An Exploratory Attribution Analysis of Attitudes Toward the World Wide Web As a Product Information Source

ABSTRACT - It has been observed that the World Wide Web provides consumers with a higher level of control over communication with marketers than traditional media. Attribution theory is used to predict when consumers may evaluate the potential for such added control more or less favorably, and thus when consumers may evaluate the World Wide Web more or less favorably as an information source. Data from an exploratory study support the key hypothesis that attributions for past failures when exercising this control have a significant impact on current attitudes toward the Web.


Michelle L. Peterman, Harper A. Roehm, Jr., and Curtis P. Haugtvedt (1999) ,"An Exploratory Attribution Analysis of Attitudes Toward the World Wide Web As a Product Information Source", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 75-79.

Advances in Consumer Research Volume 26, 1999      Pages 75-79


Michelle L. Peterman, Wake Forest University

Harper A. Roehm, Jr., James Madison University

Curtis P. Haugtvedt, Ohio State University


It has been observed that the World Wide Web provides consumers with a higher level of control over communication with marketers than traditional media. Attribution theory is used to predict when consumers may evaluate the potential for such added control more or less favorably, and thus when consumers may evaluate the World Wide Web more or less favorably as an information source. Data from an exploratory study support the key hypothesis that attributions for past failures when exercising this control have a significant impact on current attitudes toward the Web.


In recent years, the impact of the World Wide Web on advertising media planning has been significant, and it continus to grow (Advertising Age Interactive Daily 1997). For many marketers, Web advertising is evolving from a media plan add-on to an integral aspect of the overall media strategy. A key question that arises from this trend is the following: How much emphasis should Web advertising receive relative to other media options?

In contemplating this issue, consumers’ attitudes toward the new medium are an important element to consider. The more favorable an individual feels toward the Web, the more likely he or she may be to use it and thus encounter Web-based advertising. At present, research on attitudes toward using the Web has only begun to accumulate. The present paper contributes to this developing literature. Specifically, we examine attributions about prior Web experiences and the roles that they play in shaping a consumer’s current attitudes toward it as a source of product information.


Interactivity and Consumer Control

Our analysis begins rather broadly with a comparison between the Web and media in the print and broadcast families. Relative to these traditional advertising media, the Web places fewer a priori constraints on the contributions of a marketer and a consumer within a product information exchange. Both parties have at least some latitude to adapt and modify input as the exchange progresses (see Figure 1). As such, the Web can be said to support a greater degree of "interactivity" than the conventional media. Formally, interactivity has been characterized as the extent to which "users can participate in modifying the form and content of a mediated environment in real time" (Steuer 1992, p.84).

In the Figure 1 framework, the Web’s advantage over broadcast media derives from the Web consumer’s flexibility to determine what product information is covered, the order in which topics are dealt with and the pace at which information is absorbed. The superiority of the Web with respect to print media comes from the Web marketer’s ability to adjust a message during an encounter to match the specific information needs and preferences of a particular consumer. For example, content can be broadened or narrowed to suit the consumer’s wishes, tone can be calibrated to the consumer’s personality, and so on (Haugtvedt and Roehm 1998).

The Web’s high level of interactivity may have important perceptual consequences. For instance, consumers may perceive that they have more personal control over information exchange processes and outcomes than is the case with the less interactive media types (Cross and Smith 1995; Haugtvedt and Roehm 1998; Hoffman and Novak 1996). When using the Web, an individual may feel empowered to lead the exchange in any number of different directions and realize that a savvy marketer can respond appropriately to follow his or her lead. In contrast, relatively less consumer control over process and outcome may be perceived with media in the print and broadcast categories. For example, with magazines or television, at least one party’s input may be recognized as fixed in advance, thereby constraining the nature of the exchange and its results. These observations are summarized in the following hypothesis:

H1:  Consumers perceive themselves as having a greater degree of control over Web-based information exchanges than over exchanges that occur via other, traditional media.

All else equal, having a high level of personal control over an information exchange may be attractive to consumers. uch control, for instance, may convey a sense of freedom of the sort that reactance studies show individuals seek to protect (e.g., Brehm 1966, 1972). If control is desirable to consumers, then to the degree that consumers perceive the Web as enhancing their control, attitudes toward the medium may be generally favorable.

At the same time, attribution theory (e.g., Weiner 1989; Weiner, Frieze, Kukla, Reed, Rest and Rosenbaum 1971) can be used to predict that there are also conditions under which the perception of added control will produce relatively negative attitudes toward the Web. As explained below, this comparative negativity seems likely to occur in cases where a person has assumed similar control in the past (i.e., has previously used web advertisements to learn about products), has failed to achieve a desired goal at that time (i.e., failed to find crucial product information) and expects that failure to be repeated again in the future.

Attribution Theory and Attitudes toward the Web

Attribution theorists contend that human beings have a natural motivation to "attain a causal picture of the world. That is, they want to know #why’ an event has occurred" (Weiner 1989, p. 325). This presumed explanatory tendency, which has been observed and investigated in myriad studies, benefits from a certain intuitive plausibility. For instance, most of us can probably recall a time today when we have paused to consider why something happened in a particular way. Even so, it is also intuitive to realize that an individual does not stop to puzzle over every single event that he or she observes. Formulating attributional explanations requires cognitive work. In many cases, a person may lack ability (i.e., he or she may be distracted) and/or motivation (i.e., the episode may not seem especially interesting) to devote enough cognitive resources to create attributions.

Given the task demands, the generation of causal attributions may be especially likely to occur for events that have cognitively motivating features. One feature that may provide a sufficient level of motivation is failure. Realizing failure is an unpleasant sensation. As an adaptive mechanism for managing discomfort and avoiding future mistakes, a person may be prompted to produce explanations for why the failure took place. In the present context, a pertinent implication is that Web information exchanges that end in failure may be especially likely to draw causal attributions



According to research by Weiner and colleagues (Weiner 1989; Weiner, Heckhausen, Meyer and Cook 1972; Weiner et al. 1971), when Web-based information exchange failures occur, they may attributed to any of four types of causes: (a) the consumer’s own abilities, (b) the consumer’s own motivation or level of effort, (c) the difficulty of a task and (d) sheer luck. These four causes differ on dimensions of stability and locus of control. Ability and task difficulty are classified as more stable or fixed than effort/motivation and luck. With regard to locus of control, ability and effort/motivation are factors internal to an individual. In contrast, task difficulty and luck are external factors. Of the two dimensions, stability is often considered to be dominant for explanations. This position is reflected in Weiner’s assertion that "in our culture, and within the confines of academic, occupational and athletic accomplishment, it has been clearly documented that ability and effort typically are believed to be the dominant causes of success and failure" (Weiner 1989, p. 328). In keeping with Weiner’s view, we focus on the stability dimension in this first, exploratory study and defer locus of control analyses to later, follow-up work.

When a past failure is attributed to stable rather than unstable factors, a person is more likely to anticipate a recurrence of the failure in future similar circumstances (Weiner 1989). To map this observation into te current issues of interest, suppose a consumer attributes an information exchange failure to his or her personal ability with the Web or to the difficulty inherent in navigating Web sites. In such cases, attribution theory predicts that the individual will have a strong expectation of failure for future exchanges. In contrast, suppose that a consumer attributes an information exchange failure to an unstable factor like personal motivation, or even sheer luck. In this instance, the expectation for repeat failure should be relatively weak.

Causal attributions and resultant expectancies may have important impacts on attitudes toward using the Web to learn about products in the future. Previous research has related attributions and expectancies to preferences for either having personal control over events or relinquishing control to another party (Anand and Stern 1985). When a prior failure is not expected again in the future, individuals understandably feel better about maintaining personal control than they do when the failure is expected to repeat itself. Thus, individuals who blame past Web information exchange failures on stable factors (ability, task difficulty) may anticipate future repeat failures and thus may prefer to relinquish control over further exchanges. As a result, these consumers may have relatively unfavorable attitudes toward the Web, because it requires them to maintain control. In contrast, individuals who blame past Web-related failures on unstable factors (motivation, luck) may not anticipate future failures and thus may not wish to relinquish control. Instead, they may hold relatively favorable attitudes toward the Web, even in the face of prior failures.

The Moderating Role of Self-Referencing

It is possible to speculate on conditions in which the relationship between failure attributions and Web attitudes may be more or less pronounced. We begin with the observation that in order for past failures to impact current attitudes, some mechanism is required for bringing the failure episodes into working memory. One process through which this might occur is via self-reflection on one’s past experiences. As past experiences in general are brought to mind, the subset that constitutes past failures will also be retrieved. If this logic holds, then we can predict that the relationship between Web attitudes and prior information exchange failures should be especially apparent in persons who naturally engage in a great deal of reference to past personal experiences. A different prediction would be made for individuals who do not have a strong tendency to review prior personal experiences. For these consumers, past failures with the WebCand more broadly, past occurences per seCmay rarely be contemplated. As a result, attributions for past failures may receive very little attention, and these attributions may thus have a comparatively small effect on Web attitudes.

Differences in habits of personal reflection can be captured in an individual difference measure called the Self-referencing Scale (Haugtvedt 1994). This eight-item scale measures the degree to which a person perpetually connects thoughts to the self (see Appendix). As one facet of self-referencing, several items on this scale inquire particularly about propensities to think about past personal experiences. Examples include "I often find myself using past experiences to help remember new information," and "I think it is easier to learn anything if only we can relate it to ourselves and our experiences." As a function of constantly ruminating on past experiences in general, a high self-referencer may spend a substantial amount of time thinking about Web-related information search failures. The associated attributions and future expectancies may thus be significantly related to the high self-referencer’s Web attitudes. Conversely, because a low self-referencer is less likely to reconsider own past history, the subset of personal events related to the Web sould receive fairly little attention. Therefore, the low self-referencer’s attributions and expectancies may often have little bearing on attitudes toward the Web. The preceding observations suggest these specific hypotheses:

H2a:  Among high self-referencers, attitudes toward the World Wide Web are more favorable when past Web failures are attributed to unstable factors than when past Web failures are attributed to stable factors.

H2b:  Among low self-referencers, attitudes toward the World Wide Web do not differ according to whether past Web failures are attributed to unstable factors or stable factors.

Relationships with Traditional Media

As a final note, it seems unlikely that these hypothesized relationships will hold for attitudes toward other, more traditional media. At the heart of Hypotheses 2a and 2b are consumers’ assessments of whether having a high level of control over an information exchange will be essentially a good thing or a bad thing. If, as Hypothesis 1 suggests, the less interactive media are perceived as conferring relatively little control upon the consumer, then judgments of whether having control will have positive or negative implications should have little impact on attitudes toward these media. These contentions suggest one additional hypothesis:

H3:  Attitudes toward traditional media will not differ according to self-referencing tendencies or whether past failures are attributed to unstable factors or stable factors.


Subjects and Procedures

The hypotheses derived above were investigated in an exploratory study. Questionnaires were administered to 62 undergraduate students as a part of an in-class extra credit assignment at a medium-sized, Southeastern university. The primary purpose of the questionnaire was to assess subjects’ attitudes and attributions with respect to using the World Wide Web to learn about products. For comparison purposes, parallel items about using magazines to learn about products were also included, in order to allow a contrast between findings associated with the Web and findings associated with a representative of the more traditional media. [The order of survey questions related to the Web and to magazines was manipulated so that half of the study participants received Web-oriented questions first, and half received magazine-oriented questions first. No order effects were obtained on any dependent measures (all ps>.20). Thus, the order variable will not be discussed further.]

On the first page of the questionnaire, respondents were asked to list thoughts about positive and negative aspects of gathering product information via the World Wide Web and via magazines. These questions were intended as a "warm up" exercise that would stimulate thinking about these two types of media.

On the second page of the questionnaire, respondents provided ratings for the World Wide Web and for magazines on separate sets of three, seven-point semantic differential scales. Anchors for these measures were "necessary/unnecessary," "useful/not useful" and "good/bad." These items were designed to measure participants’ attitudes toward the two media.

A series of filler questions followed, which served to distance forthcoming attribution questions from the attitude measures. Immediately prior to the attribution questions of interest, respondents were directed to "think back to the last Web page (magazine ad) that you did not find informative." To solidify this experience of failure respondents’ minds, they were then asked to report "what the Web page (magazine ad) was about." Participants were then asked to choose from the following four possible explanations for the failure to garner desired information: (a) "I would have found it more informative if I were mre motivated." (b) "I would have found it more informative if I had better abilities." (c) "I would have found it more informative if it weren’t for bad luck." (d) "I would have found it more informative if it weren’t for the difficulty of using the site." These four possible attributions were selected specifically to provide explanations related to stable (ability, difficulty) and unstable (motivation, luck) factors.

After responding to additional filler items, subjects encountered questions relating to perceived control over information exchanges. In particular, respondents were asked to rate the degree of control a consumer has over communications that are conducted via the Web (via magazines) on a seven-point semantic differential scale anchored by "complete control/no control."

Following a final batch of filler items, subjects completed the Self-referencing Scale (Haugtvedt 1994; see the Appendix) and provided some classification information including gender.


Testing Hypothesis 1.

In Hypothesis 1, it was predicted that consumers would perceive themselves as having especially high control over Web-based information exchanges, as compared to those in other media. Support for Hypothesis 1 would be obtained if participants’ ratings of consumers’ control over Web communication were higher than their ratings of consumers’ control over magazine communications.

A one-way, repeated measures analysis of variance was conducted on respondents’ Web and magazine control ratings. Means differed significantly in the anticipated directions, at 3.66 and 3.19 for Web and magazine, respectively (F (1, 123)=11.05, p<.01). The scales were keyed such that higher values indicated more consumer control. Thus, the higher mean for the Web indicates that a greater degree of consumer control was perceived for the Web than for magazines. Hypothesis 1 thus received support.

Preliminary Data Processing for Testing Hypotheses 2a and 2b.

In the context of the current study, Hypotheses 2a and 2b specifically suggest that high self-referencers who blame past failures on stable factors (ability, task difficulty) should report less favorable attitudes toward the Web than high self-referencers who blame past failures on unstable factors (motivation, luck). However, low self-referencers’ attitudes should not differ by attribution type. In order to test these expectations, a dependent variable was required to represent attitude toward the Web, and independent variables were required to represent attribution type (stable vs. unstable) and self-referencing level (high vs. low). Several steps were necessary to create these variables prior to the main analysis.



First, a factor analysis was conducted on the three Web attitude items. Examination of the scree plot suggested that these three items loaded on one factor. The individual item scores were averaged to produce a single Web attitude score for each respondent. This factor was reliable (a=.82).

Having created the necessary dependent variable, attention was then turned to developing independent variables. For self-referencing level, the eight Self-referencing Scale items were factor analyzed (one factor supported, (a=.85), scores were averaged for each respondent, and then the distribution of individual scores was split at the median (median=5.26) to designate high and low self-referencing groups. The second independent variable, attribution type, was derived from the explanations that subjects chose for their Web information search failures. As previously mentioned, participants were given four choices, relating to their own motivation, their own abilties, bad luck and difficulty of site usage. Following customs within he attribution literature, ability and difficulty explanations were classified as "stable attributions" and motivation and luck explanations were classified as "unstable attributions."

After completing these preliminary data processing tasks, a two-way analysis of covariance was conducted, with Web attitude as a dependent variable and self-referencing level (high versus low) and attribution type (stable versus unstable) as independent variables. In addition, a covariate, gender, was included to increase the sensitivity of the analysis (Keppel 1991). [Table 1 illustrates that results without the covariate entry were quite similar in terms of the pattern of means across cells. Without the covariate, the differences in these means approached significance and thus lent directional support to the primary theoretical hypotheses.] Means from this analysis conformed to expectations and are listed in Table 1.

Aside from the gender covariate (F (1, 47)=8.28, p<.01), the two-way interaction between self-referencing level and attribution type was the only significant effect (F (1,47)=3.99, p<.05). Examination of simple effects revealed that for high self-referencers, Web attitudes were significantly more positive when a past failure was attributed to an unstable factor than when it was attributed to a stable factor (F (1, 47)=5.71, p<.02). This finding supports Hypothesis 2a. However, for low self-referencers, Web attitudes did not differ as a function of attribution type (F (1, 47)=.34, p<.56). This latter result supports Hypothesis 2b.

Testing Hypothesis 3

The specific implication of Hypothesis 3 is that the effects on Web attitudes that were described immediately above should not hold for magazine attitudes. To test this prediction, a two-way analysis of covariance was conducted that was similar to the prior ANCOVA in several respects. The first independent variable, as before, was self-referencing level. The second independent variable, attribution type, was developed by classifying failure attributions for magazines into stable and unstable factors, as previously described for Web attributions. Gender was again included as a covariate. The dependent variable was a measure of magazine attitudes that was developed by factoring (one factor supported, (a=.83) and averaging scores on individual magazine ratings.

In contrast with the result for Web attitudes, the ANCOVA model for magazine attitudes was not statistically significant (F (4, 57)=.83, p<.52). [The model without the covariate was also not statistically significant (F, 3, 58)=1.11, p<.35). Thus, as predicted in Hypothesis 3, magazine attitudes did not differ as a function of self-referencing or failure attributions.


The results of our study lend empirical support to two intuitive notions. The first is that consumers do seem to feel more in control of the information exchange process when using the Web to learn about products than when other media are used for the same purpose. The second is that this control sometimes works in the Web’s favor, yet also sometimes works against it. Past failures when exercising this control seem to loom large in the minds of consumers, particularly for those who are predisposed toward self reflection. When these failures seem likely to happen again in the future, attitudes toward using the Web are negatively influenced.

The primary contribution of these findings is to identify attributions for past experiences as one factor that can influence attitudes toward the Web as a source of product information. Future research might profitably examine the relationship between Web attitudes and other aspects of experience, such as expertise and familiarity.

Finally, certain limitations to these exploratory findings are worth noting. [The authors thank Reviewer 26 for identifying these issues and thus contributing to the future direction of this line of study.] Whereas our interpretation of the current data has emphasized the difference in felt control observed between the Web and traditional media, it is also possible that other differences between the media classes are perceived by consumers and that these differences may also contribute to the differences in Web and magazine attitude pattens. For example, consumers may perceive magazine advertising as a more biased source of product information than Web pages. Further work on the current topic of interest is needed to investigate such other potential differences in perception and their relationships to attitudes toward the Web and other media.



Advertising Age Interactive Daily(1997), "Q2 Internet Ad Spending Up" (October 8).

Anand, Punam and Louis W. Stern (1985), "A Socialpsychological Explanation for Why Marketing Channel Members Relinquish Control," Journal of Marketing Research, 27 (November), 365-376.

Cross, R. and J. Smith (1995), "Customer-Focused Strategies and Tactics," in Interactive Marketing, ed. E. Forrest and R. Mizerski, Lincolnwood, IL: American Marketing Association, NTC Business Books.

Brehm, J.W. (1966), A Theory of Psychological Reactance, San Diego, CA: Academic Press.

Brehm, J.W. (1972), Responses to Loss of Freedom: A Theory of Psychological Reactance, Morristown, NJ: General Learning Press.

Haugtvedt, Curtis P. (1994), "The Propensity to Self-Reference," paper presented at the Society for Consumer Psychology Annual Meeting, St. Petersburg, FL.

Haugtvedt, Curtis P. and Harper A. Roehm, Jr. (1998), "Beyond Description: Interactive Communications as a Tool for Theory Testing, Expansion, and Boundary Identification," paper presented at the Society for Consumer Psychology Annual Meeting, Austin, TX.

Hoffman, Donna L. and Thomas P. Novak (1996), "Marketing in Hypermedia Computer-Mediated Environment: Conceptual Foundations," Journal of Marketing, 60 (July), 50-68.

Keppel, Geoffrey (1991), Design and Analysis: A Researcher’s Handbook, Englewood Cliffs, NJ: Prentice Hall.

Roehm, Harper A. Jr. (in process), "Persuasion and the New Media," unpublished dissertation, Marketing Department, Ohio State University.

Steuer, Jonathan (1992), "Defining Virtual Reality: Dimensions Determining Telepresence," Journal of Communications, 42 (4), 73-93.

Weiner, Bernard (1989), Human Motivation, Hillsdale, NJ: Erlbaum.

Weiner, Bernard , Irene Frieze, Andy Kukla, Linda Reed, Stanley Rest and Robert M. Rosenbaum (1971), Perceiving the Causes of Success and Failure, Morristown, NJ: General Learning Press.

Weiner, Bernard , Heinz Heckhausen, Wulf-Uwe Meyer and Ruth E. Cook (1972), "Causal Ascriptions and Achievement Behavior: A Conceptual Analysis of Effort and Reanalysis of Locus of Control," Journal of Personality and Social Psychology, 21 (2), 239-248.



Michelle L. Peterman, Wake Forest University
Harper A. Roehm, Jr., James Madison University
Curtis P. Haugtvedt, Ohio State University


NA - Advances in Consumer Research Volume 26 | 1999

Share Proceeding

Featured papers

See More


Non-normative influence of self-decided prices on product-related inferences

Sudipta Mukherjee, Virginia Tech, USA
Mario Pandelaere, Virginia Tech, USA

Read More


Unexpected-Framing Effect: Impact of Framing a Product Benefit as Unexpected on Product Desire

Monica Wadhwa, INSEAD, Singapore
Christine Kim, Hong Kong University of Science and Technology
Amitava Chattopadhyay, INSEAD, Singapore
Wenbo Wang, Hong Kong University of Science and Technology

Read More


Pretty Healthy Food: How Prettiness Amplifies Perceived Healthiness

Linda Hagen, University of Southern California, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.