Building Trust with Consensus Information:
The Effects of Valence and Sequence Direction
Ray L. Benedicktus, Melinda L. Andrews
Florida State University
Abstract
Sellers' reputation, as conveyed by consensus information, influences consumers' trust in the company. Two experiments were used to investigate the effects of consensus information on consumer trust. Consensus trends were manipulated over three reporting periods. The results indicated that trust varies positively with the valence of a firm's reputation and is similarly dependent upon whether the trend decreases or increases over time. Effects of valence and sequence direction were further explored in conditions wherein variation between periods was attenuated. Results showed that valence continued to drive trust. However, effects of sequence direction ceased to be significant when the change in reputation between periods was reduced.
Introduction
Widespread adoption of the Internet as a retail channel has caused online purchasing by both businesses and consumers to become more frequent (Barrett and Pugh 2003; Faloon 2001; Sheehan and Doherty 2001). A survey conducted by Yankelovich Partners found that 93 percent of Internet users have researched products online, and 85 percent have purchased products online (Pastore 2000). Nearly two-thirds of North American consumers "cross channels" when they shop, researching online then purchasing offline in a store or over the phone (Chatham 2004). Additionally, the Internet has widely expanded consumer access to information and created new methods by which consumers can share information with others. Thus, even with the profound growth in e-commerce, the Internet functions largely as an information source (Goldsmith and Bridges 2000).
Consumers have access to information on brick-and-mortar, e-tailer, and hybrid firms via electronic feedback mechanisms. These mechanisms essentially aggregate consumer responses to single or multiple scale items that measure factors such as the valence of transaction outcomes, cleanliness of the service environment, and employee friendliness/responsiveness. This information is provided by countless transaction intermediaries (e.g. epinions.com, e-bay.com), consumer reporting agencies and watchdog organizations (e.g. American Consumer Satisfaction Index, Consumer Reports, Better Business Bureau), and can also be found in magazine articles and advertising messages (e.g. 73% of women say they want, 4 out of 5 dentists agree). Feedback information is often represented as a time series (e.g. monthly for three months, annually for five years). Although the formats vary from mean and percentage scores to star ratings and histograms, each method conveys the information in the form of a consensus of consumer responses. Consensus information aims to increase consumer trust in virtuous companies and give buyers a tool for identifying risky transactions.
The valence of the consensus information and its perceived change over time are thought to have a profound influence on the attitudes and beliefs of consumers. Although firm reputation has been linked to price premiums and sales (Lucking-Reiley, Bryan, and Reeves 2000; Resnick and Zeckhauser 2002), the relationship between the sequence direction and trust, and the interaction of consensus and sequence direction on trust, have not yet been empirically analyzed and tested.
We conducted a pilot study to lay a foundation for experimental manipulations and subsequently tested the aforementioned effects at two levels of consensus, varying the direction of change in consensus over three consecutive reporting periods. A second experiment provides some insights into the effects of these directional changes. We now turn to a brief review of the literature to provide the background for the present research.
Conceptual Background
Much of the foundation for this paper is derived from previous findings involving the effects of firm reputation on consumer attitudes and behavior. In addition, attribution and prospect theories and responses to outcome sequences also provide support for our hypotheses.
Consumer Trust
Trust is a key element in successful marketing, a vital antecedent for purchasing, and a crucial element of relationship marketing, especially in online environments where interaction between buyers and sellers is at a minimum (Gefen and Straub 2003; McCole 2002; Morgan and Hunt 1994). It is widely accepted that consumers make buying decisions on the basis of trust (Brannigan and de Jager 2003; Koehn 2003; Urban, Sultan, and Qualls 2000). Similarly, the factors that affect consumers' willingness to participate in online transactions are primarily related to trust (Doney and Cannon 1997; Farrell, Leung, and Farrell 2000; Jarvenpaa, Tractinsky, and Vitale 2000). Consumer concerns and perceived risks are much higher in online transactions than in brick-and-mortar purchasing, so much that a very high level of trust is needed to stimulate increases in online spending (Hoffman, Novak, and Peralta 1998; Jarvenpaa, Tractinsky, and Vitale 2000). Many companies have failed in their Internet marketing efforts primarily because they failed to build trust (Urban, Sultan, and Qualls 2000). Another possible reason for these failures is that knowledge of the manifestation of trust in online environments is still being developed.
Trust has been described as an attitude, relationship, measure of expectations, and cognitive belief. A slight modification of Koehn's (2003) definition would suggest that trust is the expectation of being benefited by another party. Consumers and companies have the potential for engaging in at least a temporary relationship and the firm has the ability to betray these benefit expectations. A similar definition, proposed by Geyskens et al. (1996) presents trust as the expectation that the merchant's word can be depended upon and that the seller will not take advantage of the consumer's vulnerability. Trust is also believed to be "the extent to which two parties are willing to rely upon each other" (Doney and Cannon 1997).
The present study focused on a hybrid of calculative (information-based) and knowledge-based trust, which are not entirely distinct concepts (Koehn 2003). Consumers engaging in calculation-based trust will attempt to predict a seller's intentions using evidence of the seller's trustworthiness and by examining the seller's history of fulfilling orders on time and its overall reputation. The consumer then considers the benefits and potential liabilities of trusting the seller and makes a decision whether to trust based on potential gains and losses. This form of trust is most common in relationships in which parties are unfamiliar, and is quickly violated when there is even a minor breach in expectations. Alternatively, consumers that use a more knowledge-based criterion for determining trusting beliefs are typically more familiar with the firm and are likely to be more familiar with similar purchasing situations (e.g. Internet buying). Consumers essentially rely on their knowledge of the firm and the purchase environment when evaluating the likelihood that their expectations will be met. The merging of these concepts occurs when consumers are given the benefit of knowledge in the form of others' collective experiences with a particular seller (Koehn 2003). If information on the firm is absent in a given transaction, it is impossible for consumers to develop any sense of trust (Gefen 2000). Consumers will avoid purchasing from firms that demonstrate an inadequate capacity for meeting consumer expectations or that fail to create trust (Jarvenpaa, Tractinsky, and Vitale 2000).
Linking Firm Reputation to Trust and Expectations
Several studies in the marketing literature link firm reputation to consumers' trust and expected outcomes. Trust has been shown to be more prevalent when vendors are perceived as having a reputation for fairness (Anderson and Weitz 1989; Dasgupta 1988; Ganesan 1994, Seiders and Berry 1998). Conversely, a negative reputation has been shown to reduce trust (Anderson and Weitz 1992). Expected outcomes and pre-purchase beliefs can be based on many sources of information, including expert opinion, word-of-mouth, publicity, and firm communication (Boulding et. al 1993; Olson and Dover 1979).
In Boulding et al.'s (1993) Dynamic Process Model, consumers are thought to form two distinct types of expectations, what will happen and what should happen. Should happen expectations are based on what the consumer believes to be reasonable and feasible, and are often based on communication from the firm. Will happen expectations are representative of an initial judgment, which is then adjusted by the consumer's integration of past experiences. When consumers lack the ability to integrate their own experiences they may use aggregated evaluations of other consumers as a surrogate, similar to the manner in which consumers seek advice from family and social groups when attempting to reduce purchase risk.
In new buying situations, important information, such as satisfaction judgments of prior experiences, is often missing. Previous findings in consumer inference research suggest that missing information may be estimated via consumers' inferences of available information (Brown and Dacin 1997; Dick, Chakravarti, and Biehal 1990; Lynch Marmorstein, and Weigold 1988; Simmons and Lynch 1991). Thus, trustworthiness can be inferred from a firm's reputation for acting in a trustworthy fashion (i.e. reliably satisfying customers, keeping its promises).
Consumer Attributions: Valenced Outcomes
Although attribution theory is often used to explain the past, its "real value is in its ability to influence the future" (Forlani and Walker 2003). People make constant judgments about the likely causes of events (Einhorn and Hogarth 1986). These ‘attributions' guide decisions between alternative courses of action. Attribution theory seeks to explain how individuals make decisions when given information about prior outcomes (Weiner 1985), and is explored herein to answer the question, "To what extent, do outcomes reported by a firm's customers influence subsequent consumers' trust in that firm?"
Attributions are inferences or explanations that individuals generate to account for previous outcomes (Heider 1958; Kelley 1967; Weiner 1985). The valence of an outcome indicates the outcome's "degree of goodness" (Weiner 1985), where positively valenced outcomes indicate at least acceptable performance and negatively valenced outcomes indicate less than acceptable performance. Past research has shown that extremely high and low valenced outcomes provide reinforcement of an outcome's stability, thereby supporting the assumption that future outcomes will emulate the outcomes under consideration (Anderson, Kulhavy, and Andre 1971). Outcomes with low valence prompt consumers to avoid engaging in similar behavior to that of consumers having less than satisfactory experiences with a firm (Matsui, Kakuyama, and Onglatco 1987). A consumer considering a transaction with a firm having a reputation of low performance should be less likely to trust the firm to fulfill his/her needs than a consumer considering purchasing from a firm with a much better reputation. Hence,
H1: High (low) consensus increases (decreases) consumer trust in the firm.
Sequences of Outcomes
Sequences of outcomes are evaluated by decision makers based on reference points (Kahneman and Tversky 1979), which can be part of the sequence under consideration or provided by an outside criterion. Some reference points are located at the beginning of sequences, as in principal amounts of investments or a predetermined target rate of return. Other reference points may be located in the middle or at the end of a sequence, such as in reputation sequences where consumers may be concerned with only the most recent shifts in firm performance. This recency effect (Anderson 1981; Miller and Campbell 1959) suggests that consumers weigh more heavily the information in recent periods, because final outcomes in a sequence are more salient to decision makers (Varey and Kahneman 1990).
Loewenstein and Prelec (1993) found that people evaluate sequences involving levels of positive and negative outcomes as upward and downward shifts. The loss aversion principle of prospect theory suggests that downward shifts receive disproportionate weight. Thus, people will typically prefer sequences of outcomes that improve over time to those that depict a decline (Loewenstein and Sicherman 1991; Ross and Simonson 1991). It follows that consumers will have a ‘preference for happy endings' in that they will have higher trust in a firm whose performance sequences improve over time than in a firm with a diminishing reputation. This proposition is further supported by Matsui, Kakuyama, and Onglatco's (1987) conclusion that consumers avoid purchasing from unsatisfying firms, especially when there is information indicating that satisfaction with the firm is decreasing over time. Thus,
H2: Consumers' trust will be higher for firms having improving consensus over the three reporting periods than for firms having decreasing sequences.
Prospect theory (Kahneman and Tversky 1979) also proposes that consumer judgments and attributions are made with a "diminishing sensitivity" in that the marginal impact of both negative and positive prior outcomes decreases as the number of similarly valenced outcomes increases. Thus, when the magnitude of changes between each reporting period are such that there is little perceived difference in reputation between the periods, the sequence direction effects of H2 are likely to be attenuated, or may even cease to exist. Therefore, we propose:
H3: Attenuation of the change in consensus between reporting periods will reduce the effects hypothesized in H2.
Research Method
The Pilot Study
A pilot study was conducted with the goal of gaining a deeper understanding of consumer perceptions of consensus scores. Specifically, a between groups questionnaire was given to a convenience sample of 60 undergraduate marketing students from a large Southern U.S. University. Each respondent was given a questionnaire that featured an Internet purchasing scenario involving either an e-tailer or a hybrid firm. More precisely, respondents were asked to consider a firm that either "sells on the Internet, but does not have any retail locations," or "also sells offline through a chain of retail stores." The scenario asked respondents to think about what percentages would best represent various levels of satisfaction with the respective firm type. The remainder of the questionnaire was identical across conditions. Following the scenario, respondents filled in a table inquiring as to the percentages that best represented an ideal, desired, average/typical, and minimally acceptable satisfaction. Next, the respondents were asked to quantify the categories on a more visual measure, depicting each of the satisfaction levels on a Y-axis. These techniques allowed us to estimate appropriate manipulations with a method called absolute magnitude estimation (Gescheider 1988). Analysis was conducted after removing one subject that could not recall the type of firm discussed in the scenario.
The results of the pilot study yielded confidence intervals for the means of each satisfaction level. The confidence intervals for these means were very similar for both types of firms. Ideal satisfaction was perceived as being represented by consensus percentages in the mid to high 90's, respondents desired to do business with firms having at least 84-88% positive evaluations, and the typical/average Internet firm was shown to have a score in the mid to high 70's. The results are consistent with logic that consumers desire to do business with firms having above average reputations, but often engage in transactions with firms having less than an ideal score. Additionally, respondents indicated that the lowest acceptable satisfaction percentage varied between the mid 60's and low 70's.
Overview of the Experiments
Using the results from the pilot study, consensus sequences were formed for the experiments. The goals of the manipulations were to: (1) replicate consensus perspectives observed in the pilot study across sequences that could be reasonably expected to occur in the marketplace, and (2) examine whether trust can be causally linked to sequence characteristics.
The first experiment was designed to test the first two hypotheses. Consensus sequences were required that varied substantially in their overall valence and wherein respondents would be likely to detect a change in consensus over the periods included in the sequence. Three time periods were presented to subjects as the three most recent months to the month of data collection. The second investigation retests H1, but was primarily designed to test H3. In this design, changes in the valence of consensus were reduced to vary within the confidence intervals of the high and minimally acceptable consensus levels. The sequences shown in table one were presented using customer feedback tables similar to those found in online purchasing environments.
Table 1. Consensus Sequence Manipulations: % of Satisfied Customers During Period
Internet purchases were selected as the setting for the studies because: (1) the Internet has become a large catalyst of both B2C and C2C Internet transactions, (2) respondents were readily available who had experience with Internet venues, (3) Internet sites are used by consumers to purchase and research an enormous variety of products and service, (4) Internet intermediaries actively encourage customer feedback and integrate disclosure of company reputation into the online buying process, and (5) reputation is presented at the forefront of many Internet transactions, often being viewed by shoppers prior to product or service descriptions.
Prior to administration, all questionnaires were randomized. Undergraduate marketing students at a large Southern U.S. university were given course credit in their classes for volunteering to participate in the studies. A relatively equal number of male and female students participated in both experiments. The age of the students ranged from 19 to 55 (median = 21) in experiment one and 19 to 48 (median = 21) in experiment two. No students that participated in the pilot study were included in these samples. During debriefing, no subjects indicated that they engaged in hypothesis guessing while responding to the questionnaire, although some mentioned that the items were related to their trust in the firm. No subjects revealed awareness that they were participating in an experiment or that their questionnaire differed from others'.
Measurement, Control, and Manipulation Check
Trust. Trust items from a two-dimensional brand trust scale developed by Delgado-Ballester (2004) were slightly modified to reference an Internet seller in lieu of a brand and to refer to the respondents' expectations of the seller rather than a first-hand experience. For example, "[X] is a brand that is honest and sincere," was changed to "This firm is honest and sincere," and "[X] is a brand that would make an effort to satisfy me," became "This seller would make an effort to satisfy me." The remaining items encompassed beliefs of reliability, credibility, and benevolence. A seven-point Likert scale anchored by "strongly disagree" and "strongly agree" was applied.
Prior Experience. The discounting principle proposed by Kelley (1973) summarizes how causal attributions may be affected by prior beliefs. This principle states that the effects of an attribution are minimized when an alternative explanation for the outcome is plausible. Consumers with more Internet purchasing experience may be more aware of alternative explanations for negative outcomes, and thus, take a more cognitive approach to evaluating the information. Additionally, this group of consumers is inherently more likely to have left feedback for Internet sellers in the past than consumers that only use the Internet for communication and information search purposes. Moreover, those consumers that shop online have been shown to differ significantly on measures of risk perception from those consumers that do not shop online (Donthu and Garcia 1999). McCole and Palmer (2002) investigated trust in relation to frequency of conducting Internet transactions. Respondents conducting fewer than five transactions were found to have low trust in the Internet as a purchasing mechanism. Persons reported that they "could not count on the Internet to be trustworthy" and that they "did not believe that the Internet fulfilled their purchase requests competently." These findings prompted us to test for prior experience effects. Thus, we included a frequency of Internet purchase factor as a covariate in each of the two experiments.
Sequence Manipulations. The sequence manipulations were checked with a single item at the end of the questionnaire, reading, "The reputation of this firm has changed a lot in the past three months." Results from this check, shown in table one, reveal that there was no statistical difference in reputation change over the three periods within either experiment (all p>.40). As intended, the perceived change between periods in experiment two was substantially lower than in the first set of manipulations.
Analysis and Results
Experiment One
A 2 (sequence: improving, diminishing) x 2 (consensus: low, high) between-subjects design was used to test the first two hypotheses. Thus, subjects were grouped into a low-diminishing condition (n=31), a low-improving condition (n=45), a high-diminishing condition (n=33), and a high-improving consensus condition (n=39).
Before analyzing the differences between subject groups, confirmatory factor analysis was performed to evaluate the psychometric properties of the dependent measure. All items were constrained to load on two dimensions of trust as specified by Delgado-Ballester (2004). The reliabilities of each dimension were above .89. Convergent validity was achieved as item to dimension loadings all fell above .75 and the average variances extracted (.789 and .866) exceeded Fornell and Larcker's (1981) criteria. Dimension to trust factor loadings were .879 and .958. Additionally, several fit indices confirmed that the measurement model fit very well to the data (CFI=.99, TLI=.99, RMSEA=.07). Homogeneity of error variance was assessed using Levene's test (F(3,144) =1.125, p = .341). Visual inspection of histograms yielded no considerable departures from normality and no outlying cases were detected. Studentized residuals plotted against both valence and sequence direction indicated linear independence of error terms. To test for covariate effects, a median split was used to group subjects by the number of Internet purchases made each year. Subjects were considered to be frequent Internet buyers if they purchased over four products per year online. This median split matches the aforementioned purchase frequency groups defined by McCole and Palmer (2002). ANCOVA results illustrated that trust responses did not vary significantly according to the frequency of Internet purchases (F(20,13.9) = 1.198, p = .371). After removing the covariate from the final model, univariate analysis of variance (ANOVA) was used to examine mean differences in trust across groups.
ANOVA tests indicated significant effects of
valenced consensus (F(1,144) = 36.197, p<.001), and sequence
direction (F(1,144) = 15.045, p<.001), and no interaction
effects (F(3,144) = 0.816, p=.368). Group means revealed that
differences in trust were in the hypothesized direction, supporting
H1 and H2. Specifically, the mean for the high consensus group
(µ = 5.210, S.E. = 1.49) was larger than the lower consensus
group (µ = 3.952, S.E. = 1.47) and the mean for the improving
sequence group (µ = 4.987, S.E. = 1.38) was larger than
the mean of the diminishing sequence group (µ = 4.175,
S.E. = 1.57). The estimated effect size (partial
) of valence on consumer trust was .201 and for sequence direction
was .095 (adjusted R2 = .254). The results are illustrated in
figure one and summarized in table two at the end of this section.
Experiment Two
The second experiment was designed to confirm the effects of valenced outcomes, while examining the change in sequence direction effects when the variation between the initial and most recent periods is attenuated. Thus, low diminishing (n= 41), low-improving (n= 48), high-diminishing (n= 46), and high-improving (n= 47) sequences similar to those used in the first experiment were applied to this investigation. However, the magnitudes of consensus variations in the manipulations were modified from 90-99% and 65-79%, to 95-99% and 64-70%, respectively. These sequences were selected to probe the boundaries of the effects found in experiment one.
Consumers have thresholds for which they perceive a difference between any two stimuli. They might react similarly to a 97% satisfaction rating as they would to a 95% rating, but perceive a greater difference in consensus as the distance between these numbers increases, such as in experiment 1, wherein consensus varied from 99% to 90%. Thus, we expect that when the sequence does not vary outside the thresholds identified in the pilot study, that the valence effects will still hold, but directional effects will be reduced or become insignificant because there is little perceived difference in reputation between the periods. Analysis of these reduced magnitude sequences allows for a more comprehensive understanding of when changes in consensus information should be considered critical for the firm
The same preliminary analyses employed in experiment one were repeated for the second experiment. Levene's test demonstrated that error variances could be considered equal across groups (F(3,178) =2.298, p = .071). Plots of residuals indicated no substantial departures from normality and studentized residual plots did not indicate linearly dependent error terms. Covariate effects were tested using a median split to group subjects by the number of Internet purchases made each year. Subjects were similarly considered to be frequent Internet buyers if they purchased over four products per year online. Trust responses do not vary significantly according to frequency of Internet purchases (F(14,11.8) = 1.205, p = .378).
ANOVA results indicated a significant effect
of consensus valence on consumer trust (F(1,178) = 83.056, p
< .001). However, neither the overall effect of the sequence
direction (F(1,178) = 0.098, p = .754), nor the interaction
effect (F(1,178) = .541, p = .463) was found to be significant.
These results support H3. Subjects were shown to exude higher
trust in transactions wherein consensus was more positively
valenced (µ = 5.33, S.E. = .114) than when the firms had
lower valenced sequences (µ = 3.84, S.E. = .117). However,
trust did not vary with the direction of the sequences when
the overall change over the three periods was abbreviated. The
estimated effect size (partial )
of valence on trust was .318 (adjusted R2 = .307).
Table 2. ANOVA Results
Discussion
One popular topic in word of mouth literature is examination of its impact on attitudes, perceptions, and behavioral intentions. The primary focus of our study was to assess the relative impact of consensus information and fluctuations of firm reputation on a critical antecedent of purchase behavior. Creating consumer trust can require very precise strategies, especially if the firm has developed a poor reputation in the marketplace. These strategies should attempt to maximize both consumer trust and purchase intentions. In order to properly evaluate the effectiveness of strategy implementations, researchers and practitioners must have an accurate impression of expected outcomes prior to allocation of resources.
Managerial Implications
Results in the two experiments illustrate a profound distinction between firms with excellent and poor reputations and emphasize the impact of dramatic versus meager reputation improvements. From a resource allocation perspective, the results demonstrate the importance of a firm's total commitment to implementing planned improvements. Firms that do not fully commit resources to improving their ability to satisfy consumer needs may only slightly improve their reputations. Unless substantial improvements are readily noticeable by consumers, allocation of funds and efforts to advance a firm's reputation are likely to be frivolous. Similarly, if a firm intends on changing slowly over time, it should not expect consumers to recognize improvement efforts in the short-term. Trust in the firm will not increase until the firm's reputation has crossed a certain threshold and has remained above that threshold for several periods. The confidence intervals of the consensus categories identified in the pilot study give some indication of what these thresholds might be.
Our findings also highlight the importance of maintaining a first-rate reputation. Sellers should not be unsettled if their reputation is diminishing so long as consensus does not decline to a lower category (e.g. from ideal to average, from average to less than acceptable). Providing the firm's reputation has not declined greatly in recent periods, swift recovery is possible. In this case, the firm should focus on problem identification and prevention, perhaps through encouragement of complaining behavior, rather than making substantial alterations to service delivery processes. Organizations should also be mindful that customers now have a very convenient way to inform a large number of consumers about service failures. Many more periods of positive evaluations are required to rebuild trust than are required to damage it (Ross, Lepper, and Hubbard 1975).
Because firms can influence customer expectations with marketing communications, and they have control over their own performance, it follows that Internet firms are also capable of at least influencing consumer trust. Firms must ensure accurate communication of product descriptions, establish accurate service expectations, create opportunities for customer-employee interaction, address customer concerns promptly, and ultimately generate the word-of-mouth necessary to maintain or improve its reputation. Companies selling on the Internet must make efforts to disclose as much unambiguous information about a product as possible. The presentation of a larger amount of up-front information should give consumers a more accurate representation of the product and the firm's policies. Many Internet companies participate in guarantee and insurance programs similar to bond insurance used by many service firms. These activities have also been shown to be drivers of trust (Lansing and Hubbard 2002). However, participation in such programs may not be necessary to demonstrate trustworthiness. Firms can reduce costs associated with subscribing to these services by building consensus of high performance on attributes such as ability to satisfy customers and deliver outstanding service.
Research Directions
Prospect theory notes that losses are more heavily weighted than gains of equal magnitude (Kahneman and Tversky 1979). The literature suggests that there will be more of a change in consumer attitudes when the direction of the change is negative than when the change of the direction is positive (Ross and Simonson 1991). Results from the pilot study suggest that disproportionate weighting effects would hold in online shopping environments. Specifically, the presence of disproportional differences between each of the consensus levels implies that improvements from low consensus must be larger than improvements in high consensus in order to identically influence changes in consumer beliefs. Future research might attempt to confirm these effects. New sequences could be created wherein the initial periods' consensus is the same and the magnitude between periods is identical, but varying in opposite directions. Additionally, these effects might vary according to the valence of the consensus in the initial period.
The present investigations explored effects of consensus on consumer trust, but made no attempt to estimate the variance in customer purchase intentions or other consumer attitudes, such as willingness to pay a price premium. The study also omitted exploration of individual difference traits such as normative and informational influence and predispositions to trust and seek risk. Examination of consensus effects moderated by typology of Internet consumer (Kau, Tang, and Ghose 2003) or history of successful Internet based transactions would also be of interest. This would allow researchers to link known characteristics of these groups with specific trust building strategies.
From the perspective of information search, individual differences and purchase risk may affect the likelihood of consumers to consider qualitative comments that often accompany consensus information. Many of these qualitative comments identify the type of performance failure and may be important in examinations of consumers' attitude formations. Companies having multiple complaints citing analogous failures may be viewed differently than firms receiving only one such complaint. The nature of the failures (e.g. shipping, product quality, responsiveness) may also have differential effects on consumers' attitudes.
Concluding Comments
In this paper we focused largely on consumer expectations and the perceived ability of a firm to meet those expectations. Shapiro (1982) suggests that expectations are derived directly from the firm's reputation, which Ha (2004) demonstrates to be a factor that influences brand trust in online environments. Reputation becomes especially critical in buying situations wherein single sources of information are capable of convincing a consumer to purchase from a particular firm (Gremler, Gwinner, and Brown 2001).
The present findings show that the valence of a firm's reputation is a primary influencer of trust when reputation is made available with consensus information. Improvement or deterioration of a firm's reputation over time will also affect the level of trust that consumers have in the selling firm. However, the positive impact of reputation improvements will only occur if the change in reputation is substantial enough to create expectations that the firm will act accordingly in subsequent transactions. Lastly, Boulding et al. (1993) recommend that firms attempt to raise predictive expectations, stating that even if expectations are negatively disconfirmed there is still likely to be an overall positive effect on the consumer's future behavior.
References
Anderson, Erin W. and Barton Weitz (1989), "Determinants of Continuity in Conventional Industrial Channel Dyads," Marketing Science, 8 (Fall), 310-323.
--- and --- (1992), "The Use of Pledges to Build and Sustain Commitment in Distribution Channels," Journal of Marketing Research, 24 (February), 18-34.
Anderson, Norman H. (1981), Foundations of Information Integration Theory, New York: Academic Press, 144-154.
Anderson, Richard, Raymond Kulhavy, and Thomas Andre (1971), "Feedback Procedures in Programmed Instructions," Journal of Educational Psychology, 62, 148-156.
Barrett, Robert T. and Robert E. Pugh (2003), "Procurement Auctions in E-Commerce," Southern Business Review, 29 (Fall), 1-14.
Boulding, William, Ajay Kalra, Richard Staelin, and Valarie A. Zeithaml (1993), "A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions," Journal of Marketing Research, 30 (February), 7-27.
Brannigan, Colm and Peter de Jager (2003), "Building e-Trust," Computerworld, 37 (8), 40-41.
Brown, Tom J. and Peter A. Dacin (1997), "The Company and the Product: Corporate Associations and Consumer Product Responses," Journal of Marketing, 61 (January), 68-84.
Chatham, Bob (2004), "Web Site Analytics Go Cross-Channel: Applying Clickstream Analytics To Multi-channel Scenarios," <http://www.forrester.com> (accessed on 10/11/2005).
Dasgupta, Partha (1988), "Trust as a Commodity," in Trust: Making and Breaking Cooperative Relations, Diego G. Gambetta, ed., Oxford: Blackwell, 49-72.
Delgado-Ballester, Elena (2004), "Applicability of a Brand Trust Scale Across Product Categories," European Journal of Marketing, 38 (May), 593-592.
Dick, Alan, Dipankar Chakravarti, and Gabriel Biehal (1990), "Memory-Based Inferences During Consumer Choice," Journal of Consumer Research, 17 (June), 82-93.
Doney, Patricia M. and Joseph P. Cannon (1997), "An Examination of the Nature of Trust in Buyer-Seller Relationships," Journal of Marketing, 61 (April), 35-52.
Donthu, Naveen and Adriana Garcia (1999), "The Internet Shopper," Journal of Advertising Research, 39 (May/June), 52-58.
Einhorn, Hillel J. and Robin M. Hogarth (1986), "Judging Probable Cause," Psychological Bulletin, 99 (January), 3-19.
Faloon, Kelly (2001), "B2B Adoption of Online Activities Expanding," Supply House Times, 44 (September), 30.
Farrell Vivienne, Ying Leung, and Graham Farrell (2000), "A Study on Consumer Fears and Trust in Internet Based Electronic Commerce," in Proceedings of the 13th Bled Electronic Commerce Conference, Slovenia, 647-658.
Forlani, David and Orville C. Walker, Jr. (2003), "Valenced Attributions and Risk in New-Product Decisions: How Why Indicates What's Next," Psychology and Marketing, 20 (May), 395-432.
Fornell, Claes and David F. Larcker (1981), "Evaluating Structural Equation Models with Unobservable Variables and Measurement Error," Journal of Marketing Research, 18 (February), 39-50.
Ganesan, Shankar (1994), "Determinants of Long-Term Orientation in Buyer-Seller Relationships," Journal of Marketing, 58 (April), 1-19.
Gefen, David (2000), "E-Commerce: The Role of Familiarity and Trust," International Journal of Management Science, 28 (December), 725-737.
--- and Detmar Straub (2003), "Managing User Trust in e-Services," E-Service Journal, 2 (Winter), 7-24.
Gescheider, George A. (1988), "Psychophysical Scaling," Annual Review of Psychology, 39, 169-200.
Geyskens, Inge, Jan-Benedict Steenkamp, Lisa K. Scheer, and Nirmalya Kumar (1996), "The Effects of Trust and Interdependence on Relationship Commitment: A Transatlantic Study," International Journal of Research in Marketing, 3 (October), 303-317.
Goldsmith, Ronald E. and Eileen Bridges (2000), "E-tailing vs. Retailing: Using Attitudes to Predict Online Buying Behavior," Quarterly Journal of Electronic Commerce, 1 (3), 245-253.
Gremler, Dwayne D., Kevin P. Gwinner, and Stephen W. Brown (2001), "Generating Positive Word-of-mouth Communication through Customer-Employee Relationships," International Journal of Service Industry Management, 12 (March), 44-59.
Ha, Hong-Youl (2004), "Factors Influencing Consumer Perceptions of Brand Trust Online," Journal of Product and Brand Management, 5 (13) 329-342.
Heider, Fritz (1958), The Psychology of Interpersonal Relations, New York: Wiley.
Hoffman, Donna, Thomas Novak, and Marcos Peralta (1998), "Building Consumer Trust in Online Environments: The Case for Information Privacy," <http://elab.vanderbilt.edu/research/papers/pdf/manuscripts> (accessed on 10/11/2005).
Jarvenpaa, Sirkka L., Noam Tractinsky, and Michael Vitale (2000), "Consumer Trust in an Internet Store," Information Technology and Management Journal, 1 (January), 45-71.
Kahneman, Daniel and Amos Tversky (1979), "Prospect Theory: An Analysis of Decision Under Risk," Econometrica, 47 (March), 263-291.
Kau, Ah Keng, Yingchan E. Tang, and Sanjoy Ghose (2003), "Typology of Online Shoppers," Journal of Consumer Research, 20 (2/3), 139-156.
Kelley, Harold H. (1967), "Attribution Theory in Social Psychology," in Nebraska Symposium on Motivation, David Levine, ed., Lincoln: University of Nebraska Press, 192-238.
--- (1973), "The Process of Causal Attribution," American Psychologist, 28 (February), 107-128.
Koehn, Daryl (2003), "The Nature and Conditions of Online Trust," Journal of Business Ethics, 43 (March), 3-19.
Lansing, Paul and John Hubbard (2002), "Online Auctions: The Need for Dispute Resolution," American Business Review, 20 (January), 108-116.
Loewenstein, George and Drazen Prelec (1993), "Preferences for Sequences of Outcomes," Psychological Review, 100 (January), 91-108.
Loewenstein, George and Nachum Sicherman (1991), "Do Workers Prefer Increasing Wage Profiles?" Journal of Labor Economics, 9 (January), 67-84.
Lucking-Reiley, David, Doug Bryan, and Daniel Reeves (2000), "Pennies From eBay: Determinants of Price in Online Auctions," Working Paper No. 00-W03, Vanderbilt University <http://www.vanderbilt.edu/Econ/wparchive> (accessed on 10/11/2005).
Lynch, John G., Howard Marmorstein, and Michael F. Weigold (1988), "Choices from Sets Including Remembered Brands: Use of Recalled Attributes and Prior Overall Evaluations," Journal of Consumer Research, 15 (September), 169-184.
Matsui, Tamao, Takashi Kakuyama, and Mary Lou Uy Onglatco (1987), "Effects of Goals and Feedback on Performance in Groups," Journal of Applied Psychology, 72 (August), 407-415.
McCole, Patrick (2002), "The Role of Trust for electronic Commerce in Services," International Journal of Hospitality Management, 14 (March), 81-87.
--- and Adrian Palmer (2002), "Transaction Frequency and Trust in Internet Buying Behaviour," Irish Marketing Review, 15 (2), 35-50.
Miller, Norman and Donald T. Campbell (1959), "Recency and Primacy in Persuasion as a Function of the Timing of Speeches and Measurements," Journal of Abnormal and Social Psychology, 59 (July), 1-9.
Morgan, Robert M. and Shelby D. Hunt (1994), "The Commitment-Trust Theory of Relationship Marketing," Journal of Marketing, 58 (July), 20-28.
Olson, Jerry C. and Phillip Dover (1979), "Disconfirmation of Consumer Expectations through Product Trial," Journal of Applied Psychology, 64 (April), 179-189.
Pastore, Michael (2000), "The Changing Face of E-Commerce," <http://www.clickz.com/stats/sectors/retailing/print.php/366201> (accessed on 10/11/2005).
Resnick, Paul and Richard Zeckhauser (2002), "Trust among Strangers in Internet Transactions: Empirical Analysis of e-Bay's Reputation System," in The Economics of the Internet and E-Commerce, M.R. Baye, ed., Amsterdam: Elsevier Science, 127-157.
Ross, Lee, Mark R. Lepper, and Mark Hubbard (1975), "Perseverance in Self Perception and Social Perception: Biased Attributional Processes in the Debriefing Paradigm," Journal of Personality and Social Psychology, 32 (November), 880-892.
Ross, William T., Jr., and Itamar Simonson (1991), "Evaluations of Pairs of Experiences: A Preference for Happy Endings," Journal of Behavioral Decision Making, 4 (4), 273-282.
Seiders, Kathleen and Leonard L. Berry (1998), "Service Fairness: What It Is and Why it Matters," Academy of Management Executive, 12 (May), 8-20.
Shapiro, Carl (1982), "Consumer Information, Product Quality, and Seller Reputation," The Bell Journal of Economics, 13 (Spring), 20-35.
Sheehan, Kim B. and Caitlin Doherty, (2001), "Re-weaving the Web: Integrating Print and Online Communications," Journal of Interactive Marketing, 15 (April), 47-59.
Simmons, Carolyn J. and John G. Lynch (1991), "Inference Effects Without Inference Making? Effects of Missing Information on Discounting and Use of Presented Information," Journal of Consumer Research, 17 (March), 477-491.
Urban, Glen L., Fareena Sultan, and William J. Qualls (2000), "Placing Trust at the Center of Your Internet Strategy," MIT Sloan Management Review, 42 (Fall), 39-48.
Varey, Carol and Daniel Kahneman (1990), "The Integration of Aversive Experiences Over Time: Normative Considerations and Lay Intuitions," Journal of Behavioral Decision Making, 5, 169-186.
Weiner, Bernard (1985), "An Attributional Theory of Achievement Motivation and Emotion," Psychological Review, 92 (October), 548-573.
About the Authors
Ray L. Benedicktus is a doctoral student in the Department of Marketing, College of Business, Florida State University. Prior to attending Florida State University, he worked as a small business consultant in North Carolina and earned his MBA at Fayetteville State University. His research interests include multi-channel marketing, branding, and consumer behavior. Email:rlb04d@cob.fsu.edu
Melinda L. Andrews is a doctoral student in the Department of Marketing, College of Business, Florida State University. Prior to attending Florida State University, she worked in the direct mail industry and earned her MBA at the University of Southern Mississippi. Her research interests include consumer behavior and branding. Email: mla04c@cob.fsu.edu
Acknowledgements
The authors would like to thank the anonymous JIAD reviewers for their helpful comments, and Dr. Ronald Goldsmith and Dr. Michael Brady for their feedback and encouragement during preparation of this paper.
Appendix: Pilot Scenario and Experimental Manipulation
Several weeks prior to the beginning of the semester, you visit a course website and see that your instructor has posted information on the textbook required for the course in which you have enrolled. Hoping to avoid long lines at the bookstore, you visit several Internet sites and find a listing for the textbook that you need. The description of the textbook notes it is used and in excellent condition. You also notice that the bookseller is an Internet company that has no retail locations.
To determine whether or not you will purchase the book, you decide to click on a link to the seller's profile that includes information on the satisfaction of other people that have purchased from the same company in the past. You are particularly concerned with the percentage of other buyers that have been satisfied with the seller. While waiting on the profile to appear on the page, you start to think about three things:
(1) What is the ideal satisfaction
percentage for an Internet company that has no retail locations?
(2) What is a desirable satisfaction percentage for an Internet
company having no retail locations?
(3) What would a minimally acceptable satisfaction percentage
be that the seller could have in order for you to still make
this purchase?
Below is a table that depicts these three levels of satisfaction along with a field for your opinion regarding the Internet company that does not have any retail locations. Please fill in the percentages that you consider to be an "Ideal" seller satisfaction level, a "Desirable" seller satisfaction level, and the "Minimum" level of satisfaction where you would still make the purchase. Also indicate what you think the typical satisfaction level would be for a company that only sells products online.
Similar to the previous table, there is a graph below that depicts four levels of satisfaction along with a field for your opinion regarding an Internet company that does not have any retail locations. Please fill in percentages that you consider to be an "Ideal" seller satisfaction level, a "Desirable" satisfaction level, "Typical" satisfaction level, and the "Minimum" satisfaction level where you would still make the purchase.
Experimental Manipulations: This section of the questionnaire asks you to consider a rating table similar to those found on Seller Profile pages at epinions.com, Amazon.com, Ebay.com, etc. Please consider the seller profile below when responding to the following statements and select the response that would best represent your attitude towards the seller if you were purchasing an item on the Internet.
