elationship marketing is important in business, but it is not clear whether the same applies in private education. Traditionally, people perceive business and education differently. Business is for profit and seeks competitive advantage in its dynamic environment (Jaworski et al., 2000; Hemsley-Brown and Oplatka, 2010). Business organizations are usually run by private individuals, and offer products or services to customers. They are efficient and responsive to the changing needs because of the competition in market traditionally provided by governments (i.e. public education) and is non-for-profit. As education brings a better future to the society, countries and governments put tremendous efforts in developing education in order to strengthen the human capital of the society.

Besides public education, in order to raise the education level of citizens, many countries have also encouraged private organizations or parties to provide education in recent decades. Due to reduction in government funding, and the growing interest in education of private parties, there has been rapid growth in private educational organizations (Li, 2010). The education sector can be classified into two categories: one is owned by the government and heavily relies upon government funding (i.e. public education), and the other is owned by private parties and heavily relies upon students' tuition fee (i.e. private education). Private educational institutions rely heavily on tuition fee income and are accountable to students and students' families, while public educational institutions rely heavily on government funding and are accountable to the general public (Li, 2010;Levy, 2010). Besides, management style of private educational institutions is more business-like and they emphasize customer-first attitude while public educational institutions have more bureaucratic styles of management and provide services more for the well-being of the society than forprofit (Li, 2010;Kwong, 2000). According to Hennig-Thurau et al. (2001), the organization structure and culture are also different in public and private education; greater flexibility is found in private institutions.

Research from relationship perspective in the private education sector has been minimal, this research investigates whether relationship marketing concepts are applicable to private education. According to marketing concepts, having long-term relationships with students may provide competitive advantages to educational institutions because students provide a stable source of income to the institutions and they recommend their institutions to friends and relatives (Nguyen and LeBlanc, 2001). It is worthwhile to investigate whether relationship commitment is a factor affecting student loyalty in private education, and to get a better understanding of the key determinants of relationship commitment because this can help better resource allocation.

In relationship marketing, a higher level of relationship commitment leads to higher intention of the parties remaining in a relationship (Gronroos, 1990; student loyalty from the relationship perspective is important for managements of educational institutions when building a stable and quality relationship with students. As relationship commitment is crucial to customer retention, a better understanding of key determinants of relationship commitment is also necessary (Hocutt, 1998). The main objectives of this research are: to examine the main direct effect of relationship commitment on student loyalty; and to examine key determinants of relationship commitment.

ii.

In the present era of demanding customers and intense competition, relationship marketing has drawn attention from practitioners and academics (Sheth and Parvatiyar, 1995). Relationship marketing is considered as "establishing, developing and maintaining successful relational exchanges" (Morgan and Hunt, 1994, p. 22). "Relationship marketing is an integrated effort to identify, maintain and build a network with individual consumers and to continuously strengthen the network for the mutual benefit of both sides, through interactive, individualized and value-added contacts over a long period of time" (Shani and Chalasani, 1992, p. 44).

Education is people-based, involves a lengthy and formal relationship between education providers and students, and requires high level of customization in service delivery (Mazzard and Soutar, 1999). Educational institutions are considered as service organizations Kotler and Fox, 1995), building relationships with students is important. Students' satisfaction is based on a stable relationship (Gruber et al., 2010).

Providing benefits and value to customers is the means to encourage them to stay in their relationship with a particular company (Berry, 1983;Bitner, 1995;Kolter and Armstrong, 2004). The ability to provide superior benefits and value to customers is a prerequisite when establishing relationships with customers (Ravald and Gronroos, 1996). The relationship marketing theory suggests that in the competitive global marketplace, partner selection may be a key element in competitive strategy (Morgan and Hunt, 1994). Morgan and Hunt (1994) considered relationship benefits as the quality of services and goods relative to other suppliers. Relationship benefits are the superior benefits provided to customers, which are highly valued by customers.

Students are customers of education and expect to get benefits in the relationship (Finney and Finney, 2010).

A common assumption in relationship marketing is that termination has switching costs and seeking an alternate relationship leads to dependence (Heide and John, 1988;Jackson, 1985). "Termination costs" and "switching costs" are often interchangeable terms in research studies. Though Morgan and Hunt (1994) considered switching costs to be of an economic nature only, switching costs may also comprise psychological and emotional costs (Sharma and Patterson, 2000). Adidam et al. (2004) defined relationship termination costs as the perception of net losses (financial, emotional, or time) that may result from dissolution of the relationship. In their public education study, the perceived costs to a business student include both economic and non-economic sides of switching costs; costs might include the loss of friendships or loss of credits on switching to another educational institution. The losses cannot be made good by an alternate supplier.

"Shared values" is a shared code or a shared paradigm that facilitates a common understanding or perception of collective goals and actions (Tsai and Ghoshal, 1998). Shared values are defined as "the extent to which partners have beliefs in common about what behaviors, goals and policies are important or unimportant, appropriate or inappropriate, and right or wrong" (Morgan and Hunt, 1994, p. 25). It means two parties having similar perceptions can enhance their communications and avoid misunderstanding. Holdford and White (1997) found that pharmacy students who shared the same goals, ideals and codes of conduct with their public schools were more likely to commit to a relationship with the school.

A trustworthy party is one that is considered reliable and has high level of integrity and associated qualities of competence, consistence, fairness, honesty, responsibility, helpfulness and benevolence. Morgan and Hunt (1994) used reliability and integrity together to define and conceptualize trust. Morgan and Hunt (1994, p. 23) defined trust as "when one party has confidence in an exchange partner's reliability and integrity".

In public education, Adidam et al. ( 2004) conceptualized trust as confidence in an exchange partner's reliability and integrity basing on personal experiences individual student has had with his/her education institution.

The building of relationship commitment is very important because the level of commitment determines relationship strength and the intention of the parties to (Hocutt, 1998). Relationship commitment is defined as "an exchange partner believing that an ongoing relationship with the other is so important as to warrant maximum efforts at maintaining it, that is, the committed party believes the relationship is worth working on to ensure that it endures indefinitely" (Morgan and Hunt, 1994, p. 23). Relationship commitment entails a desire to develop a stable relationship and confidence in the stability of the relationship (Anderson and Weitz, 1992).

This research adopts Moorman et al.'s (1992) concept of relationship commitment as an enduring desire to maintain a valued relationship, and investigates the key determinants of relationship commitment in private higher education.

Loyalty is defined as "a deeply held commitment to rebuy or repatronize a preferred product/service consistently in the future, thereby causing repetitive same-brand or same brand-set purchasing, despite situational influences and marketing efforts having the potential to cause switching behavior" (Oliver, 1999, p. 34). Loyalty comprises of repurchase intention and word-of-mouth (Zeithaml et al., 1996). Repurchase intention implies doing more business with the company in future and considering the company to be the first choice. It is a customer's judgement about buying again a product or service from the same company while taking into account the current situation (Hellier et al., 2003). Word-of-mouth is to say positive things about the company and recommend the company to others. Some studies on student loyalty have adopted the two aspects of loyalty identified by Zeithaml et al. (1996), i.e. repurchase intention and word-of-mouth. This research adopts the repurchase intention aspect of student loyalty because it aims to ascertain whether the existing sub-degree students of private higher education institutions would continue to pursue bachelor degree courses at their current education institutions in future.

The conceptual framework was based on concepts and findings from relationship marketing literature. The model was modified from the studies of Morgan and Hunt (1994), Adidam et al. (2004), and Holdford and White (1997), sought to illustrate the relationship between the factors: relationship benefits, relationship termination costs, shared values, and trust, and relationship commitment, and the relationship between relationship commitment and student loyalty (Figure 1).

Relationship benefits generate positive impact on relationship outcomes, such as, continuation of a relationship (Gwinner et al., 1998;Patterson and Smith, 2001), site commitment (Park and Kim, 2003), commitment to the service business (Hennig-Thurau et al., 2002), exporter's commitment to importers in exporter-importer relationships (Obadia, 2010), and satisfaction in retail banking (Dimitriadis, 2010). Therefore, it was proposed the same in private higher education:

Hypothesis H1: Students' perception of relationship benefits has a significant positive impact on relationship commitment.

Dwyer et al. (1987) suggested that anticipation of high switching costs by customers generates commitment to an ongoing relationship. Besides the economic side of switching costs, they also consider socio-psychological costs, such as worry and loss of reputation, which contribute to the commitment. Switching costs affect customers' commitment in the financial services industry (Yanamandram and White, 2010). In industrial marketing and distribution channels, extant literature suggests that the relationship may continue to exist because of the high switching costs perceived by the buyer (Porter, 1980;Ping, 1994). Findings of Vasudevan et al. (2006), Burnham et al. (2003), and Patterson and Smith (2001) suggest that relational switching cost that involves psychological and emotional discomfort due to breaking of bonds and loss of identity is positively associated with commitment. Therefore, it was proposed the same in private higher education:

Hypothesis H2: Students' perception of relationship termination costs has a significant positive impact on relationship commitment.

Shared values have been found to have positive impact on relationship commitment (Morgan and Hunt, 1994). The parties share similar beliefs in behaviors, goals and policies. Similar perspectives, including shared language and shared narratives are important for sustaining ongoing relationships (Chua, 2002; Nahapiet and Ghoshal, 1998). Therefore, it was proposed that: Hypothesis H3: Students' perception of shared values has a significant positive impact on relationship commitment.

Trust enhances commitment to a relationship by reducing transaction costs in an exchange relationship, reducing risk perceptions associated with the partner, and increasing confidence that short term inequities can be resolved in the long run. Trust has been found to be a factor affecting commitment in many previous studies (Spake and Megehee, 2010; Nusair and Li, 2010; Cassab and MacLachlan, 2009; Cater and Zabkar, 2009; Morgan and Hunt, 1994).Therefore, it was proposed that:

Hypothesis H4: Students' trust in the education institution has a significant positive impact on relationship commitment.

Tino (1975,1993) suggested that commitment directly affects student loyalty. Adidam et al. (2004) and Holdford and White (1997), based on the model of Morgan and Hunt (1994), investigated the relationship between students and their public education institutions.

The findings suggested that relationship commitment had great positive impact on acquiescence and cooperation between students and education institutions and negative impact on propensity to leave. Therefore, it was proposed that: Hypothesis H5: Students' relationship commitment to the education institution has a significant positive impact on student loyalty.

A quantitative research study using questionnaire was adopted to examine the key factors affecting relationship commitment, and the relationship between relationship commitment and student loyalty. 480 copies of questionnaire were distributed to current private sub-degree students in one of the largest higher education provider.

For the purpose of this research, a private higher education institution was identified from the list of higher education institutions available on the website of Education Bureau of the HKSAR Government. Enrolment of students in this institution accounted for approximate 11% of the total number of private subdegree students in 21 higher educational institutions in Hong Kong. This institution was approached and it agreed to allow the researcher to administer the questionnaire survey to sub-degree students at the campus. Convenience sampling technique was used to approach the students because students are the direct customers of the education institutions.

Student loyalty was measured with three items, adopted from a previous study in education context (Nguyen and LeBlanc, 2001). Relationship commitment was measured with three items, and trust was measured with four items, adopted from Holdford and White (1997), a previous study in public education. Four items of relationship benefits and three items of shared values were adopted from previous studies in public education (Adidam et al., 2004;Holdford and White, 1997). Questions for measuring relationship termination costs were adopted from Sharma and Patterson (2000). The 7-point Likert-type scales were anchored by 1 (strongly disagree) and 7 (strongly agree) for all questions.

The content and construct validity of each variable had already been evaluated by the original authors, therefore, it is reasonable to assume that the content and construct validity of the multidimensionalitem scales should accurately represent the variables concerned.

Analysis And Result a) Data Analysis Confirmatory factor analysis was performed for all variables: relationship benefits, relationship termination costs, shared values, trust, relationship commitment, and student loyalty. Cronbach's alpha was used to test internal validity, the Cronbach's alpha coefficient greater than 0.7 is considered as satisfactory (Bryman, 2008). Structural Equation Model was used to test the positive association of hypotheses H1 to H5, and calculate the variance of relationship commitment explained by the factors and the variance of student loyalty explained by relationship commitment in the research model.

A total of 444 valid questionnaire copies were collected which provided a response rate of approximately 92.5% out of the 480 copies sent out. 60.4% of the respondents were female. 98.2% of the respondents were in the age range of 18 to 25. 40.1% of the respondents were associate degree students and 59.9% were higher diploma students. Almost half of the respondents were studying business courses.

Reliability and validity were assessed to ensure the information is trustworthy. Cronbach's alpha was used to measure consistency among the items in each variable of the questionnaire, and a value of 0.7 or above is considered as acceptable and having internal consistency (Shin et al., 2000). The variables of this research had Cronbach's alpha values from 0.785 to 0.877 (Table 1) were therefore acceptable.

The covariance matrix produced values ranging from 0.229 to 0.819 for each pair of construct, which are lower than the recommended level of 1.0 (Koerner, 2000) (Table 2). The result suggests that the constructs are statistically distinct within the CFA model, and provides evidence of discriminant validity.

Hypothesis H1 and H2 are supported by empirical evidence.

Relationship benefits and relationship termination costs show strong influence on relationship commitment, as indicated by high to moderate standardized coefficients 0.563 and 0.371 respectively. Shared values construct has a small direct effect on relationship commitment (standardized coefficient 0.116), Hypothesis H3 is supported. However, hypothesis H4 should be rejected, the standardized coefficient of -0.038 suggests that trust has non-significant influence statistically on relationship commitment in private higher education. Hypothesis H5 is supported by empirical evidence.

Relationship commitment has a strong influence on student loyalty, as indicated by high standardized coefficient of 0.796. It can therefore be concluded that hypotheses H1, H2, H3 and H5 are strongly supported with empirical evidence in the research model. The factors affecting relationship Y commitment together explain 71.7% of relationship commitment (R 2 = 0.717), and relationship commitment explains 63.4% of student loyalty (R 2 = 0.634).

V.

The results of this research support the direct effect of relationship commitment on student loyalty, and the direct effects of relationship benefits, relationship termination costs, and shared values on relationship commitment in private higher education which is consistent with most of previous research studies' results in business context. However, unlike the common finding in most relationship marketing literature that trust is a determinant of relationship commitment, the direct effect of trust on relationship commitment is found to be insignificant in this research. The rejection of predictive effect from trust on relationship commitment in private higher education environment provides a new angle to the application of relationship marketing concepts in education settings.

Most previous studies related to relationship marketing concepts were conducted in U.S. and Europe; little attention has been paid to Asian countries, particularly Hong Kong. This research verifies applicability of relationship marketing concepts in the East.

This research shows that relationship commitment has a substantive and positive effect on student loyalty in the private higher education industry. The higher the relationship commitment of students with an educational institution is, the higher is the student loyalty. The student will pursue further studies in the current private higher education institution if the student has high relationship commitment. Therefore, education providers have to focus on enhancing relationship commitment in order to increase student loyalty.

Education providers can use the results of the path analysis to understand preferences of private students (customers) and allocate resources to enhance the factors that affect students' relationship commitment which, in turn, enhances student loyalty. With the findings of this research, education providers can gain a better understanding of factors affecting relationship commitment, and therefore can plan to nurture them. Considering all the four factors (relationship benefits, relationship termination costs, shared values, and trust) affecting relationship commitment, the R 2 0.717 indicates that 71.7% of the variance of relationship commitment is explained by these four factors in the proposed model.

The construct of relationship benefits is the most influential determinant of relationship commitment in the private higher education. Relationship benefits include education quality, internship opportunities, placements, professional seminars, and company visits etc. (Adidam et al., 2004). Education providers have to improve these perceived relationship benefits continuously in order to raise relationship commitment of students.

The construct of relationship termination costs is the next influential factor. This provides signals to education providers that students' perceived costs, both economic and non-economic, are important consideration in building relationship commitment in private higher education. Education providers have to increase the relationship termination costs in order to raise students' relationship commitment with the education institution. The construct of shared values is also a determinant of relationship commitment in the private higher education industry. The more the staff and students have similar values on education issues, such as learning behavior, assessments and work-load, the more the students will be committed to the relationship with the educational institution (Adidam et al., 2004). Although the influence of shared values on relationship commitment is not as strong as that of relationship benefits and relationship termination costs, private education providers still have to raise the perceived shared values between students and education institution in order to increase relationship commitment of students.

Unlike the common finding in relationship marketing literature that trust is a determinant of relationship commitment (Morgan and Hunt, 1994), the direct effect of trust, characterized as having confidence in partner's reliability and integrity, on relationship commitment, is found to be insignificant in this research. The path coefficient from trust (TR) to relationship commitment (RC) is not significant (H4: ? = -0.038), which suggests trust has non-significant influence statistically on relationship commitment, in the private higher education context. This result provides new insights into relationship marketing in the private higher education context. This finding is not surprising because the primary intention of private sub-degree students is to get degree places after the 2-year sub-degree study. Sub-degree students may consider studying at the current education institution as a stepping stone to degree programmes, and their desired degree programmes can be offered in other education institutions. In the study of Grayson and Ambler (1999), the results suggested that the influence of trust on relational outcomes was moderated by length of relationship.

They commented that the length of relationship may change the nature of association between relational constructs, and there is value in future research to investigate the relational dynamics with respects to the length of relationship. In the current research, students just spend two years in the current sub-degree study and they have a strong desire to get Year degree places in their desired education institutions, therefore, trust has non-significant influence on their commitment towards their current education institution.

Apart from the length of relationship, in order to get better understanding on the influence of trust on commitment, Moorman et al. (1992) suggested that other factors, such as economic factors and power, may affect how relationship operates, and future research can examine how trust interacts with these factors in affecting relational outcomes. Ganesan and Hess (1997) also suggested future research can study the impact of moderators, such as phase of relationship, reputation of the organization, and level of environmental uncertainty, on the link between trust and commitment. In the current research, it studies the direct effect of trust on commitment, while the impact of moderators which mentioned by Moorman et al. (1992) and Ganesan and Hess (1997) has not been investigated. In future, the current research can be extended to investigate the moderating impact of these moderators on the relationship between trust and commitment in order to understand the trustcommitment link better.

Firstly, due to time constraints, a crosssectional study was conducted, which was unable to take the actual behaviour of respondents into account.

Secondly, measurement scales used were adopted from previous studies. As the features of private higher education context may be different from features of other contexts, the adopted scales might not be as effective as scales tailor-made for a particular context.

Constructs that capture contextual characteristics have not been discussed in this research. The characteristics of higher education may affect the findings of the research.

Thirdly, some constructs that were thought to affect relationship commitment in previous literature were not included in this research. Only 71.7% of variance of relationship commitment is explained by relationship benefits, relationship termination costs and shared values, implying that there should be other factors affecting relationship commitment; and only 63.4% of variance of student loyalty is explained by relationship commitment.

Fourthly, the impact of moderators on the link between trust and relationship commitment was not included in this research.

Firstly, future research can consider developing measurement scales for education in eastern environment. This may help education institutions' managements make better decisions.

Secondly, future research can consider conducting a longitudinal study to trace the changing preferences and behaviors of students (customers). The use of multiple time frames allows researchers to track behavioral intentions of students (customers) over time.

Thirdly, future research can consider adding constructs that capture contextual characteristics. This is important because of the rapid expansion of education in most parts of the world. The current results show that 71.7% of variance of relationship commitment is explained by three major factors and 63.4% of variance of student loyalty is explained by relationship commitment. Obviously, there are some unexplained portions which have not been captured in this research. The non-captured portions may be related to contextual characteristics.

Fourthly, future research can consider investigating the impact of moderators on the link between trust and relationship commitment. This may help better understanding of the influence of trust on commitment.