Believe It or Not:Factors Influencing Credibility on the Web
C.Nadine Wathen
Faculty of Information and Media Studies,The University of Western Ontario,London,Ontario, Canada N6A5B7.E-mail:cwathen@uwo.ca
Jacquelyn Burkell
Faculty of Information and Media Studies,The University of Western Ontario,London,Ontario, Canada N6A5B7
This article reviews selected literature related to the credibility of information,including(1)the general mark-ers of credibility,and how different source,message and receiver characteristics affect people’s perceptions of information;(2)the impact of information medium on the assessment of credibility;and(3)the assessment of credibility in the context of information presented on the Internet.The objective of the literature review is to syn-thesize the current state of knowledge in this area,de-velop new ways to think about how people interact with information presented via the Internet,and suggest next steps for research and practical applications.The review examines empirical
evidence,key reviews,and descrip-tive material related to credibility in general,and in terms of on-line media.A general discussion of credibility and persuasion and a description of recent work on the cred-ibility and persuasiveness of computer-based applica-tions is presented.Finally,the article synthesizes what we have learned from variousfields,and proposes a model as a framework for much-needed future research in this area.
Introduction
People seek information for comfort,support,empower-ment,the impetus and knowledge to act,or simply to learn. Information is normally seen as that which has meaning,in that it reduces uncertainty for the seeker(Belkin,1978; Buckland,1991;Dervin,1977);information can also,how-ever,increase uncertainly and create more dissonance(Rog-ers,1995).The information-seeking process is iterative,and depends largely on the seeker’s specific situation,along with the broader context(Dervin,1992,1997).The impact of information can be limited to learning:information can simply be stored and recalled.In many cases,however, there is a much larger impact.Information can be passed on to others,it can be used to make decisions,and it can affect attitudes and behaviors.We encounter a great deal of infor-mation in our everyday lives,most of which wefilter out, retaining only that which is useful to us.One of the criteria we use tofilter information is its credibility,or believability.
Although users of information face the issue offiltering out information,those who produce information have a different problem.Information producers would like theirs to be the message that is attended to,recalled,or acted upon. If a user rejects new information as not credible,that infor-mation will not be learned,nor can it have any other impact. Thus,the provision of credible information serves as a necessary,but not sufficient,component of any process designed to influence knowledge,attitudes or behavior.
In the psychological literature,a great deal of research has examined the issues of attitude formation and behavior change.In the mid-1980s,Petty and Cacioppo(1986)pro-vided a useful way for thinking about how people become persuaded—the Elaboration Likelihood Model(ELM).In their model,persuasive communications that invoke careful, cognitive evaluation lead to the most persistent behavior and attitude changes.This“central route”to persuasion, however,requires considerable effort,and is not possible for all persuasive messages encountered.By contrast,atti-tudes formed by less cognitively demanding means,such as classical conditioning,inferences,or heuristics—so-called “peripheral routes”—are less persistent,stable,and predic-tive of behaviour(Petty,Cacioppo,Sekides,&Strathman, 1988).A key early stage in the persuasion process is the receiver’s judgment of the credibility of the information.As such,credibility is an“extramessage cue”that is a critical aspect of the persuasiveness of a message(Petty&Ca-cioppo,1986).
Given that credibility strongly influences the impact of a message,it becomes important to understand how users decide what to believe.The question of what marks credible
©2002Wiley Periodicals,Inc.
DOI:10.1002/asi.10016
JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY,53(2):134–144,2002
information has been studied within various academic dis-ciplines,including information science,psychology,sociol-ogy,marketing,communications,and health sciences.Mes-sage credibility is generally agreed to result from an inter-action of source ,expertise, trustworthiness),message characteristics(related to mes-sage content,encompassing factors such as plausibility, internal consistency,and quality),and receiver ,cultural background,previous beliefs).This no-tion arises in even the earliest discussions of credibility.In an important review of credibility research and theory,Self presents a summary of early Socratic and Aristotelian ideas about credibility:“First,sources are credible because their message’s rightness is perceived by the audience.Second, sources are credible because they rightly read how to reveal them
selves to particular audiences.And,third,sources are perceived to be credible because of audience characteris-tics”(Self,1996,p.423).
Research on source credibility examines the impact of personal(or organizational)characteristics such as expertise or trustworthiness on the“believability”of the message that is delivered.Other bodies of literature examine the impact of message and receiver on credibility,while still other research investigates the interaction of these three factors on credibility assessment.The impact of delivery medium (which can be construed as an aspect of message source)on credibility assessment has not been explored to the same degree.Research that does examine media effects on cred-ibility tends simply to compare the credibility of messages in different media(Johnson&Kaye,1998)without explor-ing how these credibility differences arise.Thus,there are little data on questions such as the impact of personalized messages versus broadcast delivery,or the impact of chan-nelfidelity(high vs.low)on credibility assessment.One important question is whether new media introduce new factors into credibility assessment.For example,as we move from face to face to television as a communication medium,does image size become important in the determi-nation of message credibility?
The advent of the Internet as a new and widely used method for the delivery of information raises the question of how credibility is assessed in this medium.The number of Internet users is large,and ever-i
ncreasing:in2000,the number of North American Internet users exceeded100 million(e-land,2000).A well-documented concern of both information providers and information seekers on the Inter-net is the decline in the perceived quality of Internet infor-mation,or at least the growing belief that high-quality information will be impossible tofind amidst the vast amount of lower quality,unfiltered information(Eysenbach, 2000;Shon&Musen,1999).Thus,both information pro-viders and users of the Internet are interested in the question of how users can and will assess the credibility of informa-tion presented through that medium.
This literature review attempts,based on previous re-search,to provide some insight into this question.We begin by providing a definition of credibility.We then discuss,in turn,the factors that influence credibility in traditional me-dia(including personal interaction)and electronic informa-tion(including information on the Web).The specific case of Internet health information is then examined,followed by a discussion of the evidence that is reviewed in the article and general recommendations to help information producers ensure that their Internet information is seen as credible by audiences.
Finally,a proposed model for how users judge the cred-ibility of on-line information is presented.This model is in part an attempt to synthesize existing evidence and in part a call for empirical research to e
xamine the key gaps in the literature outlined in the review.
A Definition of Credibility
What do we mean by“credibility”?At its simplest, credibility can be defined as“believability”(Fogg,1999; Tseng&Fogg,1999).Credible sources are described as “trustworthy”and having“expertise”(Self,1996).Sources are judged as credible based on perceived competence, character,composure,dynamism and sociability(Burgoon, Bonito,Bengtsson,Cederberg,Lundeberg,&Allspach, 2000).Tseng and Fogg(1999)identify four types of source credibility.Presumed credibility arises from the assump-tions of the perceiver.For example,stereotypes about an object or source,such as“car salesmen are generally dis-honest,”can lead to a credibility judgment,such as“don’t believe what the car salesman says,”absent any other cues. Reputed credibility is based on source labels.Those sources labeled“Doctor”or“Professor,”for example,are perceived as credible by virtue of the label.Surface credibility is assigned based on a user’s simple inspection of superficial characteristics—i.e.,“judging a book by its cover.”Finally, experienced credibility is based on a user’sfirst-hand expe-rience with a source over time.This is the most complex, and reliable,method of making credibility judgments(Fogg, 1999;Tseng&Fogg,1999).
Research on credibility operationalizes the concept in one of three ways.First,credibility can be assessed directly by asking respondents to indicate whether information and/or information sources are believable.This direct as-sessment,however,is not always used.Instead,credibility of information is often inferred on the basis of proxy mea-sures.These proxy measures include:(1)knowledge change [on the relatively weak assumption that only credible infor-mation is processed enough to be ,Self, 1996)];and(2)attitude or behavior change[on the stronger assumption that information must be credible,and thus believed,before it can affect attitudes or , Petty et al.,1988)].
Credibility in Traditional Media
Message familiarity can influence judged credibility, with more familiar messages being judged as more credible (Self,1996).In some cases,receiver and source character-
JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—January15,2002135
istics interact in credibility judgments.Message sources chosen to match audience attitudes and context lead to higher perceived credibility and better recall of the infor-mation.Indeed,health behaviors
in a group that received a tailored message from a credible(and familiar)source were changed to comply with the message(Campbell et al., 1999).Early studies by Hovland and colleagues(1949, 1953,cited in Self,1996,showed that messages from both high-and low-credibility sources(those with more or less “trustworthiness”and“expertise”)were learned equally well.In assessment immediately following information pre-sentation,messages from high-credibility sources changed attitudes more than those from low credibility sources.In a follow-up4months after the initial delivery of the message, however,subjects could no longer remember the source of the message,and messages from high-and low-credibility sources assumed similar levels of persuasiveness.It may be, therefore,that the primary influence of credibility factors is at the point of“first contact”between the receiver and the message:the means by which the message is made persua-sive might not matter in the long run,only that it was persuasive in thefirst place.
Wilson and Sherrell(1993)performed a meta-analysis of empirical studies that measured factors influencing credibil-ity perceptions.Studies that qualified for the analysis mea-sured three types of effects:(1)those arising from manip-ulation of source pertise,attractive-ness,similarity to user);(2)those due to the the issue being communicated);and(3)those due to the medium or channel.Their operational definition of credi-bility involved internalization of the
message:“a commu-nication source possesses credibility if his/her statements are considered truthful and thus worthy of serious consideration.A member of a target audience may accept(internalize)the influence of the communication source because the advocated behavior is congruent with his/her own value system.”(p.102).The results of the meta-analysis indicate that(1)source effects overall have a relatively small effect on knowledge/attitude/behavior change or persuasion(ϳ9%);(2)when source characteris-tics are involved,expertise is the most important one;and (3)the source by receiver involvement interaction is signif-icant,and consistent with the predictions of the , that highly involved/motivated users take a central route, relying less on peripheral cues and more on the content of the message)(Wilson&Sherrell,1993).
Thus,the majority(but not all)of the available evidence conceptualizes credibility as a multidimensional construct whose two main source-related components are trustworthi-ness and expertise,along with dynamism,likeability,and goodwill,among others(see also O’Hara,Netemyer,& Burton,1991).In the determination of credibility,source effects interact with receiver-related factors,such as level of involvement and attitudes(Petty&Cacioppo,1990).If the audience is less involved with the topic,source character-istics will have more influence on the construct under ,knowledge/attitudes/behaviors).Source characteristics also interact with message-relat
ed factors, such as discrepancy(from previous beliefs),incongruity in content,or timing of source identification within the mes-sage.For example,if a message is high in discrepancy,low in incongruity,and if the source is identified early in the message,then high-credibility sources will tend to have a greater impact on knowledge/attitudes/behaviors than low credibility sources(Wilson&Sherrell,1993).
Slater and Rouner(1996)postulate that credibility as-sessments are based on(1)knowledge and attitudes about the source specifically;(2)attitudes about , inferred or reputed credibility);and(3)quality of the mes-sage,including its presentation,plausibility,and whether it is supported by data or good examples.The third set of factors,they claim,is not well studied.This research showed that the way a message is well-written,-produced,-organized)can influence how the source is perceived.Source credibility ratings made before message presentation can be mediated by message charac-teristics:well-presented messages lend credibility to the source,while poorly presented messages detract from cred-
TABLE1.Examples of factors influencing credibility(print and interpersonal media).
Source Receiver Message Medium Context •Expertise/Knowledge•Issue relevance•Topic/content•Organization•Distraction/“noise”
•Trustworthiness•,need for
the information)•Internal validity/consistency•Usability•Time since message
encountered
•Credentials•Prior knowledge of the
issue
•Plausibility of arguments•Presentation
•Attractiveness•Issue involvement•Supported by data or
examples
•Vividness
•Similarity to receiver beliefs/
context
•Values/beliefs/situation•Framing(loss or gain)
•Likeability/Goodwill/Dynamism•Stereotypes about source or
topicdocumented evidence
•Repetition/familiarity
•“Social location”•Ordering
136JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—January15,2002
ibility.However,if a source is initially seen as expert and credible,the mediating effect is not as strong:the message and source are more likely to retain credibility in the face of poor presentation(Slater&Rouner,1996).
Table1summarizes some of the variables related to source,message,receiver,context,and media characteris-tics discussed above in the context of traditional(mainly print and interpersonal)media.All of these factors are relevant in the on-line environment,though the interpreta-tion or application of them may differ.
Credibility in Electronic Media
Several authors in information science have developed ways to think about electronically delivered information. They identify how this type of information differs,at a cognitive,information processing level,from more tradi-tional forms.Olaisen(1990)discusses factors related to information quality and how“cognitive authority”is estab-lished for electronic information.He equates cognitive au-thority to influence,and influence to credibility.He states what is generally known in information science theory—that although personal information sources may be the most trusted,they are not necessarily the most expert:we turn to other sources when expertise or competence is required.He states:“when we are processing information we will give credit and authority to certain persons and sources depend-ing on our social location.One’s will greatly influence quality factors like ,reli-ability),relevance and perceived value of information”(p.
92).He goes on to speculate whether the electronic transfer of information has effectively changed our“social location”(our location in time and space and in a network of social relationships)and the types of social networks we inhabit (Olaisen,1990).For texts,cognitive ,credi-bility)ratings are made based on the perceived expertise of the author and/or publisher,or referral to the text from a credible source.Olaisen asks:“[given that]we can store, organize and distribute electronic information i
n a com-pletely different way than we organized printed information ...do we then give a different credibility or cognitive authority to ?”(p.92).
Olaisen(1990)among others(Fogg,1999;Petty&Ca-cioppo,1986),differentiates between factors related to sources/messages/receivers and those related to design fea-tures of the medium.Calling the former“cognitive”quali-ties,and the latter“technical”qualities,he ranks various factors from each in terms of importance to cognitive au-thority perceptions.For“cognitive”factors,influence,trust-worthiness,competence,reliability,and relevance deter-mine cognitive authority.On the“technical”side,form, novelty,accessibility,andflexibility are key.Combined, cognitive and technical features make up what he terms “institutional quality”(Olaisen,1990)—presumably,the greater the“institutional quality,”the higher the judgment of cognitive authority,and thus the more credible and per-suasive the information.
Interestingly,Olaisen goes on to say that cognitive au-thority tends to be attributed more to informal than formal sources of information.Intervening variables,such as time and extrinsic ,price)can also influence cognitive authority(the nearer in time or the more expensive the item, the more cognitive authority will be granted).
It has been proposed that the Internet combines the broad reach advantages of mass communication channels with the persuasion characteristics of interpersonal channels by al-lowing for give and take between the message source and receiver.The Internet,therefore,may have a greater ability than other mass media to make use of principles of con-sumer behavior to enhance information provision and up-take(Cassell,Jackson,&Cheuvront,1998).
In a direct comparison of“new”versus“old”media, Johnson and Kaye examined Web users’perceptions of the credibility of Web-delivered information compared to that delivered more traditionally.The study compared political information in paper-based newspapers,magazines,candi-date literature,or issue-oriented sources,to similar informa-tion sources on-line.The sample(Nϭ308)was a self-selected group(by virtue of choosing to complete the on-line survey)of regular consumers of political information who described themselves as heavy Internet users(Ͼ13 hours/week).Users found on-line newspapers and candidate literature to be significantly more credible than their tradi-tional counterparts.Magazines and issue-oriented material were found to be equally as credible whether delivered traditionally or on-line.Of interest,users found political issue-oriented Web sites more credible than either on-line newspapers or magazines.Candidate literature was not deemed credible.However,in terms of overall credibility of political information,no source,either delivered on-line or traditionally,was rated more than“
somewhat”credible—a statement more,perhaps,about the subject matter than the medium,or more indicative of the public’s general and growing skepticism about mass media(Johnson&Kaye, 1998).Of relevance to the topic at hand is that the medium, at least for political information,did not influence credibil-ity as much as the source of the information.Thus,when source and message are separated from medium,as is often not the case in the current literature relevant to this topic, there are factors that distinguish between the three,and influence credibility.
Rieh and Belkin(1998,2000),in two qualitative studies, examined cognitive authority and quality ratings used by scholars for information presented on the Web.They pos-tulated that these judgments serve asfiltering mechanisms, allowing people to more efficiently manage the amount of information they have to process.This is especially impor-tant in the Web environment,because the amount of infor-mation available is virtually limitless(Rieh&Belkin, 1998).According to these authors,the quality checks used by scholars for print materials—presence of peer review, refereeing,publisher reputation—are more difficult to apply in the Web environment.In addition,direct experience with the source is less likely to be available to users,given the
JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—January15,2002137
newness of the medium,and the often-ambiguous nature of the source.
Rieh and Belkin(1998)found that the types of informa-tion needs that lead scholars to the ,to learn about a new topic/get started in an unfamiliar area)do not differ from those triggering searches of traditional media.Quality judgments of Web information are also based on the per-ceived credibility of the source.Users in the Rieh and Belkin(1998,2000)studies sought source credibility mark-ers to replace those used for print media,identifying aspects such as suffixes to the URL ,.edu,.v). These,denoting educational,nonprofit,and government sources,respectively,were rated as more credible than were commercial sites().At the individual source level, users looked for reputed or referred credibility markers, such as a statement of the author’s credentials and institu-tional affiliations.These were seen as particularly important due to the lack of publication controls on the Web.Related to this is the use of referrals to specific sites from trusted sources(a colleague,known expert or journal article).These referrals were often cited as reasons for visiting a given site. Of interest was that,even when seemingly believable infor-mation was found on the Web using the credibility and quality checks above,the scholars serving as subjects in the first study still felt the need to verify the information by referring back to a traditional expert source.
The authors conclude that there are seven criteria that affect credibility of Web information:source,cont
ent,for-mat,presentation,currency,accuracy,and speed of loading:“since authority is not necessarily identified in the Web environment,[subjects]used many different,and new char-acteristics of Web information objects(pages)in order to make their judgments of authority and credibility”(p.288).
Fogg and colleagues(Fogg et al.,2000)conducted a large-scale(Nϭ1,410)on-line survey of Web users in the United States and Finland to examine the elements that user’s report as affecting their judgments of Web site cred-ibility.They found seven factors that influence credibility:“real-world feel”;ease of use,expertise,trustworthiness, and message tailoring had positive effects on credibility ratings.Commercial implications and amateurism nega-tively affect credibility.These authors conclude that people use the same types of criteria for assessing on-line informa-tion as they use for traditional media—indeed they seek out familiarity in this new environment(Fogg et al.,2000).
Information on the Internet may benefit from a sort of presumed credibility.Many users assume that if someone expends the effort andfinancial resources to provide infor-mation on a Web site,then the quality of that information must be at least adequate(Feightner,Quintana,Marshall,& Wathen,1999).In addition,surface credibility is evident in the assessment of Internet information.Information coupled with a well-designed interface and attractive graphics may result,in the absence of more substantive cues,in a ten-dency for users to make a positive credibility judgment (Critchfield,1998).Some surface c
haracteristics,such as errors unrelated to content,negatively influence credibility.Tseng and Fogg(1999)conclude that“small errors by computers have disproportionately large effects on percep-tions of credibility”(p.43).Links to and from other Web sites may act as referrals,implying third party endorsement, which increases the reputed credibility of the site(Fogg& Tseng,1999).First-hand experience with a Web site also informs credibility judgments(Reih&Belkin,2000),indi-cating that experiential credibility is important in assessing Web-based information.
There is some evidence of an interesting interaction between user characteristics and errors in an electronic information interface.User expertise tends to influence how computing devices and electronic information are per-ceived:a more technologically sophisticated user will tend to scrutinize the source or object much more closely and more critically,leading to greater skepticism about the object and the tendency to assign low credibility(Tseng& Fogg,1999;note,however,that there are some mixed results in this area,reviewed in Tseng&Fogg,1999).In evaluation of Web-based information,there also appears to be a utilitarian aspect to credibility assessment.User need for information is correlated to credibility judgments—the greater the need,the more likely the user is to accept the information as credible(Tseng&Fogg,1999).These char-acteristics interact with two types of evaluation errors:(1) gullibility errors,defined as“blind faith”in a technology or source;and(2)incredulity errors,defined as“blin
d skepti-cism”in the source.Simply speaking,novices to both the topic and the computing process and those with a greater need for information will tend to make gullibility errors, while expert users will tend to incredulity.Fogg and Tseng point out that most educational and research efforts are directed at decreasing gullibility errors by educating nov-ices.However,the goal should be tofind the right balance of knowledge,experience,and needs to decrease gullibility errors without increasing incredulity errors—i.e.,to design systems that match the needs of users and instill confidence in the information being provided.
Internet Health Information
An estimated43%of Internet users go on-line to gather health information on over34,000health related Web sites (DMOZ Open Directory Project,2000).There are numerous advantages to using the Internet to seek and provide health information.Consumers can access information in the pri-vacy of their own home and at convenient times.Further-more,the information can be tailored to the specific needs and desires of the consumer,based on such factors as client ,a user can specify the language in which the information is provided)and learning , a choice between graphical or written forms of the same information).In addition,the Internet presents the opportu-nity to support written material with multimedia features, and provides a level of interactivity not available in other for
mats.The Science Panel on Interactive Communication and Health(SciPICH)identified the following advantages
138JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—January15,2002

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。