Tsai-fa Yen
Department of Tourism and Leisure Management Fortune Institute of Technology, Taiwan (R.O.C.)
Hsiou-hsiang J. Liu
Department of Tourism Management
National Kaohsiung University of Applied Sciences, Taiwan (R.O.C.)
Yung-Chieh Chen*
Department of Tropical Agriculture and International Cooperation, National Pingtung University of Science and Technology, Taiwan (R.O.C.)
*Corresponding Author: yungchieh@hotmail.com
Abstract
With the active growth of agritourism in Taiwan, the total amount of leisure farms is keeping growing. However, it is now facing some marketing difficulties. For instance, the higher churn rate of the tourists or low revisiting rate despite of high revisit intention. Evaluating the reasons they switch destination and understanding how to achieve tourists’ commitment is needed for the development of Taiwanese agritourism. The current study develops and tests a model of tourists’ commitment that incorporates such contingencies between satisfaction and switching costs. A core proposition is that the effect of satisfaction on commitment depends on the magnitude of switching costs in the agritourism context. Data was collected by ques-tionnaire survey from tourists in leisure farms. The findings show that affective commitment is a mediator between satisfaction and continuance commitment; switching costs is a modera-tor on satisfaction-affective commitment relationship. Finally some more managerial implica-tions are drawn.
Keywords: agritourism, leisure farms, switching costs, commitment
Introduction
Within the agritourism industry in Taiwan, three types of experience service are offered by leisure farms i.e. agricul-tural experience activities, food and bever-age service, and accommodation service.
Each of the farms can adapt one or more service as its business model. Those adopt three types of experience service are full-service farms and others are limited-service farms. There are clearly defined business segments, which vary in their level of service and amenities, and attract different customer types with different needs. However, with the active growth of agritourism, the total amount of leisure farms has been more than 1,244 in 2007, recording a total revenue of an estimated 4.5 billion NT dollars in 2004 and the gross output value of agritourism was es-timated to be 21.7 billion NT dollars (Tuan, 2011; Chen, Tuan, Lin and Xie, 2007). Taiwanese agritourism is now fac-ing some marketfac-ing difficulties. For in-stance, the higher churn rate of the tourists or low revisiting rate despite of high revisit intention (Cheng, 2003; Lin, Chen and Wang, 2007).
Reflecting upon these problems iden-tified above, researchers have found that the total cost of bringing a new customer to a comparable level of profitability to that of the lost customer is approximately sixteen times greater (Lindgreen et al., 2000), and customer switching has delete-rious effects on organizations’ market
share, profitability, viability, and future revenue stream in today’s competitive marketplace (Ganesh et al., 2000; Keave-ney, 1995; Rust et al., 1995). Evaluating the reasons they switch destination and understanding how to achieve visitors’
commitment is needed for the develop-ment of Taiwanese agritourism.
Moreover, there have been studies indi-cating that providing good service behav-ior and professional interpretation may help leisure farms improve relationship quality (RQ) and revisit intention (Wei, 2008); good service quality presents posi-tive impacts on customer loyalty (Lee, Chou and Lin, 2006). Furthermore, studies out of agritourism have explored how to obtain the customers’ satisfaction with higher service quality (Kim and Lee, 2011;
Park, Robertson, and Wu, 2004) and with higher perceived value (Santiago, Ramon, Javier, and Luís, 2012; Chen and Chen, 2010; Reisinger and Turner, 2003). One has demonstrated the higher satisfaction visitors’ perceived could lead to higher affective commitment (Yen, 2009) and revisit intention (Han, Back, and Barrett, 2009). Some of them have understood their behavioral intentions (Santiago et al., 2012; Chen and Chen, 2010), and achieved their loyalty (Santiago et al., 2012) when visitors perceived the higher satisfaction.
There is very little scholarly research evaluating the role of switching cost.
Switching costs refer to visitors’ percep-tions of the time, money, and effort
associ-ated with changing service providers (Jones, Mothersbaugh and Beatty, 2000).
It is one of switching barriers and can positively enhance continuance commit-ment and decline switching intention to leave of customers (Bansal et al., 2004;
Yen, 2009). Another study states that it is positively related to affective commitment and can moderate the relation between satisfaction and affective commitment (Yen et al., 2010). It is important because it may generally foster greater retention and help companies weather short-term fluctuations in service quality that might otherwise result in defection. However, managers and scholars are confused be-cause it acts a contingency role on satisfac-tion and commitment relasatisfac-tions. Clarifying its role on the relations of those variables will have higher probabilities to benefit decision making of managers and to fill up the theoretical gaps.
Therefore, the current study develops and tests a model of visitors’ commitment that incorporates such contingencies be-tween customer satisfaction and switching costs. A contingency approach has been called for by a number of researchers (e.g., Anderson and Fornell, 1994), but has gen-erally not been adopted in studies of com-mitment for agritourism. A core proposi-tion is that the effect of satisfacproposi-tion on commitment depends on the magnitude of switching costs in the agritourism context.
Satisfaction should play a lesser role when exit costs are high and a greater role when exit costs are low. This proposition, if
sup-ported, would (1) augment existing visitor-commitment models which focus mostly on satisfaction, (2) help to explain variabil-ity in the satisfaction-commitment rela-tionship evidenced in prior research, and (3) provide guidance to leisure farms in developing visitor-commitment programs.
Literature Review
Satisfaction
Satisfaction, according to Oliver (1980) and Tse and Wilton (1988), is an evaluation made by a person between pre-viously created expectations and the result obtained from the consumption of a prod-uct or service; i.e. the final psychological state resulting when the feeling around the disconformity of expectations meets the previous sentiments about the consump-tion experience (Oliver, 1981). In tourism context, satisfaction is primarily referred to as a function of pre-travel expectations and post-travel experiences. When experi-ences compared to expectations result in feelings of gratification, the tourist is satis-fied. However, when they result in feelings of displeasure, the tourist is dissatisfied (Reisinger and Turner, 2003).
Furthermore, satisfaction itself has occasionally conceptualized as emotional responses to product/service experiences (Han and Back, 2007). This emotional response is a critical determinant of com-mitment (Kyle, Theodorakis, Karageor-giou, and Lafazani, 2010) and can enhance
affective commitment of a tourist (Liu, Yen and Huan, 2010).
Commitment
Commitment has been defined as a force that binds an individual to a course of action of relevance to one or more tar-gets and is distinguishable from exchange based forms of motivation and from target-relevant attitudes and can influence behav-ior even in the absence of intrinsic motiva-tion or positive attitudes. (Meyer and Her-scovitch, 2001; Bansal et al., 2004). The role of commitment in loyalty has been well documented. Affective commitment, also referred to as emotional or relation-ship commitment, can be described as an emotional attachment that creates a sense of belonging and personal identification and a desire to maintain a long term rela-tionship with the provider (Allen and Meyer, 1990; Baloglu, 2002; Benapudi and Berry, 1997; Bowen and Shoemaker, 2003; Fullerton, 2003, 2005; Mattila, 2001, 2006; Sui and Baloglu, 2003; Liu et al., 2010; Tanford, Raab, and Kim, 2012).
Affective commitment is considered key to building relationships within the hotel in-dustry (Bowen and Shoemaker, 2003;
Shoemaker and Lewis, 1999).
A second type of commitment has been termed calculative commitment (Mattila, 2006), value commitment (Tan-ford et al., 2010), or Continuance mitment (Bansal et al., 2004). Value com-mitment is less enduring and associated
with greater price sensitivity and willing-ness to switch hotels than affective com-mitment (Tanford et al., 2010). Continu-ance/calculative commitment refers to a cost-based attachment where an employee feels he or she has to stay with the organi-zation (i.e., employees remain with the organization because they need to) (Bansal et al., 2004). Continuance commitment represents a constraint-based force binding the consumer to the service provider out of need. It reflects the fact that consumers stay with a service provider because they feel they have to; it reflects a sense of be-ing “locked in” to the service provider (Meyer and Herscovitch, 2001).
Numerous studies have identified the important roles of satisfaction in building commitment (Back, 2005; Han and Back, 2008; Kim and Han, 2008; Yuksel et al., 2010; Kyle, Theodorakis, Karageorgiou, and Lafazani, 2010; Liu et al., 2010). Sat-isfaction is a critical determinant of com-mitment (Kyle et al., 2010), when a desti-nation predicts tourist behavioral loyalty.
Kim and Han (2008) found that satisfac-tion enhances customers’ favorable inten-tions toward a restaurant firm. In the hotel industry, Kim et al. (2001) demonstrated that satisfaction, as a central part of rela-tionship quality, was an important predic-tor of commitment and behavioral inten-tions. In agritourism, Liu et al. (2010) and Yen (2009) have evidenced that satisfac-tion can enhance affective commitment.
Hence, the following hypothesis is devel-oped:
H1: the higher satisfaction a tourist per-ceived could lead to the higher affective commitment.
Moreover, commitment can be related to purchase decision factors (Tanford et al., 2012). Hospitality research has docu-mented the role of affective commitment and relationship quality in loyalty to hotels (Mattila, 2006; Tanford et al., 2010), casi-nos (Baloglu, 2002; Sui and Baloglu, 2003), and restaurants (Hyun, 2010; Mat-tila, 2001). Affective commitment is posi-tively associated with continuance com-mitment in agritourism (Wu, Yen, Tsai, 2009; Yen, 2009). In those studies, affec-tive commitment has been shown to be a stronger determinant of loyalty than other forms of commitment. Hence, the follow-ing hypothesis is developed:
H2: the higher affective commitment a tourist perceived could lead to the higher continuance commitment.
Switching Costs
Perceived switching costs are con-sumer perceptions of the time, money, and effort associated with changing service providers (Jones et al., 2000). Ones state that switching costs are the costs/sacrifices that may be incurred when changing pro-viders, including monetary and non- monetary costs/sacrifices (i.e., time, psy-chological costs) (Dick and Basu, 1994;
Han et al., 2009). Monetary costs are sunk cost and non-monetary costs refer to a
perceived risk. Such costs may entail search costs resulting from the geographic dispersion of service alternatives, as well as learning costs resulting from the cus-tomized nature of many service encounters (Guiltinan, 1989). As the perceived costs of an activity increase, the likelihood of consumers engaging in such behavior should diminish (Yen et al., 2009). Switch-ing service providers is likely to involve various behavioral and psychological costs, and such costs should act to dimin-ish switching tendencies (Jones et al., 2000).
Economic models of buyer behavior generally posit that consumers weigh both the costs and benefits of a particular deci-sion (Hauser and Wernerfelt, 1990). One implication is that as perceived switching costs increase, the perceived costs of switching should eventually outweigh the perceived switching benefits arising from dissatisfaction with the core service. When perceived switching costs are low, dissatis-fied consumers should be more likely to defect than are satisfied customers. Alter-natively, when perceived switching costs are high, customers may remain despite their dissatisfaction due to perceptions that switching costs outweigh switching bene-fits (Jones et al., 2000). Hospitality re-search has been offered the evidence that dissatisfied guest will not switch the ser-vice firm because the switching costs are high (Han et al., 2009). The probability for developing SAT-AC relation might be
moderated by switching costs. Therefore, this study hypothesizes that:
H3: As perceived switching costs increase, the relationship between satisfaction and affective commitment will diminish (i.e., switching costs × satisfaction interaction).
Methodology
Having considered the data collection requirements of this study such as a need of large sample of customers and quanti-ties of Taiwanese agritourism, it would be appropriate to employ the field survey with a self-administered questionnaire as the primary data collection technique for this study. The field study method was chosen in order to gain information di-rectly from individuals at the leisure farm settings. As such, their feelings and per-ceptions about the setting with respect to relational satisfaction, affective commit-ment, continuance commitment and switching cost are likely to be clearly in mind (Danaher and Mattsson, 1994).
To ensure the content validity of the scales, the items selected constructs are mainly adopted from prior studies. The study uses exiting scales for measuring satisfaction, affective commitment, con-tinuance commitment and switching cost.
Three items for SAT were drawn based on the studies of De Wulf et al. (2001) and Yen and Liu (2009). Three items for SWC were drawn based on the studies of Han et al., (2009) and Jones et al. (2000). Three
items for CC were drawn based on the studies of Bansal et al. (2004) and Jones et al. (2000). Three items for AC were drawn based on the studies of Tanford, Raab, and Kim, (2012) and Yen and Liu (2009). The initial items were confirmed and corrected by the managers of leisure farms and pre-tested was done by EMBA (Executive Master of Business Administration) stu-dents in NPUST (National Pingtung Uni-versity of Science and Technology), Tai-wan. For items, responses were ratings from 1 to 7. The anchors are “strongly disagree” (1) and “strongly agree” (7) for measuring SWC (switching cost), CC (continuance commitment), and AC (affec-tive commitment). The anchors for SAT (satisfaction) are “strongly displeased” (1) and “strongly pleased” (7), “strongly dis-gusted” (1) and “strongly dis-gusted” (7) and
“strongly dissatisfied” (1) and “strongly satisfied” (7).
It was decided that the model would be tested by collecting data from leisure farms in Taiwan. The criteria for farms’
selection were based on their service qual-ity of experience, food and beverages, and accommodation certified by Taiwan Lei-sure Farms Development Association (TLFDA). Finally, a total of 23 farms were drawn and could be categorized into full-service farms and limited-full-service farms.
They were selected expecting adequate diversity of quality and loyalty to allow a model to be estimated. A questionnaire was prepared for collecting rating and other information. Items measuring the
various constructs were distributed about in the questionnaire to reduce halo effects.
Because the goal was to develop a model, random sampling was not seen as necessary. Surveyors were collecting data from visitors they did not know. Quota sampling was adapted to ensure that re-spondents were distributed across age and sex groups. Having enough respondents in certain categories was seen as important for data to be appropriate for estimating the model of concern. Data was collected by personal contact with respondents at rest area of the farm. In collecting data, respondents were asked to complete a printed questionnaire. The data collectors, as necessary, clarified the meaning of questions and answers. In other words they dealt with any problems encountered while answering questions. Data were collected during the April to May in 2012. A total of 351 valid questionnaires were received. Of 351 questionnaires (166 respondents were drawn from limited-service farms and 185 respondents were drawn from full-service farms) obtained about 50% were from fe-male respondents (50.4%). At about 16.8%
of respondents were below 20 years of age and 13.6% of respondents were higher than 40 years of age. The majority of re-spondents were between 20-40 years of age (69.8%). Approximately 7.1% of re-spondents were graduated from junior high school and 26.5% of respondents were graduated from junior high school. At about 65.4% respondents were graduated from college or above. With regard to the
frequency visited, 55.6% of respondents were first time to the destination and 44.4% were revisit.
Results
A confirmatory factor analysis (CFA) using AMOS 17.0 and SPSS 17.0 were conducted to test the measurement model and hypothesis. Before testing the model, the data were examined. For making maximum likelihood (ML) estimates for path models (Kline, 1998), there are prob-lems if certain conditions arise. There are likely to be outliers if the absolute value of skewness is greater than 3. Also, there is a distribution problem if the absolute value of kurtosis is larger than 10. One wants data that is approximately normally dis-tributed for making ML estimates. For this research the skewness of variables ranges between -0.605 and 0.311 (Table 1) so the
< 3 criterion is met. The kurtosis values are between -0.842 and 0.650 so the < 10 criterion is met. Therefore, this enables authors to proceed in evaluating the meas-urement models.
The chi-square (115.06) is significant (p < 0.05; Bollen, 1989), a finding is not unusual with large sample sizes (Doney and Cannon, 1997). The ratios of chi-square to degrees of freedom (df= 48) are 2.39 for measurement model within the acceptable range of 2 to 5 (Marsh and Hovecar, 1985). The values for GFI (0.948), AGFI (0.915), CFI (0.966), and RMSEA (0.063) are acceptably close to
the standards suggested by Hu and Bentler (1999) 0.9 for GFI, 0.9 for AGFI, 0.95 for CFI and 0.08 for RMSEA. Given that these batteries of overall goodness-of-fit (GFI) indices were accurate and that the model was developed on theoretical bases, and given the high level of consistency samples, no respecifications of the model were made. This enables authors to pro-ceed in evaluating the reliability and valid-ity.
This study assesses the quality of measurement efforts by investigating uni-dimensionality, convergent validity, reli-ability, discriminate validity. Evidence for the uni-dimensionality of each construct included appropriate items that loaded at least 0.573 on their respective hypothe-sized component and loaded no
Table 1. Reliability and convergent validity
Concept Items Mean SD Skewness Kurtosis Loading SMC CR AVE SAT sat1 4.60 1.47 -.605 .397 .720 .519 .89 .72
sat2 4.85 1.36 -.220 -.438 .779 .607 sat3 4.95 1.32 -.442 .067 .815 .664
AC ac1 3.97 1.60 -.101 -.683 .834 .696 .92 .80 ac2 3.92 1.59 -.057 -.592 .835 .697
ac3 4.16 1.73 -.081 -.842 .831 .691
CC cc1 3.29 1.60 .311 -.410 .816 .666 .87 .70 cc2 3.46 1.70 .248 -.688 .818 .670
cc3 4.13 1.55 -.190 .350 .622 .386
SWC swc1 4.29 1.25 -.068 .406 .573 .329 .87 .69 swc2 4.08 1.22 -.222 .650 .875 .766
swc3 4.11 1.29 .011 -.346 .797 .635
Note: χ2=115.06(p=.000); df=48; GFI=.948; AGFI=.915; CFI=.966; RMSEA=.063 SMC: Squared Multiple Correlation; CR: Composite Reliability; AVE: Average Variance Extracted
larger than 0.30 on other components in a factor analysis (see Table 1). In addition, the overall goodness of fit supports uni-dimensionality (Steenk-amp and van Trijp, 1991). Convergent validity was supported by all loadings being significant (p < 0.01) and nearly all SMC (square of multiple correla-tion) exceeding 0.30 (Hildebrandt, 1987). This study assesses reliability
jointly for all items of a construct by computing the composite reliability (C.R.) and average variance extracted (AVE) (Baumgartner and Homburg, 1996; Steenkamp and van Trijp, 1991).
For a construct to assess good reliabil-ity; composite reliability should be higher than 0.70, and the average vari-ance extracted should at least be 0.60
(Bagozzi and Yi, 1988). All scales demonstrate good reliabilities.
To examine discriminant validity, this study first checks the coefficients of correlations between factors whether they are significantly lower than 1 and then compared the correla-tions between factors with their AVE (Gaski and Nevin, 1985). The results show that all of coefficients of
correla-tions between factors are significantly lower than 1 and the correlations be-tween factors are lower than their AVE, thus confirming discriminant validity (see Table 2). In summery, the measurement model demonstrates ade-quate uni- dimensionality, convergent validity, reliability, and discriminant validity. This enables authors to pro-ceed in evaluating hypotheses testing.
Table 2. Discriminate validity
Concept Mean SD SAT AC CC SWC
SAT 14.40 3.55 .72
AC 12.04 4.39 .64** .80
CC 10.88 4.07 .47** .72** .70
SWC 12.47 3.16 .39** .40** .42** .69
** Correlation is significant at the 0.01 level (2-tailed).
Diagonal elements are AVE. Off-diagonal elements are correlations between factors.
Before testing the hypotheses, the role of SAT, CC, and AC should be clarified to make sure their relations.
Three steps were adapted to examine the mediated effects of AC. The results (see Table 3) showed that the effects of SAT on CC (t= 6.85; β=.46) and on AC (t= 6.85; β=.46) were significant when the only independent variable SAT was considered in the model; the effects of AC on CC (t= 8.74; β=.72) was significant but SAT on CC (t=
.117; β=.01) was not when the
inde-pendent variable SAT and AC were considered in the model. This indicated
inde-pendent variable SAT and AC were considered in the model. This indicated