Transaction frequency and trust in internet buying behaviour

Transaction frequency and trust in internet buying behaviour

McCole, Patrick

There has been much discussion of the role of trust in Internet environments and it has been argued that a lack of consumer trust is the primary barrier to the full commercial development of the Internet. Trust has been conceptualised as a crucial antecedent and outcome of buyer-seller relationships but there is debate as to whether established models can be readily applied to online environments. This research adopts a relational perspective on trust in business-to-consumer environments by examining the effects of repeated usage of the Internet on consumer trust in the medium as a generic instrument for transacting business. It is hypothesised that trust in the Internet as an exchange medium is positively related to consumers’ experience of the medium.

Starting with established approaches to measuring trust, focus group analysis was used to refine and develop a multidimensional measurement approach which was then embodied in a questionnaire survey, administered by post and over the Internet. Responses from 470 Internet users are analysed and it is found that, along many dimensions, trust increased with increasing use of the medium.

Introduction

The concept of trust has long been discussed by marketers, building on an extensive body of knowledge developed within the discipline of sociology, psychology and economics (e.g. Gambetta, 1988; Fukuyama, 1995; Giddens, 1987; Misztal, 1996). Recently, there has been much discussion about the role of trust in Internet environments where the absence of face to face contact and other tangible cues can cause partners to an exchange to be uncertain about the trustworthiness of their potential exchange partner. Hoffman et al. (1997) noted that the primary barrier to the successful commercial development of the Internet is the current lack of consumer trust in this new commercial medium. Participation in a highly developed technological society demands a relatively high level of trust and confidence in the system due to the multiplication of dependency relationships. One must believe that vital goods and services will not be maliciously withheld and that their flow will not be disrupted by accident or incompetence. Many consumers have concerns about credit card fraud using the Internet, and so long as the perception of fraudulent use remains, the full potential of electronic commerce may not be achieved.

Trust has been conceptualised as a crucial antecedent and outcome of buyer-seller relationship development. A number of models of trust development within the context of relationship marketing have been proposed (Morgan and Hunt, 1994; Geyskens et al., 1998). However, it is widely reported that users of the Internet have high levels of distrust of the medium as an exchange mechanism (Morrison and Firmstone, 2000; Hoffman et al., 1998). The construct of trust has been widely interpreted and doubts raised about whether established models of trust can be readily applied to online environments, or whether new conceptualisations of trust are called for. Although efforts have been made towards a consensual definition of trust, some researchers have argued that the resulting conceptualisations are so ‘stretched’ that they have limited usefulness for conceptual and/or empirical work (Bigley and Pearce, 1998).

This paper takes a relational perspective on the degree of trust present in business-to-consumer (B2C) Internet environments by examining the effects of repeated usage of the Internet on consumers’ perceptions of trust. There is extensive evidence that trust results from the cumulative effects of interactions between partners to a relationship (Lewicki and Bunker, 1996; Shapiro et al., 1992).

This study is concerned with the relationship that exists between a consumer and the Internet as a generic instrument for transacting business. The research issue focused on the Internet as a transaction medium rather than the specific transaction partner. Of course, in reality it is difficult to separate the effects of the partner from the communication medium, for example the medium of the Internet may be perceived as being more trustworthy if a known and trusted partner was the focus of communication, rather than a partner whose trustworthiness had not yet been established. The subsequent research framework sought to isolate the effects of partners’ trustworthiness from trust in the medium by eliciting respondents’ general views about the medium rather than specific partners.

With this caveat, the hypothesis can be specified as:

A private consumer’s trust in the Internet as an exchange medium is positively related to the number of interactions that they experience with the medium.

The hypothesis was tested with respect to a sample of Internet users who have used the medium to purchase travel related services. There is considerable evidence of the widespread use of the Internet by consumers to purchase travel related services. The intangibility and commoditytype nature of many travel services make them ideal for distribution through the Internet. However, consumers buy on trust and cannot fully evaluate service delivery until their travel services have been consumed. Despite the existence of legislative and trade association guarantees against fraud, consumers remain vulnerable to fraudulent or negligent delivery of the services, which they requested. The characteristics of intangibility, commodity type products and vulnerability to non-delivery are shared by many other services (for example, financial and information services) thereby allowing some generalisability from this study to other service sectors.

Methodology

Trust has often been regarded a multidimensional construct, yet few studies have specifically looked at its multidimensional nature, preferring to focus on a unidimensional view of a very complex issue. A review of literature revealed ten conditions cited most frequently. They were:

* Availability – the goods and services promised are actually available for the purchaser to buy (Butler, 1991).

* Competence – the other party has the competence to deliver on his/her promises (Butler, 1991; Dwyer and Lagace, 1986).

* Consistency – a belief that the promised standard of delivery will not deviate from the norm (Butler, 1991; Larzelere and Huston, 1980).

* Discreetness (each party will respect the privacy of the other) (O’Malley et al., 1997).

* Fairness – the relationship between exchange partners will be governed by principles of equity (Rotter, 1971).

* Integrity – a belief that the partner will act with soundness and honesty (Morgan and Hunt, 1994; Butler, 1991).

* Loyalty – the other party will be faithful (Berry, 1993; Andaleeb, 1991; Butler, 1991).

* Openness – exchanges between partners will take place without the existence of hidden agendas (Butler, 1991; Gabarro, 1978).

* Promise fulfilment – promises made by the other party will be acted upon and goods and services delivered as promised (Anderson and Narus, 1990; Anderson et al., 1987; Crosby et al., 1990).

* Receptivity – the other party will react quickly to needs as they emerge (Butler, 1991; Fletcher and Peters, 1997).

Butler (1991) added an eleventh condition of overall trust, which was conceptualised as the cumulative effect of a number of dimensions of trust.

In this study, trust was measured using scales adapted from Butler’s ‘conditions of trust inventory’ (CTI) (Butler, 1991). Butler’s study represents one of the best frameworks for measuring trust in a specific manner. He developed a multidimensional measure of trust that best encapsulates all definitions and variances of trust mentioned in the literature. The original scales were developed from secondary data (Jennings, 1971; Gabarro, 1978) and a series of qualitative interviews in organisational settings. For each condition Butler (1991) developed four scale items and conducted a series of confirmatory analyses guided by Jackson’s (1984) four principles for scale construction and validity. His results indicated that the scales tapped nine different conditions of trust and that the a priori scales he developed had high alphas and test-retest reliabilities. Butler’s (1991) CTI was chosen over other trust-based scales as it focused exclusively on understanding and measuring conditions that lead to trust in a specific person. The distinction between conditions leading to trust, dimensions, and the antecedents and consequences of the phenomenon is an important one. In this study, conditions of trust refer to factors that are present in a buyer-seller relationship and are associated with the existence of trust. The conditions of trust inventory items do not in themselves make any presumptions about the processes by which trust is developed.

Butler developed 44 scale items, which were indicators of eleven emerged constructs describing conditions of trust. Although Butler’s scales were developed in a business-to-business context, it was believed that these scales would be a good starting point for measuring causes of trust in the context of consumers’ use of the Internet. It was therefore necessary to define the constructs in order to be in a position to search for relevant indicators, to write or select items. Having closely followed the contribution of Butler (1991), focus groups were held to assess the applicability of his conditions for measuring trust in Internet buying behaviour. Although the researcher did not want to deviate too much from the original scales, a certain amount of deviation was inevitable. Using Butler’s 11 conditions of trust, the focus groups were analysed so as to support the already established construct and content validity of the original scales. The focus groups were mainly concerned with understanding participant perception of online buying and to clarify exactly the meaning of the reworded items developed.

Given the inductive dimension to this phase of the research, the criteria for analysis were developed simultaneously. While there was an element of ‘analysis as you go’, it was necessary to develop an initial analytical framework. This framework was developed from both the predetermined and emergent topic areas from the literature. The focus groups were recorded on audiocassette and unabridged transcripts were then generated. The process of data analysis followed Wolcott (1994) in that it incorporated the three major phases of description (relying heavily on verbatim quotes from respondents), analysis (identifying important factors, themes and relationships) and interpretation (making sense of meanings in context). To end the session, and to aid in analysis, participants were asked to talk about the most important thing that was raised during the session.

The transcripts were content analysed under the original conditions. In an attempt to locate these major themes within the data, each interview transcript was read and re-read (Hill et al., 1999; Hill and McGowan, 1999). A separate file, which is similar to the diary method recommended by Easterby-Smith et al. (1991), was used as a means of recording insights into the data as they occurred. Once several readings of the data had taken place the process was temporarily halted and the file scanned through to check for emerging patterns. With these patterns in mind, the transcripts were read once more. The process was again halted, this time to attempt to bring together the various separate themes identified into one coherent pattern. Once a logical pattern had been established it was necessary to code data in some way.

The discussion guide was used as the basis of this coding frame. According to Hart (1989), coding is the process of flagging key issues within each transcript so that they can be compared. Colour coding as advocated by Griggs (1987) was deemed the most user friendly method. Each discussion topic was therefore given an identification colour.

The material was then reviewed to check for missing information or wrongly coded information. Once the researcher was satisfied the codes were correct, it was necessary to categorise these topics and sub-topics physically. This involved manually cutting and pasting the information. Analysis was carried out using Microsoft Word instead of computer assisted qualitative data analysis software (CAQDAS) packages such as NUD*IST or NVivo. The researcher did not believe in using quantitative vectors to analyse qualitative data, as these packages do. Using Microsoft Word, each topic area was brought together within a separate file. At the end of the third focus group it was decided that the data had reached saturation point.

Establishing indicators of reliability and validity in qualitative research is difficult, if not impossible with the exception of ecological validity which relates to the naturalness of the research approach in qualitative research (Bryman, 2001). Ecological validity therefore has parallels with Kirk and Miller’s (1986) adaptation of reliability and validity and Lincoln and Guba’s (1985) notion of trustworthiness. They state that each element of trustworthiness (credibility, transferability, dependability and confirmability) has a parallel with quantitative research criteria. Hammersley (1992) proposed the notion of relevance as a criterion to judge qualitative research. To establish the credibility of the qualitative findings in this phase of the research, a panel of experts was assembled. As a great deal of subjective judgement was involved in this part of the research (and indeed with qualitative research in general), it was noted that there was the possibility that there could be a lack of credibility or relevance in the coding and subsequent translation of qualitative transcripts. Therefore, a copy of the researcher’s qualitative findings and transcriptions were given to an independent person so as to verify the researcher’s perceived meaning of the statements. Therefore a panel of experts was asked to validate and to ensure that the newly revised items had construct and content validity (for a full account of the refinement process, see McCole, 2002).

In keeping with the results of the focus group analysis, all eleven constructs were used in the new instrument with between two and four scale items written for each. Seven causes of trust from Butler’s original study (availability, competence, consistency, discreetness, integrity, overall trust, and promise fulfilment) were used using slightly different wording of the items (see Appendix A).

Four of the causes of trust in the original study (fairness, loyalty, openness, and receptivity) were used as constructs only, with new scales borrowed/developed as appropriate based on focus group analysis and various previously validated scale items.

For example, it was decided to create a new scale for fairness because consumers regarded fairness in the Internet context differently to that intended in Butler’s original context (for example, fairness of pricing was not the meaning used in the original research). Several studies were consulted for scales that measured fairness. Kumar et al. (1995) fostered a few ideas for measurement criteria, many studies involving trust in marketing channels tended to deal with B2B commerce, and few drew on the work of Butler (1991). It was decided to develop new scales based on focus group transcriptions. Singh and Sirdeshmukh (2000) proposed that benevolence trust/distrust would influence pre-purchase price fairness. The main concern uncovered in focus groups was the availability and appeal of price differentials, which were, on occasions, not as good as conventional channels. Therefore, as an explanatory exercise, the current research project needed to draw correlations of price and trust effects, and the best way to do this is to develop new scales. A two-item, five-point Likert scale was developed.

It was decided to create a new scale for loyalty because loyalty to repurchase was not the meaning used in the original research. Several studies were consulted for scales that measured loyalty. Despite the plethora of loyalty scales consulted in various handbooks of marketing scales (for example, Bruner and Hensel, 1992; Bearden et al., 1993), few were successful because loyalty, like trust, is both attitudinal and conative and is best examined as a separate study. Loyalty is a vast construct and there are a lot of ongoing empirical studies of the phenomenon. The focus groups believed that branding was important for deciding to purchase online with a particular site, or the availability of that brand online. However, when purchasing a hospitality and tourism related commodity a certain degree of brand presence is expected. For example, all airline service principals operate under a brand name. Most accommodation providers operate under a brand, or are at least affiliated to a brand via consortium membership. Increasingly, Internet intermediaries are developing brands in their own right. Thus it seemed inappropriate to ask consumers how brand influenced trust, as there are too many brands and this is not the focus of this particular study. It is accepted that for purchases of hospitality and tourism related commodities brand consumption is unavoidable. A consumer’s evoked set, however, would contain various branded options. Other important points raised by the focus groups were that often, after a negative experience with one site, they would not use it again. This negative experience included poor information, technical difficulties or logistical problems of delivery and receipt. As loyalty involves an investment of one’s interest, it was decided to examine Internet loyalty in terms of the presence of a recognised brand, and how one ‘bad experience’ with the Internet would effect re-purchase intentions. It was decided that it would be best to devise new scales using a two-item, five-point Likert scale.

With regard to openness, participants felt that when purchasing online, especially for the first time, contact details of the supplier firm should be available on screen. Openness in the original questionnaire dealt with explicit one to one communication between manager and subordinate, and this was not the interpreted meaning in the new research. In addition to the focus groups, one particular study (Smith and Barclay, 1997) was useful for developing a new scale to measure openness of information perceived by consumers over different media relating to their purchase. The focus groups were concerned about the information content they received from different media. The vast majority of the focus group participants were happy with the content of information, especially that provided by EasyJet. It was again decided that it would be best to devise new scales using a two-item, five-point Likert scale. The openness scale that was developed therefore needed to examine the manner in which there was a provision of rele- vant information together with how responsive agents were to people who needed more information. Therefore a two-item scale of openness based on previous work by Smith and Barclay (1997) was developed.

Finally, scales for receptivity were adapted from a UK study conducted by Fletcher and Peters (1997)) which examined ‘offer receptivity’ and trust in direct marketing. This study also drew on Butler’s original scales. Items generated from the qualitative phase, including notes (where applicable) are presented in Appendix A.

There were issues raised during the focus groups such as ability to contact a tangible person via telephone, and influence of security initiatives on websites, that could not be easily ‘categorised’ into original scale items. It was decided to include these in the questionnaire under no particular condition, as additional questions that would yield interesting correlations in analyses.

In total six other items were added to the questionnaire after discussions with the supervision team and based on comments made by focus group participants. They included the effect that security Web initiatives such as ‘TRUSTe’ had on propensity to purchase (Web initiatives such as ‘Which? Web Trader’, ‘VeriSign’ (BT TrustWise), ‘ChamberSeal’, or ‘TRUSTe’ that specialise in addressing user privacy and security concerns are important to me for buying online). Another question that had to be included was the effect that recommendation had on the decision to use the Internet in the first place – ‘I would only use the Internet after it has been recommended to me by another person’. The third examined the number of people who would make a small purchase over the Internet before carrying out a large one – ‘I would carry out a small purchase over the Internet before buying a service which costs a lot of money’. The fourth examined the number of people who would ring the company to see if everything had been processed in the absence of a tangible receipt – ‘I would always ring the company providing the service to confirm that everything is OK after having bought online’. The fifth asked respondents whether they considered themselves to be a trusting type of person – ‘I would consider myself a trusting type of person’. The sixth examined the number of people who would abandon their online transactions because too much information had to be disclosed before the completion of a transaction – ‘Often I abandon my online purchase because of the amount of information I have to disclose about myself before completing a purchase online’.

One item was developed based on advice from Chircu et al.’s (2000) study to measure adoption intention. Propensity to purchase, which can be used as a proxy for actual purchase behaviour, was measured by one item based on verification from qualitative usage intention data from the focus groups – ‘I can see myself using this way of buying to purchase the majority of my future purchases of this service’.

Frequency of using the Internet was defined in terms of the number of occasions during the previous six months that a respondent had used the Internet to make purchases. Respondents were also asked about the place (for example, office, home) where most transactions took place.

The final survey instrument comprised 45 scale items used to measure trust. This was effectively a modified and extended conditions of trust inventory. Questionnaires were distributed via electronic and ‘traditional’ means through a probabilistic and quasi-probabilistic sampling frame respectively. Due to limitations of data protection legislation, cooperating companies administered the dispatch of questionnaires. 1,007 postal questionaires were dispatched, and this resulted in a 28 per cent response rate. This should be considered a good response rate, compared to typical response rates. The existence of a prior relationship between the cooperating company and potential respondents may have been important in increasing the response rate. An electronic questionnaire was hosted by an Irish low fares airline and generated 1,877 ‘hits’. A total of 188 electronic responses resulted in a 10 per cent response rate (based on the percentage of hits which were translated into completed surveys).

Overall the sample comprised 470 usable responses. A summary of the demographic characteristics of the sample is given in Table 1. It must be recognised that the survey samples possessed many characteristics of a convenience sample. This was probably especially true in the case of the Web-based survey, as it is not known how many people were exposed to the link which led to the questionnaire, and furthermore, it is not known whether the characteristics of respondents were representative of all visitors to the host site. However, the profile of the sample is comparable to other Internet studies (see amarach.ie; nua.ie, UK Office of National Statistics). The sampling methodology was felt to be sufficiently robust to allow inferences to be made about the conditions of trust as they applied to the sample studied. The survey was essentially concerned with the attitude as the unit of analysis.

Data from the two samples (Internet and postal questionnaire) was aggregated for analysis. This can be justified as no significant differences emerged between the two samples when assessed by responses to most of the scale item questions. The survey was administered during summer 2000 as part of a larger study.

Results

To test the hypothesis, the Kruskal-Wallis test was used to evaluate whether the population means of the dependent variable were the same across all levels of a factor. This technique was chosen over its parametric alternative (ANOVA) for two reasons. First, the distributional assumptions associated with parametric procedures were not met. Secondly, as the nature of the dependent variable is ordinal, it is preferable to conduct a Kruskal-Wallis or median test because the medians are a more meaningful measure of central tendency than means for ordinal data (Green et al., 2000).

Descriptive statistics classified respondents according to the number of transactions that they had undertaken using the Internet during the previous six months (see Table 1).

A number of notable differences were observed between respondents who had conducted ‘infrequent’ and ‘frequent’ transactions over the Internet. So that the results are useful and meaningful, they were arranged into a classification similar to that devised by Shapiro et al. (1992) and Lewicki and Bunker (1996). These authors argued that the development of trust occurs in stages with deterrence-based being the first and identification as the last stages having the highest trust. Ratnasingham (1998) believed that the development of trust is the same for all types of relationship be it romantic, manager-employee, or among peers and trading partners in electronic commerce. Thus this research attempts to classify respondents based on bivariate analyses into low-intermediate-high trusting tendencies. This research classifies transaction frequency into low (1-5 transactions); intermediate (11-15 transactions); and high (16+ transactions)

In many ways, respondents exhibiting low trust may be similar to the deterrence/calculus dimension by Shapiro et al. (1992) and Lewicki and Bunker (1996). Similarly respondents exhibiting intermediate trust may be similar to the knowledge-based dimension by Shapiro et al. (1992) and Lewicki and Bunker (1996). Finally, respondents exhibiting high trust may be similar to the identification-based dimension by Shapiro et al. (1992) and Lewicki and Bunker (1996).

Interestingly, there was variation between the dimensions of trust in the differences between these groups. The main results of this study are presented in Appendix B. Due to the wealth of data, only significant differences are presented. An interpretation of these results is presented next.

Low Trust

Respondents who considered themselves to be lowly skilled in using the Internet and who had conducted fewer than five transactions are considered to be typical of those exhibiting low trust towards Internet buying behaviour.

The rationale for this may be calculus based. Calculus based trust is often ensured both by the rewards of being trusting and by the threat that if trust is violated when one’s investments may not be honoured, then consumers are either going to be hesitant about purchasing online or simply will not have the courage to commit to the ‘purchase click’. This calculative process emphasises utilitarian considerations that enable one party to believe that another will be trustworthy, because the costly sanctions in place for breach of trust exceed any potential benefits from opportunistic behaviour (Ring and Van de Ven, 1992; Shapiro et al., 1992). As trust at its narrowest bandwidth, this calculative process has the qualities of low distrust/low trust where costs of breaching trust are high and the involvement between the parties is limited in the first place.

Let us look at the first type of consumers in the low trust classification. Respondents who have conducted fewer than five transactions may exhibit low trust tendencies due to the yet unfamiliar environment offered by Internet enabled e-commerce. These types of consumers:

* Did not believe that the Internet fulfilled their purchase requests competently.

* Did not believe that the Internet could be trusted.

* Did not believe that the Internet followed through on promises made.

* Did not believe that they could count on the Internet to be trustworthy.

* Did not believe that the Internet would not make false claims to them during purchasing actions.

* Did not believe that the Internet kept private information disclosed to it.

* Reported that they seldom knew what the Internet would do next in carrying out requests.

* Reported that they did not believe that confidential information was kept confidential via the Internet.

* Reported that they did not trust the Internet.

* Stated that sometimes they could not trust the Internet.

* Did not feel that the Internet did the same thing every time the situation was the same.

* Did not feel that the Internet did not tell others about things if they asked for it to be kept private.

* Stated that after one bad experience with the Internet they would be more cautious of using it again.

* Could not see themselves using the Internet to purchase future hospitality and tourism related commodities.

* Did not believe that the Internet offered services at the best price.

* Would only use the Internet after it was recommended by another person.

* Would carry out a small purchase first.

* Were more inclined to ring the company to confirm that everything was OK after having bought online.

* Were more inclined to abandon online purchases because of the amount of information needed to complete a transaction.

There are really no surprising consumer types in this category. These consumers may exhibit low trust for a variety of reasons. They may be the late majority or laggarts of society (Rogers, 1962) in terms of innovativeness. They may have a lower trusting disposition in terms of Web security, or may simply prefer traditional means of purchasing hospitality and tourism related commodities. However, it is likely that these consumers may move to other categories with time. For example, Kemp (2000) states that as people become experienced online buyers, their expectations of online vendors increase, in terms of service and the ‘buying experience’. As people demand more in the way of assurances for safety, security, and confidentiality of information, the industry continues to respond with better solutions. Efforts such as CPA WebTrust, Shop Safe!, TRUSTe and P3P are certain to expand and flourish. The study suggests that greater experience with a website is associated with lower trust and higher risk in a particular Internet agent.

Intermediate Trust

Interestingly, respondents who conducted between 11 and 15 transactions are considered to be typical of those exhibiting low trust towards Internet buying behaviour. The rationale for this may be knowledge-based and is consistent with rational choice motivations. This is grounded in the knowledge of the other trading partner (Internet agent or agents) which allows the trustor (consumer) to understand and predict the behaviour of the trustee. The key factor in this type of trust is the information derived over time that allows one trading partner to predict the behaviour of another trading partner. It is reasonable to assume that these types of respondents may accept the possibility of Internet opportunism, but are equally aware that they can have recourse to ‘institutional safety’ (credit card protection, company guarantee, etc.). This is similar to Fukuyama’s (1995) belief in fiduciary responsibility. Similarly, in this vein Luhmann (1979) has proposed that the development of institutional trust acts as a mechanism for the reduction of complexity and stated that interpersonal mechanisms are no longer adequate for trust production in such a society.

Respondents who completed between 11 and 15 transactions:

* Did not believe that the Internet could be trusted as much as respondents who conducted 16-20 transactions.

* Seldom knew what the Internet would do next in carrying out requests compared with respondents who conducted 16-20 and 21+ transactions.

* Believed that the Internet did not deal with their requests in a consistent manner to a higher degree than respondents who conducted 16-20 transactions.

* Stated that they could not trust the Internet as much as respondents who conducted 16-20 transactions did.

* Stated that after one bad experience with the Internet they would be more cautious than did respondents who conducted 16-20 transactions.

* Were more inclined to ring the company to confirm that everything was OK after having bought online than respondents who conducted 16-20 transactions.

Consumers in this category may exhibit intermediate trust for a variety of reasons. They may use the Internet more regularly because it is part of their daily lives. Thus consumer convenience may be an important influence. However, there is evidence that they are aware of potential risks and are confident that they have recourse to institutional safeguards. These consumers are also unlikely to have loyalty towards the medium, they display fickle habits in terms of purchasing via Internet enabled e-commerce and are likely to have hybrid marketing channels in their evoked consideration set. They may also represent the early majority society (Rogers, 1962) in terms of innovativeness.

High Trust

Respondents who conducted more than 16 transactions are considered to be typical of those exhibiting high trust towards Internet buying behaviour.

The rationale for this may be identification-based. This is based on empathy and common values with the other trading partner’s desires and intentions, to the point that one trading partner is able to act as an agent (in terms of recommending that Internet site (part-time marketer)) over time. Consumers exhibiting high trust in Internet buying behaviour are those who have repeated interactions over time. Consumers in this category have a long history of interaction with this medium and have often been in the other two categories before being ‘promoted’ to this category. High trust may be the result of objective/subjective assessments on the behaviour of the Internet agent and the history of interactions with it. Consumers in this classification are believed to be the most loyal to purchasing over the Internet through previous satisfaction and previous consumption of the services bought. This high trust tendency is related to the way in which any given transaction is carried out from which the customer consciously evaluates the service provided against pre-determined standards and expectations. Andaleeb (1992) referred to this dimension as the firm’s ‘motive or intent’. Gronroos (1994) believed that in relationship marketing this functional quality dimension would grow in importance.

As stated, consumers in this classification will not just exhibit high trusting tendencies. Respondents would start with one transaction and then judge the satisfaction of that against personal expectations. This would occur several times. Reliability and dependability in previous interactions with the trustor give rise to positive expectations about the trustee’s intentions. Emotion enters into the relationship between the parties, because frequent, longer-term interaction leads to the formation of attachments based upon reciprocated interpersonal care and concern (McAllister, 1995). For this reason, scholars often refer to this form of trust as ‘affective trust’ (McAllister, 1995) and as ‘identity-based trust’ at its broadest scope (Coleman, 1990).

High trust therefore is built through ‘spiral reinforcement’, in that low trust can move from being fragile to being resilient (high trust) based on post-purchase evaluation. Consumers in this type of trust may depend more on symbolic and less formalised types of governance (Etzioni, 1988; Ouchi, 1980). Granovetter (1985) suggested that these are antecedents of the kinds of social relations that he describes as necessary conditions for trust and trustworthy behaviour. In turn, high levels of trust within a society or community may be accompanied by less formal control systems. Thus consumers in this classification tend to be less concerned with security issues and exercise habitual trust in their purchasing episodes.

Specifically, respondents who have conducted more than 16 transactions:

* Believed that the Internet fulfils their purchase requests competently.

* Believed that the Internet could be trusted.

* Believed that the Internet followed through on promises made.

* Believed that they could count on the Internet to be trustworthy.

* Believed that the Internet would not make false claims to them during purchasing actions.

* Believed that the Internet kept private information disclosed to it.

* Knew what the Internet would do next in carrying out requests.

* Believed that confidential information was kept confidential via the Internet.

* Trusted the Internet.

* Believed that the Internet dealt with their requests in a consistent manner.

* Felt that the Internet does the same thing every time the situation is the same.

* Felt that the Internet does not tell others about things if they asked for them to be kept private.

* Believed that after one bad experience with the Internet they would be willing to try it again.

* Could see themselves using the Internet to purchase future hospitality and tourism related commodities.

* Believed that the Internet offered services at the best price.

* Would carry out large purchases.

* Were not inclined to ring the company to confirm that everything was OK after having bought online.

* Were not inclined to abandon online purchases because of the amount of information needed to complete a transaction.

Consumers in this category may exhibit high trust for a variety of reasons. They may be inclined to use the Internet more regularly because it is part of their daily lives. Thus consumer convenience may be an important influencer on these consumers. These types of consumers are also likely to have loyalty towards this medium, and be more forgiving in the event of an undelivered promise. They display habitual trust in terms of purchasing via Internet enabled e-commerce and are likely to have Internet only marketing channels in their evoked consideration set. These types of consumers are likely to enjoy Internet shopping and have sufficient skills (optimal stimulation levels) and means to use the medium more frequently. They may also represent the innovators and/or early adopters in society (Rogers, 1962) in terms of innovativeness.

‘Grey Areas’

Respondents who conducted between 6 and 10 transactions were difficult to classify as there were non-consistent answers to individual items measuring trust in Internet buying behaviour in relation to this segment of respondent profile. This classification tended to exhibit intermediate to high trust tendencies, but seemed to be sceptical about various websites (‘I seldom know what this way of buying will do next in carrying out my requests’ (z = -2.01; p = .041)) compared to respondents who had conducted more than this amount. Further, respondents in this profile would be more cautious of using the Internet in the future to buy a similar service (z = -2.20; p = .028) compared to respondents who had conducted more than this amount. Finally, respondents in this profile did not believe that it was easier for them to find a service at the best price using the Internet (z = -2.03; p = .042) to a higher degree than respondents who conducted more than this amount (specifically those who conducted more than 21 transactions). It is likely that this classification is in the ‘spiral reinforcement stage’, whereby the future of their Internet use is unclear. The results suggest that this classification may either resort to low trust tendencies or be promoted to either stable intermediate or high trust tendencies in Internet buying behaviour in their evaluation of future interactions.

Discussion and Management Implications

Mistrust has been frequently cited as inhibiting a more widespread adoption of the Internet for business-to-consumer transactions. Newspapers have published stones of credit card numbers being intercepted and used fraudulently. Although the chances of this happening are low – and probably much lower than fraud, which results from traditional credit card transactions – consumers’ perceptions of the trustworthiness of the Internet may be at variance with reality. Mistrust in many situations (for example, personal relationships, commercial rivalry, diplomatic relations between countries) may be based on fear of the unknown. This research would suggest that mistrust in the Internet might be based on respondents’ lack of familiarity with the medium. It is possible that as consumers become more familiar with the benefits of undertaking transactions through the Internet, benefits figure more prominently than the perceived risks. This study has shown a relationship between the level of trust and level of experience of using the Internet.

Of the 45 scale items used to measure causes of trust, 20 showed significant differences between respondents who exhibited low-high trust tendencies. Most of these items are newly developed items as presented in Appendix A. This modified CTI represents a new insight into causes of trust in the Internet domain. Butler’s (1991) study provided reliable constructs that needed to be included to measures causes of trust. Thus, Butler’s work provided theoretical reliability, but ecological validity (Bryman, 2001) was established during the refinement phase and as such comparison between the original and modified CTI would be unfair.

This study has produced some evidence to confirm that repeated exposure to the Internet as a means of conducting transactions increases users’ level of trust in the medium. Although it was not within the scope of this study to examine the psychological and behavioural processes by which trust is developed, the results are consistent with previous studies of trust which have shown it to be a developmental concept. In this sense, trust in Internet environments is similar to trust in more traditional environments. However, the processes by which trust is developed may differ between the two. Internet environments in themselves present relatively few cues by which trust can be assessed, for example body language and having sight of products being offered are absent. It is also likely that different types of product are associated with different levels of trust. The Internet has been used extensively to distribute intangible services, where the relative absence of tangible goods can give a competitive advantage over more traditional distribution channels. However, purchases of such services involve a much greater element of faith and vulnerability on the part of the buyer than is the case with tangible commodity goods whose properties (or replicas of them) can be examined prior to purchase.

It must be recognised that distinguishing trust in the medium from trust in the exchange partner can be difficult. This study focused questioning on the medium and it must be accepted that responses may have been influenced by an individual’s predisposition towards a specific exchange partner. However, across the range of respondents, this study has found significant differences between experienced and inexperienced Internet users in their levels of trust. It may be assumed that the trustworthiness of exchange partners was randomly distributed across inexperienced and experienced groups, so the observed effects are due to trust in the medium.

The study has been unable to say whether prior trust in a partner increases trust in the medium. For companies seeking to migrate their existing face-to-face customers to an online environment, an existing trusting relationship with the company may be one factor which facilitates trust in the medium. However, this research would suggest that more experience of using the Internet per se is likely to be effective in increasing levels of trust.

Further analysis of the data set is under way to identify further antecedents of trust in Internet trading environments, in particular the influence of demographic characteristics and product type.

Given the observed association between experience of Internet use and trust in the Internet as a medium, it might appear that the best long-term strategy for companies to develop trust in their Internet based trading systems is to increase the general level of Internet usage, whether with their own websites or those of other organisations.

While the development of greater general Internet experience is beyond the capability of most companies, there are a number of methods which a company can use to take site visitors through the stages of gaining experience and thereby trust. Essentially, websites should seek to give users experience prior to commitment.

In the case of inexperienced Internet users, who by implication may be the least trusting, website marketers should:

* Promote onsite navigation with user friendly graphics and appropriate content.

* Develop a Web adviser to help consumers during a purchase episode that consumers can choose to use or not.

* Encourage repeat transactions by providing open, unbiased, relevant and clear information.

* Incorporate customer testimonials, and provide contact details.

* Market the Internet company as if it is actually managed by people.

* Provide receipt and authentic proof of purchase.

* Incorporate trust seals (VeriSign, TRUSTe).

* Inform customers of tracking technology.

* Provide details of what will be done with information gathered.

* Offer a facility to build relationships, i.e. adopt a permission marketing approach.

* Provide guarantees.

For consumers with intermediate levels of experience, website marketers should continue this trust building by additionally offering:

* Website profiling based on number of transactions, skill, and sex (Nairn and Evans, 1998).

* Involve the consumer in online discussion groups.

* Encourage consumers to voice their opinions.

* Provide search and comparison facilities for alternative commodities.

* Provide facilities for ‘one-stop’ shopping.

* Provide incentives to return (some sort of loyalty initiative).

For experienced users of the Internet, website marketers should additionally:

* Offer personalised service webpages.

* Recommend present and future purchases based on interaction history.

* Provide complementary services with affiliates.

* Involve the consumer in the future direction of the company.

References

Andaleeb, S.S. (1991), ‘Trust and dependence in channel relationships: implications for satisfaction and perceived stability”, in AMA Summer Educators Conference Proceedings, American Marketing Associaton, Chicago, pp. 249-50.

Anderson, E., L.M. Lodish and B.A. Weitz (1987), ‘Resource allocation behaviour in conventional channels’, Journal of Marketing Research, vol. 24, February, pp. 85-97.

Anderson, J.C. and J.A. Narus (1990), ‘A model of distributor firm and manufacturer firm working partnerships’, Journal of Marketing, vol. 54, January, pp. 42-58.

Bearden, W.O., R.G. Netemeyer and M.F. Mobley (1993), Handbook of Marketing Scales, Sage, London.

Berry, L., (1993), ‘Improving America’s service’, Marketing Management, vol. 1, no. 3, pp. 28-38.

Bigley, G.A. and J.L. Pearce (1998), ‘Straining for shared meaning in organisational science: problems of trust and distrust’, Academy of Management Review, vol. 23, no. 3, pp. 405-21.

Bruner, G.C. and P.J. Hensel (1992), Marketing Scales Handbook, AMA, Chicago.

Bryman, A. (2001), Social Research Methods, Oxford University Press, New York.

Butler, J.K. (1991), ‘Toward understanding and measuring conditions of trust: evolution of a conditions of trust inventory’, Journal of Management, vol. 17, no. 3, pp. 643-63.

Chircu, A.M., G.B. Davis and R.J. Kaufman (2000), ‘Trust, expertise, and e-commerce intermediary adoption”, Proceedings of the Americas Conference on Information Systems, Long Beach, CA, 10-13 August.

Coleman, J.S. (1990), Foundation of Social Theory, Belknap Press, Cambridge, MA.

Crosby, L.A., K.R. Evans and D. Cowles (1990), ‘Relationship quality in services selling: an interpersonal influence perspective’, Journal of Marketing, vol. 54, July, pp. 68-81.

DeVaus, D.A. (1996), Surveys in Social Research, 4th edn, UCL Press, London.

Dwyer, F.R. and R.R. Lagace (1986), ‘On the nature and role of buyer-seller trust’, in T. Shimp et al. (eds), AMA Summer Educators Conference Proceedings, American Marketing Association, Chicago, pp. 40-5.

Easterby-Smith, M., R. Thorpe and A. Lowe (1991), Management Research, An Introduction, Sage, London.

Etzioni, A. (1988), The Moral Dimension: Towards a New Economics, The Free Press, New York.

Fletcher, K.P. and L.D. Peters (1997), ‘Trust and direct marketing environments: a consumer perspective’, Journal of Marketing Management, vol. 13, pp. 523-39.

Fukuyama, F. (1995), Trust: The Social Virtues and Creation of Prosperity, The Free Press, New York.

Gabarro, J.J. (1978), ‘The development of trust influence and expectations’, in Athos, A.G. and J.J. Gabarro (eds), Interpersonal Behaviour: Communication and Understanding in Relationships, Prentice Hall, Englewood Cliffs, NJ.

Gambetta, D. (1988), Trust: Making and Breaking Co-operative Relationships, Basil Blackwell, Oxford.

Geyskens, I., J.-B.E.M. Steenkamp and N. Kumar (1998), ‘Generalisations about trust in marketing channel relationships using meta-analysis’, International Journal of Research in Marketing, vol. 153, pp. 223-48.

Giddens, A. (1987), Social Theory and Modern Sociology, Polity Press, Cambridge.

Granovetter, M. (1985), ‘Economic action and social structure: the problem of emdeddedness’, American Journal of Sociology, vol. 91, pp. 481-510.

Green, S.B., N.J. Salkind and T.M. Akey (2000), Using SPSS for Windows, Analysing and Understanding Data, 2nd edn, Prentice Hall, Englewood Cliffs, NJ.

Griggs, S. (1987), ‘Analysing qualitative data’, The Journal of the Market Research Society, vol. 32, pp. 507-20.

Gronroos, C. (1994), ‘From marketing mix to relationship marketing: towards a paradigm shift in marketing’, Management Decision, vol. 32, no. 2, pp. 4-20.

Hammersley, M. (1992), What’s Wrong with Ethnography? Methodological Explorations, Routledge, London.

Hart, S.J. (1989), ‘Collection and analysis of interview data’, Marketing Intelligence and Planning, vol. 7, no. 5, pp. 23-8.

Hill, J., P. McGowan and P. Drummond (1999), ‘The development and application of a qualitative approach to researching the marketing networks of small firm entrepreneurs’, Qualitative Market Research, An International Journal, vol. 2, no. 2. pp. 71-81.

Hill, J. and P. McGowan (1999), ‘A qualitative approach to developing small firm marketing planning competencies’, Qualitative Market Research, An International Journal, vol. 2, no. 3. pp. 167-75.

Hoffman, D.L., T.P. Novak and M.A. Peralta (1997), ‘Information privacy in the marketplace: implications for commercial uses of anonymity on the web’, online, available at http://www2000.ogsm.vanderbilt.edu.

Hoffman, D.L, T.P. Novak and M. Peralta (1998), ‘Building consumer trust in online environments: the case for information privacy’, online, available at http://www2000.ogsm.vanderbilt.edu.

Jackson, D.N. (1984), Personality Research Form Manual, 3rd edn, Research Psychology Press, Port Huron, MI.

Jennings, E.E. (1971), Routes to the Executive Suite, McGraw-Hill, New York.

Kemp, R. (2000), ‘Whom do you trust?’, online, available at http://www.tmcnet.com.

Kirk, J. and M.L. Miller (1986), Reliability and Validity in Qualitative Research, Sage, Newbury Park, CA.

Kumar, N., L.K. Scheer and J.-B.E.M. Steenkamp (1995), ‘The effects of supplier fairness on vulnerable re-sellers’, Journal of Marketing Research, vol. 32, February, pp. 54-65.

Larzelere, R.E. and T.L. Huston (1980), ‘The dyadic trust scale: toward understanding interpersonal trust in close relationships’, Journal of Marriage and the Family, vol. 42, August, pp. 595-604.

Lewicki, R.J. and B.B. Bunker (1995), ‘Trust in relationships: a model of development and decline’, in Conflict, Cooperation and Justice, Bunker, B.B. and J.Z. Rubin (eds), Jossey-Bass, San Francisco.

Lewicki, R.J. and B.B. Bunker (1996), ‘Developing and maintaining trust in work relationships’, in Kramer, R.M. and T.R. Tyler (eds), Trust in Organisations: Frontiers of Theory and Research, Sage Publications, Thousand Oaks, CA, pp. 114-39.

Lincoln, Y.S. and G. Cuba (1985), Naturalistic Inquiry, Sage, London.

Luhmann, N. (1979), Trust and Power, John Wiley, New York.

McAllister, D.J. (1995), ‘Affect and cognition based trust as foundations for interpersonal co-operation in organisations’, Academy of Management Journal, vol. 38, pp. 24-9.

McCole, P. (2002), ‘The role of trust for electronic commerce in services’, International Journal of Contemporary Hospitality Management, vol. 14, no. 2, pp. 81-7.

Misztal, B.A. (1996), Trust in Modern Societies, Polity Press, Cambridge.

Morgan, R.M. and S.D. Hunt (1994), ‘The commitment-trust theory of relationship marketing’, Journal of Marketing, vol. 58, no. 3, pp. 20-39.

Morrison, D.E. and J. Firmstone (2000), ‘The social function of trust and implications for e-commerce’, International Journal of Advertising, vol. 19, pp. 599-623.

Nairn, A. and M.J. Evans (1998), ‘Direct MailShots: the gender effect’, proceedings of the 1998 Annual Conference of the Academy of Marketing, Adding Value to Marketing, Sheffield Hallam University, pp. 289-332.

O’Malley, L., M. Patterson and M. Evans (1997), ‘Intimacy or intrusion? The privacy dilemma for relationship marketing in consumer markets’, Journal of Marketing Management, vol. 13, pp. 541-59.

Ouchi, W.G. (1980), Markets, bureaucracies, and clans’, Administrative Sciences Quarterly, vol. 25, pp. 124-41.

Ratnasingham, P. (1998), ‘Trust in web-based electronic commerce security’, Information Management and Computer Security, vol. 6, no. 4, pp. 62-166.

Ring, P.S. and A. Van de Ven (1992), ‘Structuring co-operative relationships between organisations’, Strategic Management Journal, vol. 13, pp. 483-98.

Rogers, E.M. (1962), Diffusion of Innovation, 3rd edn, The Free Press, New York.

Rotter, J.B. (1971), ‘Generalised expectancy for interpersonal trust’, American Psychologist, vol. 26, pp. 443-52.

Shapiro, D., B.H, Sheppard and L. Cheraskin (1992), ‘Business on a handshake’, The Negotiation Journal, October, pp. 288-308.

Singh, J. and D. Sirdeshmukh (2000), ‘Agency trust mechanisms in customer satisfaction and loyalty judgements’, Academy of Marketing Science, vol. 21, no. 1, pp. 150-67.

Smith, J.B. and D.W. Barclay (1997), ‘The effects of organisational differences and trust on the effectiveness of selling partner relationships’, Journal of Marketing, vol. 61, January, pp. 3-21.

Wolcott, H.F. (1994), Transforming Qualitative Data: Description, Analysis and Interpretation, Sage, Thousand Oaks, CA.

Authors

Patrick McCole is a lecturer at Dublin City University Business School, Ireland.

Adrian Palmer is professor of marketing at University of Gloucestershire, England.

Copyright Mercury Publications Ltd. 2002

Provided by ProQuest Information and Learning Company. All rights Reserved