounted for more than 52% of the


total market turnover in 2005. In this densely populated nation, metropolitan residents are the main customers of these farms. In addition to the pursuit of physical outdoor activities and enjoyment of the pleasant countryside, these consumers are attracted to these farms based on the social, cultural and educational events held there (Frater, 1983; Walford, 2001). Farm tourists in Taiwan mostly consist of family groups (approximately 60%) (The Council of Agriculture of Taiwan, 2003), which is similar to many other countries (Dernoi, 1983). Even though the BI of family groups is considered to be an important research topic (e.g. Howard and Madrigal, 1990), there is still a dearth of related studies. To examine the BI of family groups, we chose to focus on a large-scale farm situated in an agricultural county that is a major travel destination for families living in the greater Taipei Metropolitan area.

1.2 Research objective

The purpose of the study is to construct a BI model made up of the three antecedent factors of service value, satisfaction and perceived switching barriers, and their mutual associations in order to assist in predicting tourist’s intentions towards consuming farm tourism services.

1.3 Research flow

As shown in Figure 1.1, this study is presented in five chapters. Chapter 1 introduces the research background and motivations, the objective of the research, and the research flow. Chapter 2 reviews the existing literatures on, service value, satisfaction, perceived switching barriers, behavioural intentions,


and farm tourism development in Taiwan and the study case are introduced.

Chapter 3 describes the research methodology. A Structural Equation Modeling (SEM) is established to explore simultaneous and interactive effects of three constructs: service value, satisfaction and perceived switching barriers.

Subsequently, research hypotheses, measurement and data analysis method are detailed.In Chapter 4, the collected data is reviewed, and manifests associated with the constructs are first tested using confirmatory factor analysis, followed by calibration of the SEM and an examination of the multi-group analysis using LISREL 8.54 computer software. In Chapter 5, findings are summarised and managerial implications are discussed. Finally, study limitations are touched upon and suggestions for future research are presented.


Service value

Figure 1.1 Research flow Introduction

Literature review

Research methodology

Data collection and result analysis Satisfaction Switching


Behavioral intentions


Farm tourism


Chapter 2 Literature review

2.1 Service value

Service value is conceptualized as the consumers’ evaluation of the utility of perceived benefits and perceived sacrifices (Zeithaml, 1988). In very broad terms, consumers may cognitively integrate their perceptions of what they get (i.e., benefits) versus what they have to give up (i.e., sacrifices) in order to receive services. This broad representation however can change depending on the type of product because for many products, some consumers may seek good price deals while for others they seek more than monetary gains. This may be why Zeithaml’s (1988) study resulted in four meanings of value for consumers:

low price, whatever one wants in a product, quality customer receives versus price paid, and what the customer gets versus what he/she gives. Among these four definitions, the last definition has been the main interest for most perceived value studies in marketing (Petrick et al., 1999).

Analysis of perceived value literature in marketing reveals that two perspectives have been used to model consumers’ value perceptions: utilitarian and behavioural (Jayanti and Ghosh, 1996).

Utilitarian approach to perceived value. Monroe (1990) and Thaler (1985) argued that consumers’ value perceptions are the result of their comparisons among different price structures including advertised selling price, advertised reference price and internal reference price. Monroe (1990) suggested that sellers constantly introduce advertised selling prices (sale price) and advertised


reference prices (higher or regular prices) to influence buyers’ internal reference prices, which are formed after processing relevant information (Monroe, 1990).

Therefore, internal reference price helps buyers to form their price expectations and value the deals around them (Monroe, 1990). According to the utilitarian perspective, perceived value of a product is a combination of acquisition value and transaction value of that product. These two value structures are combined in a way that their subjective weights, which are placed by buyers, are constituted and a perceived value equation is reached:

PV= v1 (AV) + v2 (TV).


PV= Perceived value of a product AV= Acquisition value of a product TV= Transaction value of a product,

v1 and v2 = Subjective weights placed on AV and TV by buyers.

“Transaction value, or the merit of paying the actual price, was determined by comparing the buyers’ references to the actual price” whereas “acquisition value of the product is the perceived benefits of the product at the maximum price compared to the actual selling price” (Monroe, 1990, p. 75-76).

Acquisition value, according to Monroe (1990) is the maximum price (i.e., Pmax: the highest price that the buyer is ready to pay for the product) less the actual price of the product (p) while transaction value was the buyer’s reference


price (Pref) less the actual price of the product. Hence, PV= v1 (Pmax-p) + v2 (Pref-p)


AV= Pmax-p, TV= Pref-p

Grewal et al. (1998, p. 48) proposed that “perceived acquisition value is the buyer’s net gain (or trade-off) from acquiring the product or service” while

“perceived transaction value is the perception of psychological satisfaction or pleasure obtained from taking advantage of the financial terms of the price deal.” Therefore, acquisition value reflects a trade-off between perceived benefits and perceived sacrifice (Monroe, 1990) whereas transaction value incorporates the pleasure and satisfaction received from the financial gains from the transaction (Grewal et al., 1998).

Behavioural modelling of perceived value treats the perceived value construct as a more comprehensive construct and attempts to explain it with not only price variations but also with other factors (i.e., psychological antecedents of value perceptions). In her influential paper, Zeithaml (1988) proposed a model of perceived value that was considered an important starting point for many researchers (See Figure 2.1 for the model).



attributes High –level abstractions

Intrinsic attributes

Objective price

Intrinsic attributes


value Purchase

Perceived quality

Lower - level attributes Perceptions o flower –

level attributes

Higher-level attributes

Perceived sacrifice

Perceived nonmonetary

price Perceived monetary


Figure 2.1 A means-end model relating price, quality and value. Adapted from Zeithaml (1988)

Zeithaml (1988) proposed in her model that consumers reach quality perceptions from evaluations of product attributes. Further, consumers use their quality perceptions to form overall value judgments about the products.

Zeithaml (1988) suggested that formation of quality and value perceptions occur in a means-end way. Originally proposed to identify consumers’ product categorization processes, the means-end approach was used to show how means as objects or activities might be related to ends as desired end states or values (Gutman, 1982). From objects through values, three levels of abstraction represent the way consumers retain information in their memory (Zeithaml, 1988). Attributes represent the lowest level in the means-end hierarchy while quality and value judgments serve as the consequences of attributes. Quality and value judgments in return are used to achieve personal goals and values.


In farm tourism, benefits are largely the results of good service quality in both outcome and process domains. Sacrifices from the tourist’s perspective can be divided into two types: the price that tourists have to pay, and non-monetary costs such as time and effort spent (Gallarza and Saura, 2006). Within the identification of benefits and sacrifices, research classified service value to cover three categories: functional value, emotional value, and social value (Sheth et al., 1991; Sheth et al., 1999; Sweeney and Soutar, 2001). Functional value is defined as the perceived utility acquired from an alternative’s capacity for functional, utilitarian, or physical performance (Sheth et al., 1991). In other words, it is the quality of physical outcome of using service, which refers to how well a service serves its principal physical function consistently (Sheth et al., 1999). Sweeney and Soutar (2001) defined emotional value as the perceived utility derived from the feelings or affective states that a service generates. They maintain that services are associated with or facilitate the arousal of specific emotions or feelings; for example, pleasure, relax, enjoy, happiness (Sweeney and Soutar, 2001; Petrick, 2004; Lee et al., 2007; Sánchez et al., 2006). Finally, social value is the utility derived from the service’s ability to enhance social self-concept (Sweeney and Soutar, 2001). Consumers choose the services that convey an image congruent with the norms of their friends and associates, or that convey the social image they wish to project (Sheth et al., 1999).

2.2 Satisfaction

Satisfaction is broadly defined as the consumer’s response to the evaluation


of the perceived discrepancy between prior expectations and the actual product performance as perceived after its consumption (Ganesh et al., 2000; Lemon et al., 2002; Oliver, 1999). There has been general agreement in the consumer behaviour literature that consumer satisfaction is tied to an evaluation process, which entails a comparison of product performance and some sort of a standard in relation to this performance (Oliver, 1997). The prevailing thought suggests that consumers develop expectations and use them as standards to compare with perceived product performance. The result of this comparison is termed as disconfirmation that can be both positive and negative based on this comparison.

Positive disconfirmation is achieved if perceived performance exceeds expectations or negative disconfirmation results if performance falls short of expectations. Satisfaction or dissatisfaction with the experience or use of product then is the consumers’ disconfirmation process in which positive disconfirmation leads to satisfaction whereas negative disconfirmation leads to dissatisfaction (Oliver, 1997).

One of the pioneers of this perspective on consumer satisfaction is Oliver (1980) who suggested that consumers adapt to certain buying situations and pre-purchase expectations serve as the adaptation level. According to Oliver, consumers reach satisfaction or dissatisfaction (CS/D) after a cognitive comparison between the expectations and perceived product performance.

Despite its wide application in marketing literature, the disconfirmation paradigm has been criticized. Some of the issues raised with respect to the disconfirmation paradigm are: a. the formation of expectations and their relation


to CS/D (i.e., consumers may not engage in comparison of expectations and performance every time); b. the use of desires rather than expectations as comparison standards to perceived product performance; c. the model’s being only cognitive rather than both cognitive and affective; and, d. other antecedents might influence the formation of CS/D (Barsky, 1992).

The formation of expectations and their relation to CS/D: Barsky (1992) argued that it was not clear what sources people use to develop expectations.

According to Spreng, Mackenzie, and Olshavsky (1996), two views prevail in terms of how expectations are used to infer CS/D. One view suggests that consumers use expectations as the probability of the occurrence of an event (Bearden and Teal, 1983) while others argue that consumers use expectations of goodness or badness of an event in addition to likelihood of occurrence (Churchill and Surprenand, 1982; Oliver, 1980; Tse and Wilton, 1988). Spreng et al. (1996, p. 17) suggested, “ expectations are beliefs about the likelihood that a product is associated with certain attributes, benefits and outcomes.” In addition to predictive expectations, consumers might use ideal standards, same or similar products, market promises and industry norms as belief standards (Barsky, 1992).

Use of desires rather than expectations as comparison standard to performance perceptions: The criticism about the adequacy of expectations as comparison standards brought about the value percept model of CS/D (Westbrook and Reilly, 1983). This model proposed that the expectation performance model lacked consideration of consumers’ desires or values in the


CS/D process. Westbrook and Reilly (1983) argued that even though expectations are met, values might not be satisfied with positive disconfirmation.

Later, researchers proposed that both values or desires and expectations are necessary as comparison standards in CS/D models (Speng and Olshavsky, 1993;

Spreng et al., 1996).

Cognitive and affective perspectives on CS/D: Recent research on the disconfirmation paradigm suggests that a cognitive explanation of the CS/D concept lacks explanatory power in models (Oliver, 1993). Westbrook (1987) introduced two affect states that might influence CS/D: positive and negative affect. His analysis showed that two affects, joy and interest were more strongly associated with CS/D than other affect states such as anger, disgust and contempt. A more recent investigation of affect in CS/D studies was performed by Oliver (1993). Oliver hypothesized that attribute satisfaction and dissatisfaction with the products have both direct and indirect influences on CS/D (i.e., mediating influences through positive and negative affect). His analysis showed that an introduction of affect and an attribute-based satisfaction judgment to the disconfirmation model increases explained variance by 50%

from 35% to 85% (81% for the course example). Oliver used two samples, individuals enrolled in auto and basic marketing courses for his study. Woodruff and Gardial (1996) suggested that emotion based measures of customer satisfaction are needed to reach a fuller understanding of CS/D. According to Woodruff and Gardial customer satisfaction was enhanced or diminished by positive or negative emotions generated by product service consumption.


2.3 Perceived switching barriers

Switching barriers represent any factor that makes it more difficult or costly for consumers to change providers (Jones et al., 2000). In their empirical study they examined three types of switching barriers: strong interpersonal relationships (the strength of the personal bonds that may develop between the employees of a supplier and the customer), high switching costs (the customers perception of the time, money and effort associated with changing supplier) and attractiveness of alternatives, which refers to whether viable alternatives exist in the market. Ping (1993, 1997, 1999), following Johnson´s (1982) concept structural constraints, uses the term structural commitment as a measure of the extent to which as the customer has to remain in a relationship. Ping argues that structural commitment includes alternative attractiveness, investment in a relationship and switching cost. Fornell (1992), without proposing a formal definition of the concept, provides a list of factors that can constitute such barriers (i. e., if they are prevalent they will hinder customers to defect from a relationship): search costs, transaction costs, learning costs, loyal customer discounts, customer habit, emotional cost, cognitive effort and financial, social and psychological risk. Colgate et al. (2001), to consider good service failure recovery can lead to customers changing their mind about switching from their service provider. Hence service failure recovery is in included as an important switching barrier.

2.4 Behavioural intentions


Behavioural intentions is defined as the subjective probability that an individual will take a particular action (Fishbein and Ajzen, 1975).Consumers’

behavioural intentions have generally been measured by asking them the probability or likelihood of buying the same product again (Cronin et al., 2000;

Sweeney et al., 1999).

The behavioural intention comprises two principal forms: repurchase intentions and word-of-mouth communication in this study. The discussion concerning the two concepts is presented below:

Repurchase intentions. Since behavioural intentions are easier to measure than actual behaviour, there are numerous studies of repurchase intentions (Bolton et al. 2000). However, these studies must be interpreted with caution because behavioural intentions are subject to criticism since intentions do not always lead to actual behaviour (e.g. Gabler and Jones, 2000; Morwitz and Schmittlein, 1992).

Repurchase intentions are defined as “the individual’s judgment about buying again a designated service from the same company, taking into account his or her current situation and likely circumstances” (Hellier, 2003). From this definition, it is clear that repurchase behaviour occurs when customers purchase other products or services for the second or more times with the same company;

and the reason for purchasing again is mainly triggered by customer experience towards the products or services.

Theory suggests that increasing customer retention is a key act of the ability of a company to generate profits (Zeithaml et al., 1996). This is because


the longer consumers stay with a company, the more products or services they buy from the company and no excess marketing outlay to win new customers.

To retain customers, a company needs to improve its service quality, which in turn leads to high service value (Cronin et al., 2000). Thus, it is noticed that consumers are more likely to purchase again from the same company if they think that what they have received was worthier than what they have given up.

Word-of-mouth communication (WoM). WoM communication is defined as

“informal communications directed at other consumers about the ownership, usage, or characteristics of particular goods and services and/or their sellers”

(Westbrook, 1987). The reasons for customers doing WoM communications are because they want to ease a tension that the positive or negative experience produced, to reassure themselves in front of others, to gain support from others who share their opinions, to gain attention or to share the benefits of things enjoyed (Wirtz and Chew, 2002).

WoM has been identified in previous research as an important behaviour after consuming a product or service (e.g. Gremler et al., 2001; Wirtz and Chew, 2002). This is primarily because WoM communication provides face to face, often vivid information that is highly credible (Spreng et al., 1995). In addition, consumers frequently rely on informal and/or personal communications sources in making purchase decisions as opposed to more formal and/or organizational sources such as advertising campaigns (Bansal and Voyer, 2000). Furthermore, WoM has been widely acknowledged as an informal communication source between consumers that has great economic impact (Murray, 1991). Customers


who show up on the strength of a personal recommendation tend to be more profitable and stay with the company longer than customers who respond to conquest advertising, sales pitches, or price promotions (Reichheld, 1996). From the previous studies, it is perceived that WoM plays a key role in shaping consumers attitudes and behaviours (Harrison-Walker, 2001). Thus, WoM serves as enforcement to consumers to remain loyal to a service provider. WoM is more important and influential within a service context than strictly just product marketing scenarios, given their intangibility and higher associated risk (Mangold and Brockway, 1999). Compared to purchasers of goods, Murray (1991) found that service buyers have greater confidence in personal sources of information as well as a greater pre-purchase preference for personal information sources. A consumer may not understand a service fully before its consumption; he or she might seek WoM information from an experienced source (Bansal and Voyer, 2000). Therefore, WoM becomes especially important within the services purchase decision context.

2.5 Farm tourism development in Taiwan

In Taiwan, agricultural products now face extensive competition in the recently liberalized market of international agricultural trading since Taiwan joined the World Trade Organization (WTO) in 2002. Agricultural production values fell dramatically in 2004, and accounted for only 1.68% of the GDP, whereas the value of the service industry rose to 72.73% (National Statistics of Taiwan, 2004). In an attempt to soften the effects for farmers, the government


has actively tried to assist in diversifying agricultural operations. One of the most popular reactions was a turn to farm tourism in hopes of attracting dollars, generating jobs and supporting retail growth. The local tourism industry has taken off in recent years for two reasons: the per capita income has grown to more than $13,000 (Ministry of Economic Affairs, 2005), and a five-day workweek consisting of 40 hours has been officially implemented, which is half a day less than previously. Tourism receipts accounted for more than 5% of the GDP in 2005, which serves as another major contributing factor towards Taiwanese economic growth.

To assist the development of farm tourism, the various levels of government have largely relaxed regulations in terms of land-use and business operation, and provided substantial financial assistance. The farm tourism industry is experiencing a major upturn as a result. The number of leisure farms has in turn flourished, rising from 518 in 1999 to 1,102 in 2004 (see Figure 2.2).

The total output value is measured at $13.7 million USD, with a tourist

1974 1979 1984 1989 1994 1999 2004 Year

1 1 0 2

5 1 8

1 2 3 1 8 0

1 4 1 2 7 7 0

2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

Number of leisure farm

Figure 2.2 Evolution of farm tourism in Taiwan.


population of 8.5 million (The Council of Agriculture of Taiwan, 2005).

The farm tourism comprises two principal forms: non-accommodation and accommodation-related activities; some farms participate in both (Shaw and Williams, 1994). A simplistic list of farm tourism elements is provided by Clarke (1996a) (see Table 2.1), many are specifically used for tourism purposes.

Davies and Gilbert (1992) identified similar components, segmenting farm tourism into three distinct categories, viz. accommodation-based, activity-based, and day-visitor-based.

Table 2.1

Farm tourism elements - according to Clarke (1996a, b) Attractions - permanent Attractions - events Farm visitor centre Farm open days

Self-guided farm trails Guided walks

Farm museums Educational visits

Farm centre Demonstrations

Conservation areas Country parks

Access (rural) Activities

Stile/gate maintenance Horse-riding/trekking Footpaths/bridleways/tracks Fishing

Shooting/clay Boating

Accommodation Amenities

Bed and breakfast Restaurants

Self-catering Cafes/cream teas

Camping and caravanning Farm shops/roadside stalls

Bunkhouse barns Pick your own

Picnic sites

In this study, leisure farms can be categorized into two groups according to


the classification developed by Davies and Gilbert (1992): large-scale farms with accommodation and activity services, and small-scale farms. In Taiwan, 90% of the leisure farms are designated as small-scale farms with a land scale of less than 10 hectares and less than six staff members (The Council of Agriculture of Taiwan, 2005). These are family-run farms that only operate over a short season and generate a relatively low income. There are 61 leisure farms identified as large-scale farms with better resource capabilities. Although small in number, these large-scale farms accounted for more than 52% of the market turnover, and the annual growth rate was significant (approximately 20%). The 61 large-scale farms offer accommodation and various activity services on their farms as part of the overall package, in an attempt to meet the growing demand for the transition from farm to farm tourism. Based upon research into the farm activities offered by those operators advertising in the internet, we found the results given in Table 2.2, based on our survey. Through various activities it enables tourists to understand the farm production process (e.g. pick your own process) plus experience the excellence and the sometimes bitterness of agricultural life.


Table 2.2

Accommodation and activities provided by 61 large-scale farms Accommodation

Range of room numbers per farm Types of accommodation

Resort 26~126

Cottage 15~106


Percentage Sport and leisure

Visiting natural attractions 81%

Hiking 64%

Cycling 19%

Education, social and culture

Educational activities of natural ecology 64%

To experience rural life 60%

Agricultural festivals and cultural activities 18%

Food and beverages

Local cuisine 59%

Agriculture product retail 36%

Source: our survey

As indicated in Figure 2.3 below, most of the large-scale farms are agriculture-based (76%), whereas 21% are ranch-based, and 3% are fishery/forestry-based. In terms of geographical location, many farms are clustered together. This increases competition, as farms in the same cluster have similar offerings in terms of nature, agriculture and accessibility. The study cluster chosen for this study is situated in northern-east Taiwan. Surrounded by mountains and close to the sea, the study cluster is located in an area famous for its natural beauty, and is a very popular destination for a great number of Taipei metropolitan residents (i.e. 6.8 million) (National Police Agency, Ministry of The Interior, 2006), who account for one third of the whole Taiwanese


population. In the study cluster, we specified Tou-Chen leisure farm for our study case. According to our pre-study survey, it is the best farm to study the BI of family groups, as it hosted the greatest number of family customers within the study cluster. The Tou-Chen farm is situated on 120 hectares, and has 72 resort rooms as well as a broad range of sports, cultural, educational and social activities.

Great Taipei area Population: 6.8 million

Study cluster

Figure 2.3 Distribution of large leisure farms in Taiwan and the study case

Tou-Chen leisure farm

10 Km

Farm-based Agriculture Ranch Fishery Forestry Cluster



Chapter 3 Research methodology

3.1 Conceptual framework

In the established model, we used four latent variables: the three constructs of service value, satisfaction, and perceived switching barriers, as well as the dependent variable of behavioural intentions. The constructs used in other behavioural intentions (BI) models, such as service quality and sacrifices (costs), can be implied as part of the corresponding constructs to simplify the analytical structure (Brady et al., 2001; Oh, 1999). In accordance with the justifications of previous studies, seven hypotheses were identified to express the possible relationships between the four latent variables that are represented by 23 manifests, as indicated in Figure 3.1 they are illustrated as follows:

Service value

Satisfaction Perceived

switching barriers

H2 (+) H3 (+)

H1 (+)

H4 (+) H6 (+)

H5 (+)

H7: the difference in terms of BI between first-time and repeat tourists

Behavioural intentions




V1 V2 V3 V4 V5 V6 V7

S1 S2 S3 S4 S5 S6








Figure 3.1 Overall model of tourist behavioural intentions


3.2 Research hypotheses

3.2.1 Service value on behavioural intentions

Bolton (1998) believed that a consumer’s decision to maintain an existing relationship with a service provider is based on the value acquired. Many tourism researchers have found that customer perceptions of value directly and significantly influence their behavioural intentions (e.g. Lee et al., 2007; Petrick, 2004). Therefore, we posit the following alternative hypothesis:

H1: Service value has a positive influence on behavioural intentions.

3.2.2 Satisfaction on behavioural intentions

Past farm tourism research has suggested that satisfaction is an excellent predictor of repurchase intentions (Reichel et al., 2000). Evidence for the significant impact of satisfaction on behavioural intention comes from a wide variety of service industries including tourism (e.g. Baker and Crompton, 2000;

Gallarza and Saura, 2006). Based on these studies, we state our second alternative hypothesis:

H2: Satisfaction has a positive influence on behavioural intentions.

3.2.3 Perceived switching barriers on behavioural intentions

Academic evidence of businesses in competition indicates a positive relationship between perceived switching barriers and behavioural intentions (e.g. Balabanis et al., 2006; Burnham et al., 2003; Colgate et al., 2001). That is,


the greater the perceived switching barriers, the greater the chance the tourist will remain, which is indicated as alternative hypothesis H3:

H3: Perceived switching barriers have a positive influence on behavioural intentions.

3.2.4 Interactions between independent variables

In addition to the relationship between satisfaction and behavioural intentions, the service literature argues that customer satisfaction is the result of a customer’s perception of the value received (Hallowell, 1996). When customers perceive services to have a high degree of value, they feel satisfied with the service provider (Cronin et al., 2000; Eggert and Ulaga, 2002). As a result, we posit the following alternative hypothesis:

H4: Service value has a positive influence on satisfaction.

Further, researchers conclude that switching barriers exhibite a moderating effect on satisfaction for behavioural intentions (e.g. Fornell, 1992; Balabanis et al., 2006; Chatura and Jaideep, 2003). In other words, customer satisfaction is higher when facing higher perceived switching barriers, and vice versa. Besides, given the high correlation between perceived value and customer satisfaction, it may be assumed that a switching barrier may impose a similar impact on customer perceived value (Yang and Peterson, 2004). Thus, we propose two alternative hypotheses for examination:

H5: Perceived switching barriers have a positive influence on satisfaction.


H6: Perceived switching barriers have a positive influence on service value.

3.2.5 Control variable

Following the stating of our hypotheses, we pondered whether they hold across different tourist groups. To examine the BI consistency of farm tourists, we use a control variable of first-time and repeat tourists in the structural model.

Instructed by Um et al. (2006), null hypothesis H7 provides a relatively simple basis for segmenting tourists into two groups for efficient marketing practices:

H7: The structural coefficients posited in Figure 3.1 will additionally be invariant between first-time and repeat tourists.

In order to assess the magnitude of the role of the control variable, it is imperative that we first examine the overall fit of the model and thereby assess model invariance across the two groups with the hypothesis.

3.3 Measurement

3.3.1 Service value

After reviewing researches in Chapter 2, based on this classification, we identified seven manifests for tourist service value for analysis. Functional value included: “leisure farm greeters provide good service quality” (coded as V1);

“the leisure farm offered a well-organized tourism product” (V2); “the tourism product that the leisure farm offered was good value for the money” (V3);

“compared to the time I spent on this experience, I believe I received good value” (V ); and “the tourism service was easy to buy” (V4 5). Emotional value


consisted of “visiting the leisure farm made me feel relaxed” (V6). Finally, social value consisted of “I pursued some sort of companionship with the farm operators” (V7).

3.3.2 Satisfaction

In farm tourism, the tourist’s evaluate is based on a comparison of product performance and the tourist’s desires, need, motives and beliefs about the product’s attributes and the benefits to be derived from them. Thus, a broad range of service features are involved in improving customer satisfaction. Based mainly on the service evolutions indicated in past farm tourism research (e.g.

Busby and Rendle, 2000; Chang, 2003; Reichel et al., 2000), we summarised six satisfaction manifests for analysis including: accommodation facilities (coded as S1), local cuisine (S2); nature scenery and outdoor activities (S3); rural culture and life experience (S4); educational activities for children (S5); and finally, a safe and secure place (S6).

3.3.3 Perceived switching barrier

In accordance with this suggestion and in consideration of the specific case, we summarised the perceived switching barriers into four dimensions: tourist psychological factors, service failure recovery, tourist switching costs, and the attractiveness of alternatives. Psychological barriers mainly consist of two elements, including tourist habit (B1) and risk perception (B2) (Fornell, 1992).

Tourist habit is meant to fix customers’ decision-making process (Yu, 1990), and risk perception serves to elevate their perceived uncertainty cost to prevent


customers from changing their mind about switching from their service provider (Zeithaml, 1981). Service failure recovery (B3) is a strategic issue. It is important because mistakes and failures are unavoidable in service sectors.

Customer delight with service failure recovery will create positive word of mouth and higher repurchase intentions than is the case for dissatisfied, non-complaining customers (Andreassen, 2001). In the switching cost dimension, we identified three switching costs, including additional out-of-pocket cost and sunk costs (B4); additional travel time (B5); and additional physical and mental hassles (B6) required for a service switch (Jones et al., 2000; Jones et al., 2002). Finally, as competition is significant in farm tourism market, competitors frequently exercise promotion practices to try to gain market share. Attractiveness of alternatives (B7) has become inevitable as a component of the barrier (Patterson and Smith, 2003; Ping 1993, 1997, and 1999).

3.4.4 Behavioural intention

In terms of the manifests of the behavioural intentions, repurchase intention and word-of-mouth are often justified as appropriate indicators in many studies (e.g. Dawn et al., 2004; Oh, 1999). These indicators can be expressed in either a positive or negative manner (e.g. Morgan and Hunt, 1994; Ranaweera and Prabhu, 2003; Zeithaml et al., 1996). Thus, we identified three manifests including repurchase intention (I1), positive word-of-mouth (I2), and negative word-of-mouth (I3).


3.4 Data analysis method

Initial analysis of the data and the procedure for hypotheses testing are summarized in table 3.1.

Table 3.1

Procedure for data analysis

Stage 1. Analysis Purpose

Descriptive statistics analysis - Investigate sample characteristics - Assure overall data quality Stage 2.

Analysis of the measurement properties of the scales

- Ensure reliability and validity of the constructs (i.e. discriminant validity, convergent validity)

Stage 3.

Analysis of Structural Model - Test H , H , H , H , H and H1 2 3 4 5 6

Stage 4.

Analysis of Multi-group - Test H7

Stage 5.

Presentation of results - Discussion of findings

Two statistical programs were SPSS 13.0 and LISREL 8.54 utilized. Initial analysis of the data was performed using SPSS while confirmatory analysis was conducted by using the LISREL, structural equations modeling program. Details of each stage of data analysis are as follows.

Stage 1 of the data analysis:

In this stage of the data analysis, sample characteristics and variable descriptive were analyzed to detect data quality. One of the main purposes in this stage was to detect data to determine if important assumptions associated with further analysis hold. For example, because path analysis is a regression-based technique, outliers in the data must be detected to get unbiased



Stage 2 in the data analysis:

Before the analysis of the structural models, the reliability and validity of the constructs were tested. The confirmatory factor analysis models (CFA) were analyzed to ensure the validity of the constructs. Analysis of the measurement part of the structural models before that of the structural parts is commonplace in behavioural research, and was suggested by structural equation modeling theoreticians (Anderson and Gerbing, 1988). The purpose of testing CFA models was to reach acceptable levels of discriminant and convergent validity of the constructs.

CFA: Figure 3.2 show confirmatory factor analysis models tested before the analysis of the structural models. Wording of the measurement items used in the study are presented in Appendix A. The purpose of the analysis of CFA models was to ensure convergent and discriminant validity of the constructs. For convergent validity analysis, coefficients of indicators to their respective constructs were analyzed.


S1 S2 S3 S4 S5 S6 B1 B2 B3 B4 B5 B6 B7

Stage 3 in the data analysis:

In this stage, the structural model was analyzed for tests of hypotheses 1, 2, 3, 4, 5 and 6. Figure 4.1 shows the structural model.

Stage 4 in the data analysis:

In this stage, the Multi-group analysis basically begins by fitting a model to the data for each sample considered separately with none of the parameters constrained to be equal across groups. This unconstrained model serves as the baseline model. Subsequently, in a stepwise fashion, more stringent constraints are placed on the model by specifying the parameters of interest to be constrained across groups. The model is then examined using a chi-square (χ ) difference test between the less restrictive and more restrictive models to determine whether the model and the individual parameter estimates (e.g., factor



Service value

V1 V2 V3 V4 V5 V6 V7

Switching barriers

Behavioural intentions

I1 I2 I3

Figure 3.2 Confirmatory factor analysis


loadings, factor inter-correlations, error variance, structural relations) are invariant across the samples. A significant difference in χ2 (namely,Δχ2) represents a deterioration of the model and the null hypothesis that e parameters are equal is rejected. A non-significant χ


2 difference is consistent with model invariance; that is, the parameters examined are equal across groups.

(Koufteros and Marcoulides, 2006).


Chapter 4 Data collection and result analysis

4.1 Data collection

A questionnaire consisting of 23 items was designed to measure the constructs of service value, satisfaction, perceived switching barriers and behavioral intentions (see Appendix A for questionnaire). Data were collected using a direct face-to-face interview on account of the flexibility and higher response rate. As study design, family group was targeted for survey. We used convenience sampling and carefully chose parental respondent to represent his/her family group. Contact with the respondents was made in the restaurant after their check-out, when they were more relaxed and receptive, and they may have fully perceived the farm services. Before the survey, respondents were given a full range of explanations about the questionnaire before they were asked to rate how much they agreed with each item on a five-point Likert-type scale ranging from strongly disagree to strongly agree. In the end of interview, small toys were given to respondents as gifts for their time.

The survey was administered to tourists from 1 July to 30 September 2006.

A total of 241 interviews were performed. The number of valid questionnaires was 226. Thus, the response rate in this survey was 93.78 percent. In terms of group size, 65% of the family groups surveyed consisted of 5 to 9 people, while 20% were less than 5 and 15% were greater than 9. These tourists all came from the Taipei metropolitan area, 71% were from Taipei city (see Table 4.1).


Table 4.1

Traveler profile of respondents Number of sample valid

First-visit 115

Revisit 111

Statistics of sample (%) Family size

1-4 20

5-9 65

Over 9 15

Areas the family come from

Taipei city 71

Taipei metropolitan area (except Taipei) 29

4.2 Confirmatory factor analysis

Following Anderson and Gerbing’s (1988) suggestion, the measurement part of the research model was analyzed before the structural model.

Confirmation of the constructs in terms of their reliability and validity was necessary before the structural analysis was performed. If this purpose was not achieved then a number of problems (i.e., model fit related, significance related, etc.) might be encountered during the structural analysis of the model (Anderson and Gerbing, 1988).

Confirmatory factor analysis (CFA) was performed to justify that the measurement model achieves an acceptable fit to the data. A reliability test was followed by validity tests. The reliability test aims to examine the internal consistency of the items that are used to measure a latent construct. The composite reliability is the most commonly used evaluation index. The results of the reliability test conclude that the sampling data achieved an acceptable level,


as the composite reliabilities for the service value, satisfaction, perceived switching barriers and behavioural intention constructs were 0.88, 0.88, 0.77 and 0.80, respectively (see Table 2). These values were all over 0.7, which is the critical value recommended by Hair et al. (1998).

Validity tests were conducted to examine the accuracy of the measurements.

In most cases, convergent validity and discriminant validity were both tested.

Convergent validity is the extent to which the latent variable correlates to items designed to measure the same latent variable (Garver et al., 1999). To test the convergent validity of the measurement model, we calculated the fit indices, including χ2/df (chi-square value divided by degree of freedom), CFI (comparative fit index), NNFI (non-normed fit index), IFI (incremental fit index) and RMSEA (root mean square error of approximation). The was 390.83 with 203 degrees of freedom and provides a ratio equal to 1.9. Both the CFI = 0.97, NNFI = 0.97 and the IFI = 0.98 also indicate good model fit. Finally, the RMSEA = 0.064 suggested fair model fit. Table 4.2 indicates the result of the good fit for every model.



Table 4.2

Results of convergent validity tests

Model χ /df2 * CFI** NNFI** IFI** RMSEA***

First-timer 1.5 0.98 0.98 0.98 0.063

repeater 1.7 0.95 0.95 0.95 0.080

Overall model 1.9 0.97 0.97 0.97 0.064 Values indicate a fair fit:(Jöreskog et al., 1993; Bentler, 1990; Browne et al., 1993)

* Value of χ /df below 3 2

** Values of CFI, NNFI and IFI values close to 1

*** Values of RMSEA below 0.08

In this study, indicators of the manifests with an associated Student’s t-statistic value of less than 2.58 (0.01 significance level), standardized coefficients below 0.45, or R2 below 0.2 were removed from the analysis (Maestro, et al., 2007). Models were stepwise modified to improve the fit indices (see Table 4.3).

Table 4.3

Scale reliability and CFA analysis results (n=226) Dimension


Completely standardized loadings

T-values R2 Composite


Service value 0.88

V1 0.70 11.65 0.49

V2 0.79 13.99 0.63

V3 0.73 12.31 0.53

V4 0.80 14.23 0.65

V5 0.50 7.68 0.25

V6 0.77 13.47 0.60

V7 0.51 7.96 0.27


Table 4.3 continued Dimension


Completely standardized loadings

T-values R2 Composite


Satisfaction 0.88

S1 0.63 10.24 0.40

S2 0.71 11.83 0.50

S3 0.70 11.65 0.49

0.82 14.75 0.68


S5 0.71 11.92 0.50

S6 0.56 8.76 0.31

Perceived switching barrier 0.77

BB1 0.54 8.02 0.29

BB2 0.60 9.18 0.36

BB3 0.59 8.94 0.35

BB4 0.63 9.63 0.40

BB5 0.49 7.13 0.24

BB6 0.58 8.77 0.37

BB7 0.54 8.01 0.29

Behavioural intentions 0.80

I1 0.87 15.77 0.75

I2 0.78 13.61 0.61

I3* 0.01 0.03 0.00

* removed

Following the convergent validity tests, the discriminant validity was examined. Table 4.4 provides detailed discriminant validity and correlations among the constructs. Discriminant validity is the extent to which the items representing a latent variable discriminate the construct from other items representing other latent variables. This can be assessed for two estimated constructs by constraining the estimated correlation parameter between them to


1.0, and then performing a chi-square difference test on the values obtained for the constrained and unconstrained models (Anderson et al., 1988; Garver et al., 1999). To do so, the chi-square differences of the pairwise constructs were individually measured. As the values were all far more than 6.635 ( (1) = 6.635), we concluded that the items of one latent variable sufficiently discriminated the construct from those of the other. Finally, Evidence provided through an examination of the confidence intervals also attests to discriminant validity, as there are no intervals that contain the value of 1.00. Thus, the discriminant validity was satisfied for overall model.



Table 4.4

Results of discriminant validity analysis (n = 226) Satisfaction Service


Switching Barriers

Behavioural Intentions Service Value

Satisfaction 0.86a*

(0.81 , 0.91)b Switching


0.68a* 0.76a*

(0.58 , 0.78)b (0.67 , 0.85)b Behavioural


0.91a* 0.94a* 0.88a*

(0.86 , 0.96)b (0.90 , 0.98)b (0.81 , 0.95)b a*Correlation is significant at p<0.01.

b95% Confidence interval for factor inter-correlation (Φ) in parentheses.

4.3 Structural regression analysis

After confirming the measurement model, the structural model shown in Figure 4.1 was estimated using LISREL 8.54 – the method used was the maximum likelihood estimation procedure on the variance-covariance matrix


with the raw data as input. The LISREL analysis showed an excellent overall fit of the model as indicated by the χ2/df, CFI, NNFI, IFI and RMSEA values of 1.9, 0.97, 0.97, 0.97, and 0.064, respectively. Given the satisfactory fit of the model, the estimated structural coefficients were then examined to evaluate the hypotheses. As predicted in H1, service value had a significant positive influence on behavioural intention (β = 0.30, t = 3.04, which is greater than the critical value of 2.58 at the significance level of 0.01). The results also showed that behavioural intention was directly influenced by satisfaction (β = 0.44, t = 3.59) and by perceived switching barriers (γ = 0.34, t = 4.46). These values confirmed H2 and H3, respectively.

Between the three constructs, the proposed model conjectured that service value would directly influence satisfaction (H4). The t-value results of 6.87 provide support for this link (β = 0.64). Also as expected, perceived switching barriers had a positive influence on satisfaction (γ = 0.33, t = 4.27) (H5) and on service value (γ = 0.58, t =8.05) (H6). In sum, the interdependences identified were all significant in this study.


Service value

Behavioural intentions


β=0.44** (H2) γ=0.34** (H3)

β=0.30* (H1) γ=0.58** (H6)

γ =0.33** (H5) Perceived switching barriers

β=0.64** (H4)

* p < 0.01

** p < 0.001

Test indices X2/df 1.9 Satisfied CFI 0.97 Satisfied NNFI 0.97 Satisfied IFI 0.97 Satisfied RMSEA 0.064 Satisfied

Figure 4.1 Results of path analysis (Standardized parameter estimate)

4.4 Multi-group analysis

To test the invariance between the first-time and repeat tourists (H7), we conducted a multi-group analysis. The analysis began with an unconstrained model, i.e. none of the parameters were constrained to be equal between the two groups serving as the baseline model. Subsequently, more stringent constraints, including factor loadings, factor correlations, error variance and structural coefficients were sequentially placed on the model by specifying the parameters to be constrained between the two groups.

The results are summarized in Table 4.5 The first step was to test Model 1, the model was estimated by constraining the number of factors and the factor loadings for the specific items defining each factor across the two groups to be invariant. The difference test did not result in a deterioration of the model.

Thus, the same four latent and factor loadings for specific items defining each χ2



Figure 1.1 Research flow Introduction

Figure 1.1

Research flow Introduction p.13
Figure 2.1 A means-end model relating price, quality and value. Adapted  from Zeithaml (1988)

Figure 2.1

A means-end model relating price, quality and value. Adapted from Zeithaml (1988) p.17
Figure 2.2 Evolution of farm tourism in Taiwan.

Figure 2.2

Evolution of farm tourism in Taiwan. p.26
Figure 2.3 Distribution of large leisure farms in Taiwan and the study case

Figure 2.3

Distribution of large leisure farms in Taiwan and the study case p.30
Figure 3.1 Overall model of tourist behavioural intentions

Figure 3.1

Overall model of tourist behavioural intentions p.31
Figure 3.2 Confirmatory factor analysis

Figure 3.2

Confirmatory factor analysis p.39
Table 4.3 continued  Dimension  Items  Completely  standardized  loadings  T-values R 2 Composite reliability  Satisfaction      0.88  S 1 0.63 10.24  0.40  S 2 0.71 11.83  0.50  S 3 0.70 11.65  0.49  0.82 14.75  0.68 S4 S 5 0.71 11.92  0.50  S 6 0.56 8.76

Table 4.3

continued Dimension Items Completely standardized loadings T-values R 2 Composite reliability Satisfaction 0.88 S 1 0.63 10.24 0.40 S 2 0.71 11.83 0.50 S 3 0.70 11.65 0.49 0.82 14.75 0.68 S4 S 5 0.71 11.92 0.50 S 6 0.56 8.76 p.45
Figure 4.1 Results of path analysis (Standardized parameter estimate)

Figure 4.1

Results of path analysis (Standardized parameter estimate) p.48
相關主題 :