how task-facilitative interactive tools foster buyers’ trust in online retailers: a process view...

A

rripiptrrw©

KI

rSpbpa

ape

(

0d

Journal of Retailing 85 (2, 2009) 159–176

How Task-Facilitative Interactive Tools Foster Buyers’ Trust inOnline Retailers: A Process View of Trust Development in the

Electronic Marketplace�

Pranjal Gupta a,∗, Manjit S. Yadav b,1, Rajan Varadarajan b,2

a Department of Marketing, University of Tampa, Tampa, FL 33606-1490, United Statesb Department of Marketing, Mays Business School, Texas A&M University, College Station, TX 77843-4112, United States

bstract

While there is a sustained interest in research focusing on issues relating to trust development in the electronic marketplace, significant gapsemain in the literature. In particular, little is known of the underlying processes that may be occurring in online trust development. For example,esearch suggests that factors such as site design and navigability are among the factors that impact trust perceptions. Extant literature, however,s largely silent about why certain trust-related effects are observed in online environments. In this paper, we propose a new process-centricerspective for understanding the formation of online trust—through buyer’s assessment of the e-retailer’s assistive intent, implicitly embeddedn task-facilitative interactive tools. Specifically, we develop and test a model delineating the relationship between seller’s provision of interactiveroduct information management and product information comprehension tools, buyer’s perceptions of seller’s assistive intent, and buyer’s initialrust in the seller. The results of two studies provide support for the trust-enhancing effects of task-facilitative informational tools and the mediating
ole of buyer’s perceptions of seller’s assistive intent. Importantly, these effects occur without any explicit expressions of seller’s intentions. Theesults also suggest that the efficacy of interactive informational tools in engendering perceptions of seller’s assistive intent, and hence trust, variesith the buyer’s level of involvement with and knowledge of the product category.2009 New York University. Published by Elsevier Inc. All rights reserved.
ation;

cZcr2n

eywords: Online retailing; e-Retailing; Trust; Online trust; Initial trust formntentions and trust; Assistive intent; Helpful intentions

Introduction

A topic that has sparked considerable research interest, inecent years, is trust development in the electronic marketplace.patial separation, the physical separation between transactingarties on the Internet, and temporal separation, the elapsed time
etween when a transaction occurs and the buyer gains actualossession of purchased goods (Kollock 1999), can accentu-te online buyers’ perceptions of vulnerability and risk, and
� The authors are grateful to Raj Echambadi, Noreen Klein, Venky Shankar,nd Deepak Sirdeshmukh for helpful comments on previous versions of thisaper and to Bronius Motekaitis for programming support. They also acknowl-dge three anonymous reviewers for their suggestions to improve the manuscript.∗ Corresponding author. Tel.: +1 813 257 1788; fax: +1 813 258 7408.

E-mail addresses: [email protected] (P. Gupta), [email protected]. Yadav), [email protected] (R. Varadarajan).1 Tel.: +1 979 845 5884; fax: +1 979 862 2811.2 Tel.: +1 979 845 5809; fax: +1 979 862 2811.

i

rbp1aasnosu

022-4359/$ – see front matter © 2009 New York University. Published by Elsevier Ioi:10.1016/j.jretai.2009.02.001

Trust formation processes; Trust development; Trust development processes;

onsequently, their need for trust (Moorman, Deshpandé, andaltman 1993; Yadav and Varadarajan 2005). While advancesontinue to be made in this important and emerging area ofesearch (e.g., Bart et al. 2005; Schlosser, White, and Lloyd006; Urban, Sultan, and Qualls 2000; Yoon 2002), three sig-ificant gaps remain in our understanding of trust developmentn online buying environments.

First, many of the variables that have been the focus in extantesearch are trust antecedents that pertain to trust developmentetween new customers and existing on-line sellers. For exam-le, seller reputation (e.g., Jarvenpaa, Tractinsky, and Saarinen999; McKnight, Choudhury, and Kacmar 2002), consumer’swareness of the firm (Jarvenpaa, Tractinsky, and Vitale 2000),nd prior familiarity with the site (e.g., Yoon 2002) have beenhown to positively impact buyer’s perceptions of trustworthi-
ess of the seller. However, such factors generally have limitedr no applicability in the context of new or unfamiliar onlineellers about whom reliable priors are yet to be established. Thisnder-researched context is important from a substantive per-
nc. All rights reserved.

mailto:[email protected]



dx.doi.org/10.1016/j.jretai.2009.02.001

1 Retai

se

caNtPmos2auIctpbmtsfiBoic

pldat2pamtditra

tae“thstoctfb

otcma

rsftt(sbSitwthtttaotissttwl

psraf

aHSnaatpaL

60 P. Gupta et al. / Journal of

pective since entry by new firms occurs frequently in the rapidlyxpanding electronic marketplace.

Second, prior research has largely focused on the role ofharacteristics such as information privacy and security assur-nces (e.g., Dayal, Landesberg, and Zeisser 1999; Hoffman,ovak, and Peralta 1999; Pan and Zinkhan 2006; Yoon 2002),

hird-party seals of approval (e.g., Cheskin/Sapient Report 1999;almer, Bailey, and Faraj 2000) and streamlined order fulfill-ent capabilities (Bart et al. 2005) on trust development in

nline environments. However, as the online medium matures,uch competence-related assurances (see Wolfinbarger and Gilly003) have evolved to become commonplace and institution-lized online business practices, raising questions about theirsefulness for sustaining trust development over the long run.ndeed, the diminishing distinctive value of such cues on per-eptions of firm-level trust (Bart et al. 2005) is partly due tohe Internet maturing as a shopping medium and online shop-ers viewing such cues as baseline necessities for all web-basedusinesses. Further, such e-retailing competence cues may serveore to reduce buyer perceptions of e-retailer distrust rather than

o increase their perceptions of e-retailer trust (Cho 2006). Thus,uch cues may serve as necessary conditions, at best, but not suf-cient cues to develop buyer trust in an online seller. Further,art et al. also show that drivers of trust for various categoriesf websites and types of consumers may be different, suggest-ng a need for a better understanding of why such potentialontingencies may be triggered in online trust development.

Third, and especially important from a theory developmenterspective, very little is known about the key processes under-ying the efficacy of seller-controlled variables that may operateuring online trust development. For instance, although liter-ture demonstrates that certain website functionalities lead torusting beliefs (Bart et al. 2005; Schlosser, White, and Lloyd006), it is largely silent about the underlying trust buildingrocesses associated with these functionalities. What thoughtsre generated as buyers experience an online shopping environ-ent and attempt to update any priors they may have brought

o the online shopping task? What inferences, if any, do buyersraw about the seller based on task-related website functional-ties they encounter during the online shopping task? Researchhat sheds insights into these and other unaddressed process-elated questions are crucial for advancing knowledge in thisrea.

Against this backdrop, the overarching goal of this paper iso advance our understanding of how trust may form duringn initial online encounter between buyers and a new online-retailer, controlling for baseline competence-related factors.New” in this context implies that a retailer is either a new entryo the market or that it is an existing firm but the consumeras no prior familiarity or experience with it. Specifically, thistudy seeks to contribute to the emerging literature on onlinerust by developing and empirically testing a model centeredn a construct that we refer to as seller’s assistive intent. This
onstruct captures a buyer’s perceptions of the extent to whichhe online seller exhibits intent, implicitly embedded in task-acilitative tools aligned with the buyer’s interests, to help theuyer fulfill a specific task on the seller’s website. Our focus
MIta

ling 85 (2, 2009) 159–176

n interactive, task-facilitative informational tools stems fromhe fact that information, in its various forms, is central to theompletion of most buying related tasks and thus informationanagement and comprehension tools are critical for buyers at

ny stage of the buying decision process.We present two studies that examine the effects of task-

elated interactive informational tools embedded in an onlinehopping interface on a buyer’s initial trust in a new seller,ocusing on the hypothesized mediating role of a buyer’s percep-ions of a seller’s task-related assistive intent. We also investigatehe moderating effects of buyer’s product category involvementStudy 1) and product knowledge (Study 2) on the relation-hip between embedded, task-facilitative interactive tools anduyer’s perceptions of seller’s task-specific assistive intent.tudy 2, in addition to extending and clarifying the key find-

ngs of Study 1, uses protocol analysis to explore the underlyingrust development process in greater detail. In both studies,e find that merely embedding interactive informational tools

o solve a single task in an online shopping environment canave a significant effect on a buyer’s trust beliefs, and thathis relationship is mediated by buyer’s perceptions of seller’sask-related assistive intent. It is particularly noteworthy thathis effect occurs after controlling for competence cues gener-lly studied in the literature and with no explicit expressionsf the seller’s intentions. Furthermore, the absence of theseools appears to lead to negative trust-related inferences – whichs suggestive of the important role played by the new con-truct that we propose in this paper – buyer’s perceptions ofeller’s task-specific assistive intent. The results also suggesthat the efficacy of different sets of interactive informationalools in engendering perceptions of seller’s assistive intent variesith the buyer’s product involvement and product knowledge

evels.The remainder of the paper is organized as follows. We first

resent the proposed focal construct, hypotheses and theoreticalupport for the hypotheses. Next, we discuss the design andesults of the two experiments. In the Discussion section, weddress the limitations of the research and propose directionsor future research.

Theory and hypotheses

Trust is widely acknowledged as central to relationshipsnd long-term commitment between buyers and sellers (e.g.,arris and Goode 2004; Morgan and Hunt 1994; Sirdeshmukh,ingh, and Sabol 2002). Particularly in situations involving vul-erability (Moorman, Deshpandé, and Zaltman 1993), beliefsbout trustworthiness are likely to determine whether or nottransaction between a buyer and seller occurs. Most defini-

ions of trust revolve around the level of confidence that onearty has of the expected behavior of another and embodywillingness of one party to rely on the trusted party (e.g.,indskold 1978; Moorman, Zaltman, and Deshpandé 1992;
organ and Hunt 1994; Sirdeshmukh, Singh, and Sabol 2002).
n terms of the dimensions of trust, the literature generally dis-inguishes between credibility and benevolence (e.g., Doneynd Cannon 1997; Ganesan 1994; McAllister 1995). Credibil-

Retai

idtta1

Ba

P. Gupta et al. / Journal of

ty perceptions revolve around one’s beliefs about the honesty,ependability and integrity of the other party. Benevolence refers
o one’s perception that the other party is motivated to pro-ect the best interests of the focal party when new conditionsrise for which no prior commitments were made (Ganesan994).
ta

Fig. 1. (Panels A and B) Illustrative screenshots of interactive information ma

ling 85 (2, 2009) 159–176 161

uilding online trust: buyer’s perceptions of seller’sssistive intent

Central to our research is the premise that buyers formrust perceptions through assessments of the seller’s task-relatedssistive intent—a buyer’s perception that the seller is help-

nagement (IIM) and interactive information comprehension (IIC) tools.

162 P. Gupta et al. / Journal of Retailing 85 (2, 2009) 159–176

Cont

itsia

l

Fig. 1. (

ng or attempting to help the buyer complete a given task,
aking into account the buyer’s task-related needs. This con-truct captures a buyer’s perception that the seller’s intentions,mplicitly conveyed by the task-related tools provided, areligned with the needs of the buyer for the given task. It fol-
ists

inued ).

ows that a buyer’s perceptions of the seller’s assistive intent,
mplicitly embedded in online task-related tools that a userees as salient, will differ depending on the task and con-ext. For ease of exposition, we restrict our discussion ofuch online task-facilitative tools to those utilized to make

Retai

cu

tWoVlskaifbc

1

2

asitpt

M

ofCattt

tinawa

aS1ei

mmctmoewcbtopee

tt(lrtbpsiiasbtcthald

Hw


hoice decisions for fairly complex, information intensive prod-cts.

In the online purchasing context, task-facilitative tools oftenake the form of interactive tools embedded in the user interface.

e reviewed specific functionalities mentioned in the literaturen interface design (e.g., Häubl and Trifts 2000; Montoya-Weiss,oss, and Grewal 2003) and those featured on the websites of

eading online sellers. As expected, we found a wide range ofuch functionalities (e.g., tools for searching, acquiring productnowledge, simplifying the choice options, buying guides, etc.)nd evidence suggestive of their significance for firms compet-ng online (e.g., Montoya-Weiss, Voss, and Grewal 2003). Theunctionalities of these task-facilitative tools, particularly in theuying context, suggest that they can be grouped into two broadategories of tools.3

. Interactive information management (IIM) tools: Toolswhich enable buyers to sort through and/or compare availableproduct alternatives. For example, these tools allow buyersto limit and sort choices on levels of various attributes and/orengage in side-by-side comparisons of products in dynami-cally created tables. Fig. 1(Panel A) provides an overview ofselected functionalities provided to respondents in our studies(details are discussed in a later section).

. Interactive information comprehension (IIC) tools: Toolswhich enable buyers to understand the meaning and benefitsof product-related information. These tools include function-alities such as buying guides and glossaries of product-relatedterms. Fig. 1(Panel B) provides an overview of selected func-tionalities provided to respondents in our studies (details arediscussed in a later section).

Thus, while IIM tools facilitate the management of productnd attribute-level information, IIC tools facilitate the under-tanding and interpretation of that product and attribute-levelnformation. In a purchase context, provision of IIM and IICools by an online seller are expected to positively impact buyer’serceptions of seller’s assistive intent, and in turn, impact buyer’srust in the seller.

ediating role of seller’s assistive intentConceptual support for the relationship between perceptions

f good intentions and enhanced trustworthiness beliefs stemsrom extant literature on trust (e.g., Deutsch 1973; Doney andannon 1997; Lindskold 1978). This literature suggests that
buyer’s perception of the seller’s assistive intent in the con-
ext of the immediate transaction is likely to be diagnostic ofhe seller’s future trustworthiness. Particularly in the context ofrust in a new online seller, where other trust indicators such

3 From a theory development perspective, we investigate the role played byhese two broad categories of online tools as reflective of task-facilitative toolsn the context of buying a fairly complex product. While specific functionalitieseeded may change depending on the task, our underlying theorizing regardingssistive intent perceptions and trust formation will still be applicable. Also,hen describing the interactive tools, the terms task-facilitative and task-related

re used interchangeably.

ao

Hw(

Hiast

ling 85 (2, 2009) 159–176 163

s a history of directly observed behaviors (e.g., Sirdeshmukh,ingh, and Sabol 2002) or a good reputation (Doney and Cannon997; Jarvenpaa, Tractinsky, and Vitale 2000) do not exist, buy-rs may seek to infer trustworthiness from perceived seller’sntent derived from their experiences on the website.

Attribution research suggests that humans are inherentlyotivated to understand the causal structure of their environ-ent (Kelley 1967) and tend to ascribe stable dispositional

auses to actions and effects, as opposed to ascribing situa-ional causes (Heider 1958; Jones and Davis 1965). Thus, buyers

ay ascribe trustworthiness, which is a stable dispositional traitf the seller, as a trait underlying the seller’s assistive intent,xpressed through task-related functionalities embedded in theebsite. Further, individuals tend to take attributional “short-

uts” by ascribing causal linkages to events they experienceased on their preconceived causal schemas (e.g., good inten-ions observed at this point suggest future trustworthiness), aspposed to making more elaborate attributional analysis over aeriod of time (Ferrin and Dirks 2003). Thus, we posit that buy-rs are likely to link perceived seller intent to trustworthinessven in short, initial encounters with a new online seller.

In regard to the dimensions of trust, prior research suggestshat benevolence generally develops over a longer period ofime and is generally preceded by perceptions of credibilityMcAllister 1995). However, since an assessment of benevo-ence is meaningful and important to establish in a potentiallyisky context, buyers may be motivated to use available cueso make benevolence assessments as well. Therefore, we expectuyer’s perceptions of seller’s assistive intent to lead to enhancederceptions of both seller credibility and benevolence. Also,ince trust is context specific (Johnson-George and Swap 1982),t is important to frame the concept of “good” seller intentn the context of a specific task. In the context of choosing

risky product at a new seller’s website, for instance, theeller’s intentions to assist buyers in their choice task coulde meaningfully represented by information management toolshat facilitate that choice process. Thus, in the context of ahoice task, task-facilitative interactive tools that allow buyerso manage the information available (IIM tools) and those thatelp buyers comprehend the information available (IIC tools)re likely to be seen as reflecting seller’s assistive intent. Col-ectively, the following hypotheses stem from the precedingiscussion:

1. Buyer’s perceptions of a new online seller’s assistive intentill mediate the effects of seller-provided task-facilitative inter-

ctive informational tools (IIM and IIC) on buyer’s perceptionsf the seller’s (a) credibility and (b) benevolence.

2. Buyer’s perceptions of a new online seller’s assistive intentill be positively related to buyer’s perceptions of the seller’s

a) credibility and (b) benevolence.

3. (a) Interactive information management (IIM) and (b)
nteractive information comprehension (IIC) tools embedded innew online seller’s interface, without any overt expressions of
eller’s intentions, will positively impact buyer’s perceptions ofhe seller’s assistive intent.

1 Retai

M

fiwtterwoS

tlSa1Tdtti

stiatebsf(eetghtemaa

Htaeh

mcak

Hiael

mbuafmippHt

apbhatn1lpevopeam

teRiovvt


oderating role of buyer’s product involvementIt is important to note that the relationships between task-

acilitative tools and buyer’s perceptions of seller’s assistiventent are predicated on the assumption that the buyer has theillingness and ability to utilize (or appreciate the utility of) the

ask-facilitative tools provided. Product involvement, defined ashe personal relevance of the product category based on inher-nt needs, values and interests (see Zaichkowsky 1994), largelyeflects the requisite willingness and ability.4 Generally, buyersith higher levels of involvement will also have higher levelsf knowledge in a product category (Mitchell and Dacin 1996;ujan 1985).

Literature on information search and processing suggestshat buyers with a higher level of product involvement areikely to process more information (Punj and Staelin 1983;rinivasan and Ratchford 1991), and at a deeper level (e.g.,ttribute-level vs. holistic comparisons; Alba and Hutchinson987; Maheswaran and Sternthal 1990; Moorthy, Ratchford, andalukdar 1997; Sujan 1985), given their ability and interest inoing so. On the other hand, less involved buyers are less likelyo process attribute-level information and focus more on holis-ic evaluations of products and product attribute comprehensionnformation (Maheswaran and Sternthal 1990; Sujan 1985).

These findings suggest that the salience (or usefulness) of aeller’s action of providing interactive information managementools that enable buyers to interactively work with attribute-levelnformation, and the resultant effect on perceptions of seller’sssistive intent are likely to be greater for more involved buyershan for less involved buyers. In contrast to IIM tools (whichmphasize attribute-level data and are likely to be more valuedy more involved buyers), interactive information comprehen-ion tools which emphasize attribute-level comprehension in theorm of buying guides and glossaries are likely to be more salientor useful) to less involved buyers. Although less involved buy-rs may prefer simplifying cues or heuristics to avoid cognitiveffort (e.g., brand names, recommendations, etc.), they are likelyo utilize IICs if other simplifying cues are unavailable. Thus,iven their desire to minimize effort and seek help to compre-end attribute information, less involved consumers are likelyo seek and utilize IICs to fulfill their choice task. That is, buy-rs with less product involvement are likely to see IIC tools asore facilitating and useful, making their perceptions of seller’s

ssistive intent and corresponding trust perceptions stronger asresult of the use of IIC tools. Hence:

4. The positive relationship between provision of interac-ive information management (IIM) tools by a new online sellernd buyer’s perceptions of seller’s assistive intent, without overt
xpressions of seller intentions, will be greater for buyers withigher product involvement.
4 Other individual-difference variables (e.g., need for cognition) could deter-ine the desire to process information. However, in this initial exposition, our

onceptualization and empirical work focuses on individual-difference variablest the level of a product category—product involvement (Study 1) and productnowledge (Study 2).

tevsswstf

ling 85 (2, 2009) 159–176

5. The positive relationship between provision of interactivenformation comprehension (IIC) tools by a new online sellernd buyer’s perceptions of seller’s assistive intent, without overtxpressions of seller intentions, will be greater for buyers withower product involvement.

We conducted two studies to test the mediation (H1–H3) andoderation (H4 and H5) hypotheses presented above. While

oth studies employ the same experimental design and manip-late the same set of task-facilitative tools, we planned thems complementary investigations. Study 1 (N = 246) presentsormal tests of the mediation hypotheses (H1–H3) and theoderation hypotheses (H4 and H5) based on buyer’s endur-

ng involvement as the moderating variable. Study 2 (N = 223)rovides a more detailed look at the underlying process usingrotocol analysis, in addition to formal meditational tests for1–H3. This follow-up study investigates H4 and H5 based on

he moderating influence of product knowledge.

Study 1

A 2 (interactive information management tools: present vs.bsent) by 2 (interactive information comprehension tools:resent vs. absent) by 2 (product involvement: high vs. low)etween-subjects experimental design was utilized to test theypothesized relationships. Enduring product involvement ismeasured, rather than a manipulated factor. Two considera-

ions motivated this design decision. First, given the fairly stableature of product involvement of individuals (Zaichkowsky994), and the possibility of creating demand artifacts if manipu-ated, measuring this variable was preferred. Second, measuringroduct involvement as an individual characteristic enhances thecological validity of the findings regarding the effects of thisariable (e.g., see Sujan 1985). However, one potential limitationf this approach (stemming from lack of randomization) is theossibility that unobserved variables may mask the observedffects of product involvement. Based on a careful consider-tion of these trade-offs, we opted for measuring rather thananipulating respondent’s product involvement.Since the hypothesized relationships focus on buyer’s percep-

ions of trust in a new online seller resulting from a single, initialncounter, a cross-sectional design was considered appropriate.espondents (graduate and undergraduate students) participated

n a shopping task (choice of a laptop computer) in a simulatednline shopping environment that allowed us to systematicallyary the two manipulated independent variables (i.e., provisions. nonprovision of IIM tools and IIC tools). Pretests indicatedhat the focal product in the shopping task was (a) relevant forhe respondents in the sampling frame, (b) involved consid-rable purchase risk, and (c) was characterized by substantialariation in levels of product involvement. Based on these con-iderations, and guidance from the literature related to the use oftudent subjects in experimental research (e.g., Peterson 2001),
e concluded that the simulated online shopping task provided a
uitable context for testing the hypothesized relationships. Addi-ional details about the study are provided in the sections thatollow.

Retai

E

irtodcTnssswtc(ibts

etpriir1prwe

I

dssuipe

tCbHfspTti

ocetotai

I

mtitSiHtvi

roTrsRta“toFtIoet

I


xperimental procedure

The experiment was conducted under controlled conditionsn a dedicated computer laboratory. Upon arrival, respondentseceived a copy of the instructions and a brief verbal introduc-ion to the task and were asked to begin the online shopping tasknly after they had completely read the instructions. Respon-ents were informed that they would be shopping for a laptopomputer at the website of a new e-retailer (AtoZTronics.com).hey were asked to select a laptop that best matched theireeds. Appropriate links on the website highlighted the onlinetore’s adherence to industry best practices (e.g., informationecurity, price guarantees, shipping options, toll-free customerervice, technical support, and return policies). These linksere provided for all four experimental (manipulated) condi-

ions and created an equal baseline of e-retailing competenceues5 commonly encountered in the electronic marketplaceand examined in prior research). Introductory instructions alsonformed respondents that the e-retailer adhered to industryest practices and indicated that all product-related informa-ion available at the e-retailer’s site was obtained from reliableources such as manufacturers and Consumer Reports.

Respondents were randomly assigned to one of the fourxperimental conditions in which they completed a shoppingask (selection of a laptop computer). Prior to the actual shop-ing task, respondents answered a few introductory questionselated to computers and computer usage. In addition to serv-ng as warm-up questions to familiarize the respondents, thentroductory section also assessed the enduring involvement ofespondents with the computer product category (Zaichkowsky994). The experimental procedure ended with the onlineost-task questionnaire. We obtained usable data from 246espondents for purposes of hypotheses testing. The respondentsere 58 percent female and 42 percent male. On average, the

ntire experimental procedure was completed in about 25 min.

nteractive tools in the website design

The hypothetical online store used in the choice task wasesigned and developed by an information technology profes-ional to closely match the look and feel of real online electronicstores. The design of the online store was varied to manip-late the two independent variables of interest—interactive
nformation management tools and interactive information com-rehension tools (see Fig. 1 for an overview of each of the fourxperimental conditions).
5 Extant work shows that competence invokes higher perceptions of trustwor-hiness (e.g., Doney and Cannon 1997; Sirdeshmukh, Singh, and Sabol 2002).ompetence-related cues in the online setting such as order fulfillment capa-ilities (Bart et al. 2005) and consumer privacy and information security (e.g.,offman, Novak, and Peralta 1999; Pan and Zinkhan 2006) are all reflections of

unctional competence and impact trustworthiness. In general, they have beenhown to have diminished in value (Bart et al. 2005). Although necessary, com-etence related cues may not be sufficient to elicit trust in a new online seller.hus, we control for basic competence cues in the online buying context and

hen study the incremental impact of assistive intent embedded in task-relatednformational tools on buyer’s perceived seller trustworthiness.

vtfcrcRfittsa

a

ling 85 (2, 2009) 159–176 165

As noted earlier, our interest is in exploring the influencef these two broad categories of online tools, rather than spe-ific functionalities incorporated in these tools (which, as can bexpected, keep evolving with technology). The design of thesewo broad categories of online tools (which encompass a rangef functionalities) was guided by field research and a review ofhe pertinent literature. The product information content (i.e., thevailable product alternatives and attributes thereof) remaineddentical across all four conditions.

nteractive information management (IIM) toolsInteractivity is conceptualized as a characteristic of a com-

unication system that allows buyers to seek and gain accesso information on an on-demand basis where the content, tim-ng and sequence of the communication is under the control ofhe end user (Ariely 2000; Fortin 1997; Lynch and Ariely 2000;teuer 1992). Interactive tools provided by incumbent e-retailers

n the marketplace, and those reported in the literature (e.g.,äubl and Trifts 2000; Lynch and Ariely 2000) were reviewed

o develop an interactive tool set that would allow respondentsarious options to query and manage product-related attributenformation.

One set of tools provided was a filter system that allowedespondents to limit the consideration set on the basis of vari-us attributes and attribute levels desired (see Fig. 1(Panel A)).he second tool set was a comparison chart feature that allowed

espondents to compare laptops in their consideration set bypecifying the laptops that would enter a comparison matrix.espondents were able to sort on several attributes within the

able, either in ascending or descending order, and could views many or as few laptops and attributes as they wanted. Thepresent” conditions had the tools whereas the “absent” condi-ions did not. In the “absent” condition, respondents had to clickn each product separately to view all the attributes individually.ig. 1(Panel A) provides additional details about how these IIM

ools were incorporated in the interface of the online storefront.t is important to note that, regardless of the presence or absencef IIM tools, identical product information was provided in allxperimental conditions (i.e., the number of laptops available,heir attributes and the levels of those attributes).

nteractive information comprehension (IIC) toolsThe second manipulated factor, IIC tools, refers to tools pro-

ided by an e-retailer to enable potential customers to understandhe product, its attributes, and the benefits associated with dif-erent attribute levels. To guide the development of IIC tools foromputers, extensive searches were conducted of incumbent e-etailers of electronic goods, websites dealing specifically withomputers and computer-related publications (e.g., Consumereport’s Home Computer Buying Guides, Computers Simpli-ed and Computers for Dummies). Guided by this field research,wo IIC tool sets were developed: (1) Buying Guides that explainhe capabilities of various attribute configurations, and (2) Glos-
aries that provided an explanation and benefits of each attributend its levels.
The Buying Guides were developed as brief explanationsimed at helping customers match their needs with a set of lap-

http://atoztronics.com/


Table 1Measures of buyer’s perceptions of seller’s task-related assistive intent, credibility and benevolence.

Constructa Standardized loadings t-Valuesb Composite reliabilityc Average variance extractedd

Study 1 Study 2 Study 1 Study 2 Study 1 Study 2 Study 1 Study 2

Assistive intentThis e-retailer really wants to help

me choose the right laptop.91 .93 18.40 18.32 .96 .93 .88 .82

The intention of this e-retailer is toassist me as much as possible

.96 .91 20.41 17.48

This e-retailer is doing what it canto help me make a good laptopchoice

.95 .88 19.81 16.72

CredibilityThis e-retailer is likely to be a

trustworthy one.86 .85 16.63 15.58 .91 .92 .68 .70

I think this e-retailer will bedependable

.84 .84 15.87 15.20

This e-retailer is likely to keep thepromises it makes to me

.86 .85 16.42 15.47

This e-retailer is likely to be openin its dealings with me

.79 .83 14.59 15.04

This e-retailer seems sincere .77 .81 14.07 14.34

BenevolenceThis e-retailer is likely to always

put its customers first.84 .86 16.07 15.91 .90 .91 .69 .72

It feels like this e-retailer will beon my side, no matter whatproblems arise

.84 .87 16.14 16.22

This e-retailer is likely to alwayskeep my best interests in mind

.93 .93 18.96 18.22

This e-retailer will be like a truefriend

.70 .71 12.29 11.92

Note: N = 246 for Study 1 and N = 223 for Study 2.a Measurement scales for the mediating and trust constructs are anchored at 1 = strongly agree to 7 = strongly disagree, with lower scores representing more

favorable perceptions.

tBatgctwlitl

D

wlt(aw

r7tc

M

Bt

(ot((ss

b For all loadings p < .05.c Bagozzi and Yi (1988).d Fornell and Larcker (1981).

ops and a price range they were willing to pay (see Fig. 1(Panel)). Each glossary, incorporated as a pop-up window whenn attribute was clicked anywhere in the website, describedhe attribute and the associated benefits. These guides andlossaries were revised and finalized based on feedback fromontent experts. In the IIC tools “present” conditions, all IICools were incorporated and in the “absent” conditions theyere not. Fig. 1(Panel B) illustrates these tools. Again, regard-

ess of the presence or absence of IIC tools, identical productnformation was provided in all experimental conditions (i.e.,he number of laptops available, their attributes and attributeevels).

evelopment of stimuli

We consulted content experts to populate the online storeith alternative choice options (i.e., different configurations of

aptops, number and levels of attributes and number of alterna-
ives). Our design goals were to develop a number of choices that1) were described adequately on a set of appropriate attributesnd (2) reflected variety similar to what a computer shopperould find at an actual online computer retailer. Based on the
ttlo

ecommendations of content experts, the final dataset included0 laptops described on 25 different attributes. This set of lap-op choices was used as stimuli in each of the experimentalonditions.

easures

uyer’s perceptions of seller’s assistive intent and initialrust in seller

We developed the final set of items for these constructssee Table 1) based on a review of the literature and a seriesf pretests. Regarding the measurement of trust, extant litera-ure makes a distinction between credibility and benevolencee.g., Ganesan 1994; McAllister 1995). Following Churchill1979), the construct domain was specified first. A detailedearch of the trust literature was conducted to determine con-truct domains and generate an initial pool of items to measure
he trust dimensions (credibility and benevolence) and assis-ive intent. Next, to ensure face validity of the items, an initialist of items was sent to five leading experts in the areaf trust for assessment. They were provided with the con-

Retai

totcfifsti

t1h(oFiAi1osteaebaht.nar

B

u(Tmaechitcg

rtmwi

dp

C

rtaaTso7arn

M

otwomwtisc4fb(oη

ts

dg(tt“Pm


ext of the study and the definitions of the constructs. Basedn their inputs, the list of items was modified and an ini-ial list prepared for pretesting. A series of two pretests wereonducted in which respondents (from the planned samplingrame) responded to the test items after making a hypothet-cal laptop purchase at an incumbent e-retailer. Exploratoryactor analysis (EFA) was used to finalize the items. Theeries of pretests resulted in a three-factor solution withhe factors representing credibility, benevolence and assistiventent.

For the final data set, both EFA and confirmatory fac-or analysis (CFA) using LISREL 8 (Jöreskog and Sörbom996) supported the three-factor model. The three-factor modelad a goodness-of-fit (GFI) of .91, a comparative fit indexCFI) of .98 and a standardized root mean square residualf .047, indicating an adequate fit between model and data.urther, we assessed individual-item reliabilities by examin-

ng loadings of the measures on their respective constructs.ll items had loadings at or above .7 (see Table 1), indicat-

ng that they had good individual-item reliabilities (Hulland999). All constructs exhibited composite reliabilities of .7r more, thus indicating that the reliabilities of all the con-tructs are adequate (Bagozzi and Yi 1988). Finally, to completehe psychometric assessment of our measurement model, wexamined the discriminant validity of the constructs. Fornellnd Larcker (1981) suggest the use of average variancextracted to assess discriminant validity. Squared correlationetween any two constructs was less than the average vari-nce extracted by the constructs, and all measures loadedigher on intended constructs than on other constructs. Fur-her, average variance extracted for all the constructs exceeded50. Based on this assessment of convergent and discrimi-ant validity, the items presented in Table 1 were adopteds representing credibility, benevolence and assistive intent,espectively.

uyer’s product involvementEnduring involvement with a product category, measured

sing Zaichkowsky’s (1994) Personal Involvement InventoryPII) scale, assesses the product category’s personal relevance.he 10-item summated ratings scale, implemented with minorodifications, exhibited adequate reliability with a Cronbach’s

lpha of .89.6 For purposes of hypotheses testing, the summatednduring product involvement scale was converted into two dis-rete categories—below or above the 50th percentile score. Theigh involvement category had 125 respondents and the lownvolvement category 121 respondents. A t-test revealed that
he difference in mean product involvement levels was statisti-ally significant between the high and low product involvementroups (H = 15.0 and L = 25.7, p < .001, with lower numbers
6 On 7-point Likert scales, (1-strongly agree to 7-strongly disagree)espondents reported their perceptions of computers as being: impor-ant/unimportant; boring/interesting; relevant/irrelevant; exciting/unexciting;

eans nothing/means a lot to me; appealing/unappealing; fascinating/mundane;orthless/valuable; involving/uninvolving; and unfamiliar/familiar. Selected

tems were reverse coded, as appropriate.

(ow

tvvfip

ling 85 (2, 2009) 159–176 167

enoting greater product involvement), and appropriate for theurpose of our study.7

ontrol measuresAlthough most conceptualizations of trust suggest that expe-

ience with the other party is the most salient antecedent torust, literature also suggests that there may be some “person-lity type” trusting predispositions (e.g., McKnight, Cumminsnd Shervany 1998). We measured two such predispositions.he first predisposition, general faith in humanity, was mea-ured using the following item: “I believe that all business firmsnly want to do good for their customers” (1 = strongly agree,= strongly disagree). The second predisposition, referred to ascalculative form of trust, was measured as follows: “In the long

un, I will be better off trusting all business firms, rather thanot trusting them” (1 = strongly agree, 7 = strongly disagree).

anipulation checks

We manipulated the presence or absence of two specific typesf interactive tools in the online shopping environments: (1) IIMools and (2) IIC tools. The goal of the manipulation checksas primarily to assess whether the provision (or lack thereof)f such tools was noted by the respondents. For the IIM toolsanipulation, respondents indicated the extent to which “theebsite provides product comparison tools (e.g., comparing lap-

ops side by side) to help users manage the available laptopnformation” (1 = strongly disagree, 9 = strongly agree). A t-testhowed that the presence of such tools resulted in a higher meanompared to conditions where such tools were absent (7.40 vs..04, p < .01). To determine that this manipulation was unaf-ected by the IIC tools manipulation, an ANOVA was run withoth manipulations as factors and effect sizes were comparedsee Perdue and Summers 1986). The effect size for the sec-nd, nonmanipulated factor was considerably smaller (partial2 = .05) compared to the effect size for the manipulated fac-or (partial η2 = .37). Thus, the manipulation was considereduccessful with no confounding (Perdue and Summers 1986).

For the IIC tools manipulation (present vs. absent), respon-ents indicated the extent to which “the website provideslossaries to help users understand computer terminology1 = strongly disagree, 9 = strongly agree).” A t-test showedhat this manipulation was successful—the mean response inhe “present” conditions was significantly greater than in theabsent” conditions (6.49 vs. 3.94, p < .01). We repeated theerdue and Summers’ (1986) procedure to verify that that thisanipulation was unaffected by the other experimental factor

IIM tools). As an indication of a successful manipulation with-ut confounding, an ANOVA showed the effect size associatedith the second experimental factor was considerably smaller

7 Since our theoretical development of enduring involvement as a modera-or variable in Study 1 (and knowledge in Study 2) was at two levels of theariable, we use the median-split approach to facilitate the interpretation (andisual depiction) of the results. However, we also explored the robustness of ourndings by modeling product involvement as a continuous variable. The overallattern of results remains unchanged.


Table 2Effect of interactive information management (IIM) and interactive information comprehension (IIC) tools on buyer’s perceptions of seller’s assistive intent, credibilityand benevolence.a.

Interactive information management tools Interactive information comprehension tools

Present Absent Present Absent

Study 1 (N = 246)High product involvement 2.15 (.13)b 5.23 (.16) 3.46 (.13) 3.93 (.15)

2.43 (.10)c 3.05 (.12) 2.85 (.10) 2.62 (.12)3.61 (.11)d 4.98 (.14) 4.23 (.12) 4.36 (.13)

Low product involvement 2.78 (.13) 4.71 (.16) 3.22 (.15) 4.27 (.14)2.62 (.10) 3.27 (.12) 2.81 (.11) 3.08 (.11)4.08 (.11) 4.63 (.14) 4.05 (.13) 4.67 (.13)

Study 2 (N = 223)High product knowledge 2.76 (.17)b 4.45 (.18) 3.76 (.17) 3.45 (.18)

2.97 (.13)c 3.56 (.13) 3.51 (.12) 3.01 (.12)3.97 (.14)d 4.60 (.16) 4.62 (.15) 3.96 (.15)

Low product knowledge 2.86 (.20) 3.53 (.19) 2.95 (.18) 3.44 (.21)3.02 (.13) 2.84 (.13) 2.81 (.12) 3.06 (.14)4.01 (.16) 4.02 (.16) 3.68 (.15) 4.35 (.17)

a Measurement scales for the constructs are anchored at 1 = strongly agree to 7 = strongly disagree, with lower scores representing more favorable perceptions.b Buyer’s perception of seller’s assistive intent (mean scores with standard deviations in parentheses).

(f

dauaTcshtct

Da

abi

tanap

ci(TpciciIpo(sautrttet

c Buyer’s perception of seller’s credibility.d Buyer’s perception of seller’s benevolence.

partialη2 = .004) compared to the effect size for the manipulatedactor (partial η2 = .25).

Results

We first test specific hypotheses regarding the effects andeterminants of buyer’s perceptions of a new online seller’sssistive intent (H2–H5) and then provide a formal test of thenderlying mediation process (H1). Means and mean plots forssistive intent, credibility and benevolence are presented inable 2 and Fig. 2, respectively.8 The overall pattern of means isonsistent with the conceptual arguments advanced in previousections. Regression models 1–3 on which we base our specificypotheses tests are shown in Table 3. Effect sizes correspondingo overall model fit (see Cohen’s 1988 f2 in the table) indi-ate that the models are appropriate for purposes of hypothesesesting.

eterminants and effects of buyer’s perceptions of seller’sssistive intent

We test H2a and H2b, which posit a positive effect of seller’s
ssistive intent on buyer’s perceptions of seller credibility andenevolence, respectively, using Model 3 (see Table 3) thatncludes the effects of all variables. The results show that, after
8 The two control variables (pertaining to respondents’ trusting predisposi-ions) did not improve the analyses and were therefore dropped to retain thevailable degrees of freedom. Also, when interpreting tables and mean plots,ote that measurement scales for the trust constructs are anchored at 1 = stronglygree to 7 = strongly disagree, with lower scores representing more favorableerceptions.

apeii

i

ontrolling for other effects, seller’s assistive intent has a signif-cant positive effect on buyer’s perceptions of seller credibility.24, t = 2.53, p < .02) and benevolence (.56, t = 5.20, p < .01).hus, H2a and H2b are supported. H3a and H3b posit that therovision of task-facilitative tools will lead to favorable per-eptions among buyers of seller’s assistive intent. As shownn Model 1 (Table 3), the provision of IIM tools has a signifi-ant positive effect on buyer’s perceptions of seller’s assistiventent (−1.3, t = −17.4, p < .01)9 and the overall means for theIM present versus absent condition (2.5 vs. 5.0) provide sup-ort for H3a. Similarly, the significant effect of the provisionf IIC tools on buyer’s perceptions of seller’s assistive intent−.38, t = −5.3, p < .01) and the means for IIC present ver-us absent condition (3.3 vs. 4.1) support H3b. Finally, H4nd H5 focus on the moderating influence of buyer’s prod-ct involvement on the relationship between task-facilitativeools and buyer’s perceptions of seller’s assistive intent. Theesults of Model 1 show significant interactions between IIMools and product involvement (.29, t = 4.0, p < .01) and IICools and product involvement (−.15, t = −2.1, p < .05). Anxamination of the means (Table 2 and Fig. 2) show that IIMools lead to stronger perceptions among buyers of seller’sssistive intent for more involved than less involved buyers,roviding support for H4. Similarly, consistent with H5, theffect of IIC tools on buyer’s perceptions of seller’s assistive
ntent is stronger for less involved buyers compared to morenvolved buyers.
9 Due to the coding of variables, some negative coefficients denote a favorablempact of interactive tools. See Note in Table 3.

P. Gupta et al. / Journal of Retailing 85 (2, 2009) 159–176 169

F edibila presem

Mi

tm2fi2oimww(

ateisbaniitf

am

isoditpmtmp

imciFiIbaaaeva

ig. 2. Study 1a: mean plots for buyer’s perceptions of seller’s assistive intent, crre anchored at 1 = strongly agree to 7 = strongly disagree, with lower scores reanagement tools and interactive information comparison tools, respectively.

ediating role of buyer’s perception of seller’s assistiventent

We followed established procedures detailed in the litera-ure for conducting assessments of the hypothesized mediation

echanism (Baron and Kenny 1986; Muller, Judd, and Yzerbyt005). Specifically, three different models were run to infer theollowing (see Table 3): (1) there is a significant effect of thendependent variables on the final outcome variables (Model); (2) there is a significant effect of the independent variablesn the mediating variable (Model 1); and (3) the effects of thendependent variables are either reduced or eliminated when the

ediating variable is included in the model (Model 3).10 Sincee also posit the moderating influence of product involvement,e model the relevant interaction effects as well in Models 1–3

see Muller, Judd, and Yzerbyt 2005).First, in Model 2, IIM tools significantly impact credibility

nd the means for the IIM tools present versus absent condi-ion (2.5 vs. 3.2) are in the expected direction. The interactionffect of IIM tools and involvement on credibility is not signif-cant. Interestingly, the effect of IIM tools on credibility seemstrong for both low and high involvement buyers, perhaps sinceoth groups require IIM tools in their decision-making. Thus,lthough low involvement buyers may value IICs, they probablyeeded some of the IIM tools to work through their task. Thempact of IIC tools on credibility is not significant. However, the
nteraction between IIC tools and involvement is significant andhe means suggest that the effect of IIC tools on trust is strongeror less involved buyers. Focusing next on the effects of the
10 Following Sobel (1982) guidelines to test the statistical significance of medi-ted relationships, we find that all mediated relationships reported above thatet the Baron and Kenny (1986) guidelines were statistically significant.

tptasaap

ity and benevolence. Measurement scales for the assistive intent and trust scalesnting more favorable perceptions. IIM and IIC refer to interactive information

ndependent variables on perceptions of benevolence, the resultshow that the presence of IIM tools significantly impact benev-lence and the overall means (3.8 vs. 4.8) are in the expectedirection. The interaction between IIM tools and involvements significant and the means suggest a stronger impact of IIMools on benevolence for more involved buyers. Similarly, theresence of IIC tools significantly impacts benevolence and theeans (4.1 vs. 4.5) are in the expected direction. The interac-

ion effect of IIC tools and involvement is significant and theeans suggest a stronger impact of IIC tools on benevolence

erceptions for less involved buyers.Second, the results of Model 1 establish the condition that the

ndependent variables impact assistive intent, the hypothesizedediator. The effects of IIM tools and IIC tools on buyer’s per-

eptions of seller’s assistive intent, and their interactions withnvolvement are all significant and in the expected directions.inally, Model 3 completes the test of mediation. Here, assistive

ntent is included in the models testing the effect of IIM tools,IC tools and the interactions with involvement on credibility andenevolence. The impact of assistive intent on both credibilitynd benevolence, controlling for other influences, is significantnd the means are in the expected direction. Importantly, inccord with Muller, Judd, and Yzerbyt (2005) guidelines tostablish mediation, coefficients of all significant independentariables and appropriate interactions, observed in Model 2,re reduced in magnitude in the presence of the mediator inhe model. Overall, this pattern of results shows that buyer’serception of seller’s assistive intent partially mediates the rela-ionship between task-facilitative interactive informational toolsnd buyer’s initial trust perceptions. When task-related tools are
alient to the buyers, their beliefs of the seller’s task-relatedssistive intent are enhanced, which in turn impacts their beliefsbout the seller’s trustworthiness. Therefore, consistent with H1,erceived task-related assistive intent of the seller mediates the


Table 3Regression models for hypotheses testing (Study 1).

Construct (hypothesizedeffect)

Model 1 Model 2 Model 3

Assistive intent Credibility Benevolence Credibility Benevolence

Constant 3.72** (.07) 2.84** (.06) 4.32** (.06) 2.82** (.06) 4.28** (.06)Interactive information

management (IIM)tools [H3a, −]

−1.25** (.07) −.32** (.06) −.48** (.06) −.18** (.08) −.16* (.09)

Interactive informationcomprehension (IIC)tools [H3b, −]

−.38** (.07) −.01 (.06) −.19** (.06) .02 (.06) −.11* (.06)

Enduring involvement (I) .03 (.07) .10* (.06) .03 (.06) .09* (.05) .01 (.06)IIM tools × IIC tools .08 (.07) .10* (.06) .14** (.06) .10* (.05) .14** (.06)IIM tools × I [H4, +] .29** (.07) −.01 (.06) .21** (.06) −.04 (.06) .13** (.06)IIC tools × I [H5, −] −.15** (.07) −.13** (.05) −.12** (.06) −.10* (.05) −.07 (.06)Assistive intent [H2, +] .24** (.10) .56** (.11)Adjusted R2, Cohen’s f2

(effect size).58, 1.38 .14, .16 .24, .32 .16, .19 .31, .45

Model F(6,239) = 58.1, p < .01 F(6,239) = 7.6, p < .01 F(6,239) = 13.6, p < .01 F(7,238) = 7.6, p < .01 F(7,238) = 16.8, p < .01

Note: Measurement scales for all constructs are anchored at 1 = strongly agree and 7 = strongly disagree. Lower scores represent more favorable perceptions forassistive intent, credibility, and benevolence. Lower scores represent higher enduring involvement. IIM and IIC are coded as −1 (tool absent) and +1 (tool present).With this coding, the hypothesized regression coefficient signs are as follows: H2 (+), H3a (−), H3b (−), H4 (+), and H5 (−). See text for details. Standard errorsa ct size

ra

D

ottpfaiptt(c

aloiabk(odtso

eGsprTikkrosto

O

ftmceugf

re in parentheses. Cohen’s f2 (effect size): small effect size = .02; medium effe* p < .10.

** p < .05.

elationship between task-facilitative tools and buyer’s beliefsbout seller’s trustworthiness.

iscussion

Overall, the results of Study 1 lend empirical support tour thesis that the effects of task-facilitative tools on buyer’srust in a new online seller are mediated by buyer’s percep-ions of seller’s assistive intent. Further, the results suggest thatroduct involvement moderates the relationships between task-acilitative tools and assistive intent. When the tools providedre pertinent to buyers, their perception of seller’s assistiventent implicitly embedded in those tools (and, subsequently,erceptions of trust) is correspondingly higher. Interestingly,hese effects are observed without any explicit expressions ofhe seller’s intentions, and after controlling for various factorsgenerally studied in the extant literature) pertaining to seller’sompetence in the online shopping context.

It is also important to note that both buyer’s willingnessnd ability to utilize seller provided task-facilitation tools areikely to result in personal salience – and hence utilization –f such tools. The moderation effects of involvement observedn this study are a reflection of the enhanced willingnessnd, perhaps, ability to utilize the tools provided. Althoughuyer’s enduring involvement is generally correlated with theirnowledge (e.g., Mitchell and Dacin 1996), the Zaichkowsky1994) measure used in this study does not explicitly focusn knowledge. However, the fact that low involvement users
id see IIC tools as salient, suggests that they did utilize suchools. If their motivation levels were such that they only soughtimplifying cues or heuristics on the website (e.g., brand namesr recommendations), which were not provided, H5 (interaction
1moa

= .15; large effect size = .35. See Cohen (1988) for details.

ffect between IIC and involvement) would not be supported.iven that IIC tools require some effort to utilize, the results

uggest that in the context of tasks involving fairly complexroducts where making a poor choice would be potentiallyisky, even low involvement consumers spend some effort.hus, it is possible that the greater use of IIC tools by low

nvolvement users was perhaps driven by their relatively lowernowledge levels. However, one limitation of Study 1 is thatnowledge is not directly measured. Second, although theesults of Study 1 provide evidence that buyer’s perceptionsf seller’s assistive intent mediate the relationship betweenalient task-facilitative tools and trust, it would be instructiveo corroborate these results with actual process measures basedn protocols. Study 2 aims to address these limitations.

Study 2

verview

The product (laptop computers), choices available, task-acilitative tools and websites used in this follow-up study arehe same as in Study 1. A 2 (interactive information manage-

ent tools: present vs. absent) by 2 (interactive informationomprehension tools: present vs. absent) by 2 (product knowl-dge: high vs. low) between-subjects experimental design wastilized to test the hypothesized relationships. Subjects wereraduate and undergraduate students and individuals recruitedrom local church groups (N = 223). Note that unlike Study
, which measured subject’s product involvement, this studyeasured subject’s objective product knowledge. Further, the
rder of the dependent variable questions was randomized toccount for any demand effects of presentation order. We also

P. Gupta et al. / Journal of Retailing 85 (2, 2009) 159–176 171

Table 4Regression models for hypotheses testing (Study 2).

Construct (hypothesizedeffect)

Model 1 Model 2 Model 3

Assistive intent Credibility Benevolence Credibility Benevolence

Constant 3.40** (.09) 3.10** (.06) 4.15** (.08) 3.07** (.06) 4.10** (.06)Interactive information

management (IIM)tools [H3a, −]

−.59** (.09) −.10* (.06) −.16** (.08) .05 (.06) .10 (.07)

Interactive informationcomprehension (IIC)tools [H3b, −]

−.04 (.09) .06 (.06) −.002 (.08) .07 (.06) .02 (.06)

Product knowledge (K) .21** (.09) .16** (.06) .14* (.08) .14** (.06) .09 (.06)IIM Tools × IIC Tools .05 (.09) −.09 (.06) .004 (.08) −.10* (.06) −.01 (.06)IIM Tools × K [H4, +] −.26** (.09) −.19** (.06) −.16** (.08) −.13** (.06) −.06 (.06)IIC Tools × K [H5, −] .20** (.09) .19** (.06) .33** (.08) .14** (.06) .25** (.06)Assistive intent [H2, +] .53** (.08) .92** (.08)Adjusted R2, Cohen’s f2

(effect size).20, .25 .11, .12 .10, .11 .27, .37 .43, .75

Model F(6,216) = 10.0, p < .01 F(6,216) = 5.3, p < .01 F(6,216) = 5.1, p < .01 F(7,215) = 12.6, p < .01 F(7,215) = 24.5, p < .01

Note: Measurement scales for all constructs are anchored at 1 = strongly agree and 7 = strongly disagree. Lower scores represent more favorable perceptions forassistive intent, credibility, and benevolence. Higher scores represent higher objective product knowledge. IIM and IIC are coded as −1 (tool absent) and +1 (toolpresent). With this coding, the hypothesized regression coefficient signs are as follows: H2 (+), H3a (−), H3b (−), H4 (−), and H5 (+). See text for details. Standarde m effe

ctiies

t1qesAaticrrhnpc

1

wo

TcearpspcTs

I

cftctersc

rrors are in parentheses. Cohen’s f2 (effect size): small effect size = .02; mediu* p < .10.

** p < .05.

ollected retrospective protocols (Ericsson and Simon 1984) inhis follow-up study. Immediately after completing the exper-mental task and providing their assessments regarding trustn the online seller, we instructed subjects to provide open-nded responses to “describe your reasons for rating [the onlineeller’s] trustworthiness the way you did.”

Following recommended procedures for assessing the objec-ive product knowledge of subjects (e.g., Brucks 1985; Sujan985), content experts were employed to develop a battery ofuestions that would discriminate between high and low knowl-dge levels in the product category of computers. The finalet of questions, developed through pretesting, is provided inppendix A. The total number of correct responses was used asmeasure of objective knowledge. For purposes of hypotheses

esting, the summated product knowledge scale was convertednto two discrete categories—below and above the 50th per-entile score. The high product knowledge category had 121espondents and the low product knowledge group had 102espondents. A t-test revealed that the difference between theigh and low product knowledge categories was statistically sig-ificant and appropriate for this study (H = 8.44 and L = 3.79,< .001).11 The manipulation checks employed in Study 1 wereonducted here as well and were successful.

Results

Overall, the pattern of results was similar to those in Study, thus corroborating the key findings reported earlier (see

11 As in the case of Study 1, we also explored the robustness of our findingsith product knowledge modeled as a continuous variable. The overall patternf results remains unchanged.

sawsNrs

ct size = .15; large effect size = .35. See Cohen (1988) for details.

ables 2 and 4; mean plots are not reported in light of spaceonsiderations). The results show that subject’s product knowl-dge moderated the relationship between task-facilitative toolsnd assistive intent, suggesting that the moderating influenceseported in Study 1 are likely to be driven by differences inroduct knowledge. We also found support for the hypothe-ized mediating role of seller’s assistive intent. As Study 2 wasrimarily motivated to examine the underlying process morelosely, we first discuss insights based on protocol analysis (seeable 5) and then present details regarding the specific hypothe-es (H1–H5).

nsights from protocol analysis

We coded retrospective protocols from two experimentalonditions: IIM tools absent/IIC tools absent (i.e., no task-acilitative interactive tools provided) and IIM tools present/IICools present (i.e., all task-facilitative interactive tools provided)onditions. This approach provides the strongest contrast inerms of seller-provided task-facilitative tools and is useful forxploring the underlying thought processes linking the task-elated tools and perceptions of trustworthiness. Table 5 presentselected excerpts from the protocols from these experimentalonditions.

Two of the co-authors coded the responses on two dimen-ions. First, the responses were coded for mention of seller’sssistive intent (Yes/No). The coders were provided with aorking definition of seller’s assistive intent (“Expressions of

eller’s task-related helpfulness as perceived by the buyer”).ext, the coders noted whether the expressions of seller’s task-

elated helpfulness referred to the presence or absence of theeller’s task-facilitative intent. Table 5 reports selected proto-


Table 5Illustrative protocols reflecting presence or absence of seller’s assistive intent.

Selected protocols: when no IIM or IIC tools are provided Selected protocols: when both IIM and IIC tools are provided

“. . .the information about the laptops was very basic and there was noexplanation of what those things mean.”

“I think this e-retailer was pretty trustworthy. They seemed as though theyreally wanted to help me choose a good laptop for my needs. . .”

“Not enough information provided to compare items and make appropriatedecision.”

“It appeared more as if the way the website was set up was to ultimately informthe customer of the best possible choice for them and their type of use.”

“The only problem that I would have would be that I don’t knowEVERYTHING about computers and I know that I need a little bit moreassistance in looking for a computer than just a bunch of listings.”

“. . .The site allowed for the ability to compare many different itemseasily which was extremely helpful and makes me believe they are moretrustworthy.”

“This e-retailer is horrible. . . It’s hard to trust a site that doesn’t give youany help in shopping for something as costly and important as a laptop.”

“I liked the way you were able to search for what you wanted. The site’sdesign made me feel that this retailer had a good grasp on consumer needs.”

“. . .The site did not help me at all and made me guess which laptop link toclick on. It felt very shady.”

“AtoZTronics understands what the customer needs, and provides helpfulinformation to assist the customers that do not know a lot about computers.”

“. . .In no way did it assist me in making a sound decision for my nextcomputer. You had to search way too much to find exactly what youwere looking for. For people that are not computer savvy, they would beextremely frustrated looking at all the computer terminology and notknow what any of it means.”

“This is my first time at this site, so I don’t know enough to judge this retailer.It did try to help me as much as possible.”

“. . . I did think that this e-retailer’s motives were to help the customer makethe best decision. Putting up competitor prices as well as your own andoffering side-by-side comparing seemed to prove that. . .”

“I do not believe they were trustworthy because they did not help me withany aspect of choosing a computer, all they did was tell me a littleinformation about the computer. However, I do not know a lot aboutcomputers so I need a question answer section that could specify myneeds. They should have given me a list of laptops that fit my requiredneeds; they also needed to explain terminology.”

N ler’s]a e to im

ctaar

pOtIacpiatafsnc

ttptptm

c(sa

R

(aaifnaeaimHshttk

ote: Subjects were asked to “describe your reasons for rating [the online selssistive intent (or lack thereof). Minor editing of some protocols has been don

ols to illustrate subject’s contrasting thought processes relatedo trust assessments from both the IIM tools/IIC tools presentnd absent sites. Thoughts pertaining to seller’s assistive intentre mentioned frequently in these protocols, illustrating the keyole played by this construct.

Initial inter-coder agreement on these categorizations was 85ercent. The discrepancies were resolved through discussion.verall, 60.2 percent of the respondents made some men-

ion of seller’s assistive intent (presence or absence or both).nterestingly, subjects mentioned seller’s assistive intent (as

driver of trust assessments) with equal frequency in bothonditions: IIM tools/IIC tools absent and IIM tools/IIC toolsresent (56.8 percent vs. 65.0 percent; χ2 = .72, p < .397). Thats, regardless of whether task-facilitative tools are present orbsent, buyers seem to be seeking cues regarding seller’s assis-ive intent. This suggests that subjects’ evaluation of seller’sssistive intent is not triggered by the presence of task-acilitative tools. Even in the absence of such cues, subjectseem to evaluate seller’s assistive intent, highlighting the robust-ess of the mediating process that lies at the heart of ouronceptualization.

Next, an assessment was made to classify the mentions ofask-facilitative intent. In terms of whether the buyers perceivedhe presence of positive task-facilitative intent, the IIM toolsresent/IIC tools present condition had significantly more men-
ions than the IIM tools absent/IIC tools absent condition (76.9ercent vs. 36.0 percent;χ2 = 10.72, p < .001). On the other hand,he IIM tools absent/IIC tools absent condition had significantly
ore mentions of the absence of positive task-facilitative intent

Tba

trustworthiness the way you did.” Italicized text shows references to seller’sprove readability.

ompared to the IIM tools present/IIC tools present condition80.0 percent vs. 30.1 percent; χ2 = 14.77, p < .001). Protocolshown in Table 5 reflect these contrasting assessments of seller’sssistive intent.

eplication tests of H1–H5

The results of tests of specific hypotheses are as followssee Table 4). In Model 3, with all the variables, the effect ofssistive intent on perceptions of credibility (.53, t = 7.0, p < .01)nd benevolence (.92, t = 11.1, p < .01) are significant, provid-ng support for H2a and H2b. Turning next to Model 1 (testsor H3–H5), the effect of IIM tools on assistive intent is sig-ificant and the mean scores for the IIM present versus IIMbsent conditions (2.8 vs. 4.0) provide support for H3a. How-ver, the effect of IIC tools on assistive intent is not significant,nd H3b is therefore not supported. As regards the moderatingnfluence of objective knowledge (replacing product involve-

ent with objective knowledge in H4 and H5), in support of4, the interaction effect of IIM tools and product knowledge is

ignificant and the impact of IIM tools is stronger for buyers withigher product knowledge. Also, consistent with H5, the interac-ion effect of IIC tools and product knowledge is significant andhe effect of IIC tools is stronger for buyers with lower productnowledge.

As in Study 1, we examined the results of Models 1–3 (seeable 4) to test the hypothesized mediating role played byuyer’s perceptions of seller’s assistive intent (H1). The over-ll pattern of results is similar and supportive of a mediation

Retai

mv2iwoktlmpewrtotobiH

I

sbpipipsitstccrrstTbprda

cai

tek

lsscfcaspsnlgamttaitmo

iirfTtbiwsawsahoriowo


echanism.12 Also, the relative strength of this mediation patharies with buyer’s product knowledge. One difference in Studyis that the direct effect of IIC tools on assistive intent, credibil-

ty and benevolence is not significant. However, the interactionsith buyer’s product knowledge are significant and the effectsf IIC tools are more pronounced for buyers with lower productnowledge. Thus, it appears that the impact of IIC tools on assis-ive intent, and therefore trust, occur only for consumers withow knowledge, since such tools are salient only to them. The

ain effect of IIM tools is likely to exist because all consumersrobably used these tools, regardless of knowledge levels. How-ver, the significant interaction effect (H4) shows that IIM toolsere more salient for consumers with higher knowledge. All

emaining results are significant and lend support to our centralhesis regarding the mediating role played by buyer’s perceptionsf seller’s assistive intent in the relationship between interactiveask-facilitative tools and buyer’s initial online trust. The patternf results suggests that when the tools are salient to the consumer,uyers perceive assistive intent on the part of the seller, whichmpacts buyer’s beliefs about the seller’s trustworthiness. Thus,1 is supported in Study 2 as well.

Discussion

mplications for theory and practice

The research reported presents a new process-centric per-pective for understanding how initial trust forms betweenuyers and sellers in online shopping environments. Two studiesrovide evidence that buyers’ perception of a seller’s assistiventent, implicitly embedded in task-facilitative interactive tools,lays an important role in the formation of trust perceptionsn online shopping environments. Importantly, the trust-relatederceptions observed occur without any explicit expressions ofeller’s intentions. The results of mediation analyses (in Stud-es 1 and 2) and follow-up protocol analysis (in Study 2) showhat buyers use available task-facilitating tools provided by theeller to make an assessment of seller’s assistive intent and, inurn, ascribe trustworthiness. It is noteworthy that this signifi-ant impact on perceptions of trust occurs after controlling forompetence-related cues that have been the focus of previousesearch efforts (e.g., information privacy assurances, site secu-ity policies and returns policies). Furthermore, as noted above,eller’s intentions (or lack thereof) to help the buyer in the buyingask seem to be perceived in interactive task-facilitative tools.hese perceptions of seller intent, in turn, can strongly influenceuyer’s perceptions of initial trust in an online seller. Buyer’s
roduct involvement and product knowledge as moderators rep-esent important boundary conditions, suggesting that individualifference variables can lead to differing perceptions of seller’sssistive intent and trustworthiness for the same task-facilitative
12 As in Study 1, following Sobel (1982) guidelines to test the statistical signifi-ance of mediated relationships, we find that all mediated relationships reportedbove that met the Baron and Kenny (1986) guidelines were statistically signif-cant.

attdcphsm

ling 85 (2, 2009) 159–176 173

ools. In particular, Study 2 results suggest that these varyingffects are likely to stem from varying levels of buyer’s productnowledge.

Overall, the results of this research contribute to the emergingiterature on online trust by addressing a gap in our under-tanding of initial trust development in spatially and temporallyeparated buying environments. In the absence of (or deficien-ies in) other social cues that could serve as initial trust cues inace-to-face interactions, we show that initial trust developmentould be significantly affected by buyer’s assessments of seller’sssistive intent implicitly embedded in various task-related web-ite functionalities (after controlling for variables studied inrior research). We believe this process-centric perspective fortudying initial trust formation in online environments is bothew and promising. Conceptually, this perspective builds on aong-standing interest in the broader trust literature regardingood intentions as a potential antecedent of trust (e.g., Doneynd Cannon 1997; Lindskold 1978). Empirical support for theediating role played by buyer’s perceptions of a seller’s assis-

ive intent presented in this research, and how the strength ofhis mediation process varies with individual difference vari-bles such as product involvement and product knowledge, aremportant findings that can spur additional process-related inves-igations in this area. Research with a process-related focus is

uch needed, but largely remains absent, in the literature onnline trust. This paper, we hope, stimulates such work.

The results of this research also provide guidance to firms fornitiating and developing trust-based relationships by leverag-ng the interactive capabilities of the Internet. It is important foretailers to develop an understanding of potential mechanismsor enhancing initial perceptions of trust among potential buyers.o the extent that initial trust perceptions can impact potential

rial by a buyer, fostering such perceptions may be the differenceetween an opportunity to actually validate that trust throughnteractions and not initiating a relationship at all. Consistentith Urban (2004) call for more emphasis on developing a con-

umer advocacy strategy, our results suggest that initial trust inn online environment can be realized by empowering customersith interactive task-facilitative informational tools. However,

ince a buyer’s perceptions of assistive intent, and therefore trust,re stronger when personal relevance of the task-related tools isigh, the results of this study also provide a cautionary note tonline retailers. The effects of interactive tools are not likely toemain invariant across buying situations and buyer character-stics. If we were to change the context, the product categoryr even the task, the task-related tools required may change asell. Hitting this moving target represents a challenge and anpportunity for online retailers seeking a position of competitivedvantage by engendering trust among buyers. This underscoreshe continuing need for investment in personalization and cus-omization technologies for task-facilitation, and thereby, trustevelopment in online shopping environments. The on-going,omplex, dynamic nature of this approach for trust development
ossibly makes it difficult to implement and imitate—whichighlights its potential significance for developing a defen-ible position of competitive advantage in the electronicarketplace.

1 Retai

L

ammiwpspdtttlbtecelblsv

crpttitpif

1

2

3

4

5

6

7

8

9

1

1

1

A

A

B

B

B

B

C

C


imitations and future research directions

In addition to the general limitations and caveats associ-ted with laboratory experimental studies, several issues meritention. The first limitation pertains to the nature of the experi-ental task. Despite considerable efforts and resources invested

n making the experiment as externally valid as possible, itas not possible to fully duplicate “real world” online shop-ing behavior conditions. We recognize that the limited set ofpecific functionalities incorporated in the manipulated shop-ing interface may limit generalizability; however, this does notiminish the value of our work from the perspective of theoryesting and demonstrating the trust-building effects of interac-ive informational tools. Another limitation of our studies is thathe simulated nature of the shopping environment may haveessened the interest of respondents in finding a laptop thatest meets their needs. Under more realistic shopping condi-ions (involving actual transactions), buyers might conceivablyngage in more thorough information search and evaluation pro-esses. Thus, the results reported in our study are likely to beven stronger in more realistic settings in light of the greaterikelihood of the interactive tools being used more extensivelyy the respondents. Second, as the moderator variables (buyer’sevel of product involvement and product knowledge) are mea-ured and not manipulated, causal inferences related to theseariables should be made with caution.

An area worthy of further exploration concerns the role ofontextual factors on buyer’s trust-related perceptions in onlineetailing contexts. As reported in this research, buyer’s level ofroduct involvement or knowledge level (which could impacthe motivation and/or ability to attend to specific informa-ional tools) moderates the relationship between task-facilitativenformational tools and buyer’s perceptions of seller’s assis-ive intent. Additional process-oriented investigations of the rolelayed by other individual characteristics and contextual factorsn online retailing environments represent important areas foruture research.

Appendix A. Objective knowledge scale (computers)(correct answers in bold)

. What are the data bus widths of processors available in the consumer PCmarket today?(1) 16 bit, (2) 32 bit, (3) 16 and 32 bit, (4) 64 bit, (5) 32 and 64 bit.

. What is the primary function of the BIOS?(1) Connects the computer to the network, (2) Manages the power supplyto the computer, (3) Manages primary devices and helps bootcomputer, (4) Manages data transfers to and from memory, (5) I don’tknow.

. Which of the following acronyms is used in hard drive interfacedescriptions?(1) DIE, (2) AID, (3) IGB, (4) ATA, (5) SCI.

. How would you identify an optical mouse?(1) It has no cord connected to it, (2) It has no ball underneath it, (3) Ithas both a ball and a cord, (4) It is really small, (5) I don’t know.

. What is the key benefit of having a 64 bit processor instead of a 32 bit one?(1) It runs at more GHz and is therefore very fast, (2) It is actually worsethan a 32 bit computer, (3) It has more bits and therefore holds more data,(4) It can perform computations on a larger range of numbers, (5) Idon’t know.

C

C

ling 85 (2, 2009) 159–176

. When you start a computer, how does the computer determine the currenttime?(1) It communicates with a global time clock, (2) It holds and managestime and date on the hard drive in a clock which is powered by a smallbattery, (3) It holds and manages time and date in part of memorycalled ROM which is powered by a small battery, (4) It gets it from thenetwork when it starts up, (5) I don’t know.

. If you wanted a fast hard drive, which of the following attributes would bemost important?(1) Size, (2) Cache, (3) Interface, (4) RPM, (5) I don’t know.

. Which of the following is the correct way to measure LCD screen size?(1) Measure horizontally along the (physical) exterior of the monitor, (2)Measure diagonally along the (physical) exterior of the monitor, (3)Measure horizontally along the (viewable) interior of the monitor, (4)Measure diagonally along the (viewable) interior of the monitor, (5) Idon’t know.

. Can you put the memory from a desktop computer into a laptop computer?Yes, after all both are computers, (2) Yes, but only if both are the samebrand, (3) No, desktops and laptops have different memory slots, (4)No, desktops and laptops have different memory standards, (5) I don’tknow.

0. What is virtual memory?(1) Fake memory, (2) Memory that appears to be true, but cannot be used,(3) Memory management function that produces the illusion ofcomplete memory space for each program, (4) Just regular memory, (5)I don’t know.

1. When connected to the Internet, a computer. . .(1) Just connects and is totally unidentifiable just as other computers areonline, (2) Has a name given to it called the “NID”, (3) Has a name givento it called “computer name”, (4) Has an address given to it called “IPaddress”, (5) I don’t know.

2. You have a computer that works fine. However, when you start more 4–5programs it slows down drastically. What would you upgrade to avoid thisproblem?(1) Processor, (2) RAM, (3) Hard drive, (4) Get latest versions of theprograms you use, (5) I don’t know.

References

lba, Joseph W. and Wesley J. Hutchinson (1987), “Dimensions of ConsumerExpertise,” Journal of Consumer Research, 13 (March), 411–54.

riely, Dan (2000), “Controlling the Information flow: Effects on Consumers’Decision Making and Preferences,” Journal of Consumer Research, 27(December), 233–49.

agozzi, Richard P. and Youjae Yi (1988), “On the Evaluation of Struc-tural Equation Models,” Journal of the Academy of Marketing Science, 16(Spring), 74–95.

aron, Reuben M. and David A. Kenny (1986), “The Moderator-Mediator Vari-able Distinction in Social Psychological Research: Conceptual, Strategic,and Statistical Considerations,” Journal of Personality and Social Psychol-ogy, 51 (6), 1173–82.

art, Iakov, Venkatesh Shankar, Fareena Sultan and Glen L. Urban (2005), “Arethe Drivers and Role of Online Trust the Same for All Web Sites and Con-sumers? A Large-Scale Exploratory Empirical Study,” Journal of Marketing,69 (October), 133–52.

rucks, Merrie (1985), “The Effects of Product Class Knowledge on InformationSearch Behavior,” Journal of Consumer Research, 12 (June), 1–16.

heskin Research and Studio Archetype/Sapient Report (1999), “E-CommerceTrust Study,” [available at http://www.sapient.com/cheskin].

ho, Jinsook (2006), “The Mechanism of Trust and Distrust Formation and theirRelational Outcomes,” Journal of Retailing, 82 (1, 2006), 25–3.

ohen, Jacob (1988), “Statistical Power Analysis for the Behavioral Sciences,”2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates.

hurchill, Gilbert A. Jr. (1979), “A Paradigm for Developing Better Measuresof Marketing Constructs,” Journal of Marketing Research, 16 (February),64–73.

http://www.sapient.com/cheskin

Retai

D

D

D

E

F

F

F

G

H

H

H

H

H

J

J

J

J

J

K

K

L

L

M

M

M

M

M

M

M

M

M

M

P

P

P

P

P

S

S

S

S

S

S


ayal, Sandeep, Helene Landesberg and Michael Zeisser (1999), “How to BuildTrust Online,” Marketing Management, 8 (Fall), 64–9.

eutsch, Morton (1973), “Resolution of Conflict: Constructive and DestructiveProcesses,” New Haven, CT: Yale University Press.

oney, Patricia M. and Joseph P. Cannon (1997), “An Examination of the Natureof Trust in Buyer-Seller Relationships,” Journal of Marketing, 61 (April),35–52.

ricsson, K.A. and H.A. Simon (1984), “Protocol Analysis: Verbal Reports asData,” Cambridge, MA: MIT Press.

errin, Donald L. and Kurt T. Dirks (2003), “The Use of Rewards to Increase andDecrease Trust: Mediating Processes and Differential Effects,” OrganizationScience, 14 (1), 18–31.

ornell, Claes and David F. Larcker (1981), “Evaluating Structural EquationModels With Unobservable Variables and Measurement Error,” Journal ofMarketing Research, 18 (1), 39–50.

ortin, David R. (1997), “The Impact of Interactivity on Advertising Effective-ness in the New Media,” unpublished doctoral dissertation. Kingston, RI:University of Rhode Island.

anesan, Shankar (1994), “Determinants of Long-Term Orientation in Buyer-Seller Relationships,” Journal of Marketing, 58 (April), 1–19.

arris, Lloyd C. and Mark M.H. Goode (2004), “The Four Levels of Loyalty andthe Pivotal Role of Trust: A Study of Online Service Dynamics,” Journal ofRetailing, 80 (2004), 139–58.

äubl, Gerald and Valerie Trifts (2000), “Consumer Decision Making in OnlineShopping Environments: The Effects of Interactive Decision Aids,” Market-ing Science, 19 (Winter), 4–22.

eider, Fritz (1958), “The Psychology of Interpersonal Relations,” New York:Wiley.

offman, Donna L., Thomas P. Novak and Marcos Peralta (1999), “Build-ing Consumer Trust Online,” Communications of the ACM, 42 (4),80–5.

ulland, John (1999), “Use of Partial Least Squares (PLS) in Strategic Manage-ment Research: A Review of Four Recent Studies,” Strategic ManagementJournal, 20 (2), 195–204.

arvenpaa, Sirkka L., Noam Tractinsky and Lauri Saarinen (1999), “Con-sumer Trust in an Internet Store: A Cross-Cultural Validation,” Journal ofComputer-Mediated Communication, 5 (2),

arvenpaa, Sirkka L., Noam Tractinsky and Michael Vitale (2000), “ConsumerTrust in an Internet Store,” Information Technology and Management, 1(2000), 45–71.

ohnson-George, Cynthia and Walter C. Swap (1982), “Measurement of SpecificInterpersonal Trust: Construction and Validation of a Scale to Assess Trustin a Specific Other,” Journal of Personality and Social Psychology, 43 (6),1306–17.

ones, Edward E. and Keith E. Davis (1965), “From Acts to Dispositions: TheAttribution Process in Person Perception,” In Advances in ExperimentalSocial Psychology, Vol. 2, Berkowitz L. ed. New York: Academic Press.

öreskog, Karl G. and Dag Sörbom (1996), “LISREL 8: User Reference Guide,”Chicago: Scientific Software International.

elley, Harold H. (1967), “Attribution Theory in Social Psychology,” inNebraska Symposium on Motivation, David Levine and Lincoln , eds. NE:University of Nebraska Press, 192–238.

ollock, Peter (1999), “The Production of Trust in Online Markets,” Advancesin Group Processes,, 16.

indskold, Svenn (1978), “Trust Development, the GRIT Proposal and theEffects of Conciliatory Acts on Conflict and Cooperation,” PsychologicalBulletin, 85, 772–93.

ynch, John G. and Dan Ariely (2000), “Wine Online: Search Costs AffectCompetition on Price, Quality and Distribution,” Marketing Science, 19 (1),83–103.

aheswaran, Durairaj and Brian Sternthal (1990), “The Effect of Knowledge,Motivation, and Type of Message on Ad Processing and Product Judgments,”Journal of Consumer Research, 17 (June), 66–73.

cAllister, Daniel J. (1995), “Affect and Cognition-Based Trust as Foundationfor Interpersonal Cooperation in Organizations,” Academy of ManagementJournal, 38 (February), 24–59.

cKnight, D. Harrison, Vivek Choudhury and Charles Kacmar (2002), “TheImpact of Initial Consumer Trust on Intentions to Transact with a Website:

U

U

ling 85 (2, 2009) 159–176 175

A Trust Building Model,” Journal of Strategic Information Systems, 11,297–323.

cKnight, D. Harrison, Larry L. Cummins and Norman L. Chervany (1998),“Initial Trust Formation in New Organizational Relationships,” Academy ofManagement Review, 23 (July), 473–90.

itchell, Andrew A. and Peter A. Dacin (1996), “The Assessment of Alterna-tive Measures of Consumer Expertise,” Journal of Consumer Research, 23(December), 219–40.

ontoya-Weiss, Mitzi M., Glenn B. Voss and Dhruv Grewal (2003), “Deter-minants of Online Channel Use and Overall Satisfaction With a RelationalMultichannel Service Provider,” Journal of the Academy of Marketing Sci-ence, 31 (4), 448–5.

oorman, Christine, Gerald Zaltman and Rohit Deshpandé (1992), “Relation-ships Between Providers and Users of Market Research: The Dynamics ofTrust Within and Between Organizations,” Journal of Marketing Research,29 (August), 314–29.

, Rohit Deshpandé and Gerald Zaltman (1993), “FactorsAffecting Trust in Market Research Relationships,” Journal of Marketing,57 (January), 81–101.

oorthy, Sridhar, Brian T. Ratchford and Debabrata Talukdar (1997), “Con-sumer Information Search Revisited: Theory and Empirical Analysis,”Journal of Consumer Research, 23 (March), 263–78.

organ, Robert M. and Shelby Hunt (1994), “The Commitment-TrustTheory of Relationship Marketing,” Journal of Marketing, 58 (July),20–38.

uller, Dominique, Charles M. Judd and Vincent Y. Yzerbyt (2005), “WhenModeration is Mediated and Mediation is Moderated,” Journal of Personalityand Social Psychology, 89 (6), 852–63.

almer, Jonathan W., Joseph P. Bailey and Samer Faraj (2000), “The Role ofIntermediaries in the Development of Trust on the WWW: The Use andProminence of Trusted Third Parties and Privacy Statements,” Journal ofComputer Mediated Communication, 5 (3).

an, Yue and George Zinkhan (2006), “Exploring the Impact of Online Pri-vacy Disclosures on Consumer Trust,” Journal of Retailing, 82 (4, 2006),331–8.

erdue, Barbara C. and John O. Summers (1986), “Checking the Success ofManipulations in Marketing Experiments,” Journal of Marketing Research,23 (November), 317–26.

eterson, Robert A. (2001), “On the Use of College Students in Social Sci-ence Research: Insights from a Second-order Meta-analysis,” Journal ofConsumer Research, 28 (3), 450–61.

unj, Girish N. and Richard Staelin (1983), “A Model of Consumer InformationSearch Behavior for New Automobiles,” Journal of Consumer Research, 9(March), 366–80.

chlosser, Ann E., Tiffany Barnett White and Susan M. Lloyd (2006), “Con-verting Web Site Visitors into Buyers: How Web Site Investment IncreasesConsumer Trusting Beliefs and Online Purchase Intentions,” Journal ofMarketing, 70 (April), 133–48.

irdeshmukh, Deepak, Jagdip Singh and Barry Sabol (2002), “Consumer Trust,Value and Loyalty in Relational Exchanges,” Journal of Marketing, 66 (Jan-uary), 15–37.

obel, M.E. (1982), “Asymptotic Confidence Intervals for Indirect Effects inStructural Equations Models,” in Sociological Methodology, Leinhart S. ed.San Francisco: Jossey-Bass.

rinivasan, Narasimhan and Brian T. Ratchford (1991), “An Empirical Test of aModel of External Search for Automobiles,” Journal of Consumer Research,18 (September), 233–42.

teuer, Jonathan (1992), “Defining Virtual Reality: Dimensions Deter-mining Telepresence,” Journal of Communication, 42 (Autumn),73–9.

ujan, Mita (1985), “Consumer Knowledge: Effects on Evaluation StrategiesMediating Consumer Judgments,” Journal of Consumer Research, 12 (June),31–46.

rban, Glen (2004), “The Emerging Era of Customer Advocacy,” MIT SloanManagement Review, 45 (2), 77–83.

rban, Glen, Fareena Sultan and William Qualls (2000), “Placing Trust at theCenter of Your Internet Strategy,” MIT Sloan Management Review, Fall (1),39–48.

1 Retai

W

Y

Y


olfinbarger, Mary and Mary C. Gilly (2003), “eTailQ: Dimensionalizing,
Measuring and Predicting Retail Quality,” Journal of Retailing, 79 (3),183–98.
adav, Manjit S. and Rajan Varadarajan (2005), “Understanding Product Migra-tion to the Electronic Marketplace: A Conceptual Framework,” Journal ofRetailing, 81 (2), 125–40.

Z

ling 85 (2, 2009) 159–176

oon, Sung-Joon (2002), “The Antecedents and Consequences of Trust in
Online Purchase Decisions,” Journal of Interactive Marketing, 16 (2),47–63.
aichkowsky, Judith L. (1994), “The Personal Involvement Inventory: Reduc-tion, Revision, and Application to Advertising,” Journal of Advertising, 23(December), 59–71.

how task-facilitative interactive tools foster buyers’ trust in online retailers: a process view...

Documents