Stacy L. Wood & C.

Page Moreau

From Fear to Loathing? How Emotion Infiuences the Evaluation and Eariy Use of innovations
Innovation adoption is rarely a short process for consumers; accordingly, recent research has explored adoption as a dynamic process that is characterized by changing patterns, or diffusion, of consumer use of the innovation. This research suggests that adoption is rarely a neutral process and that consumers can experience strong emotions in the initial use of innovations. However, given such emotions, two opposing arguments can be made as to whether the inclusion of emotional responses increases the predictive power of traditional models of diffusion. On the one hand, experienced emotion may simply be a function of gained benefits and. as such, may already be captured in extant models through cognitive assessments of net benefits. On the other hand, and as data from two empirical and longitudinal studies demonstrate, the learning process is potentially emotion generating (independent of net benefits), and this emotion colors product evaluations. The emotional influence is sizable and, importantly, not a straightforward case of "easier is better." In this work, the authors present the E3 (expectation ~> emotion -^ evaluation) model, which describes how managers can better predict and influence the successful diffusion of complex technological products.

s technological innovations reach the consumer market at an unprecedented pace, recent reports have highlighted Ihe influence of consumers" expectations of and actual experiences in using a product on a product's ultimate success. For example, perceptions of usage difficulty have caused a significant number of consumers to delay purchases (e.g., 48% of potential digital camera buyers), and actual usage difficulty has caused them to retum previously purchased products (e.g., 30% of all home networking products: Endt 2004, p. E8). Epitomizing a company caught by these trends. Royal Philips Electronics, which has suffered as a result of consumers' rejection of complex new products, recorded huge losses in 2000 and 2001 and reduced its workforce by 55,000 employees (The Economi.st 2004). To understand tbe scope of the problem. Chief Executive Officer Gerard Kleisterlee setit 100 Philips managers home one weekend wilb aji assignment to get various Philips gadgets lo operate. Many failed and "retumed frustrated and some even angry," wbereas

Stacy L. Wood is Associate Professor of Marketing. Moore School of Business, University of South Carolina (e-mail: [email protected]). C. Page Moreau is Assistant Professor ot Marketing, Leeds School of Business, University ot Colorado, Boulder (e-mail: [email protected]). The authors thank Bill Bearden, Meg Campbell, Bill Dillon. Paul Herr, Amna Kirmani. Don Lehmann. and John Lynch for their comments and assistance in all phases of this research, and they thank Jen Dale for her research assistance. They also gratefully acknowledge research funding from the Leeds School of Business and the Moore School of Business and the insightful comments of the three anonymous JM reviewers. Both authors contributed equally to this work. To read or contribute to reader and author dialogue on this article, visit http://www. tnarketingpower.com/jmt^.

"anotber group that succeeded returned quite proud" (Endt 2004, p. E8). These managers were struck by tbe strengtb of tbeir own response to product complexity in use. but we note the emotional nature of the managers' responses. As tbis example and our own experiences sbow. learning to use a new product can evoke an emotional response, independent of the emotions produced by the attributes or benefits of tbe product itself. Tbe goal of this researcb is to understand the underlying causes of this type of emotional response and study its ramifications lor complex product innovations.' Altbougb tbe majority of prior innovation research has focused on a point of adoption, tbe field bas observed recent calls for more process-oriented research (e.g.. Golder and Tellis 1998; Rogers 1996) to look beyond purchase and address how early use afiecls ultimate adoption (Shih and Venkalesh 2004). This usage focus is increasingly important as managers strive to understand why some consumers fail to keep, or to keep using, new products. To better understand tbis consumer-centric process of innovation, we develop the E^ (expectation > emotion > evaluation) model of emotional influence, test Ibe model in two studies, and describe how managers can use tbe fmdings to help predict and influence new product .success.

lAtthtnigh many of the 4(XX) diffusion articles and books published to date have demonstrated that perceived produci complexity causes consumers to delay or avoid the purchase of a new product, fewer Ihan \% have examined the influence of perceived complexity on pttstadopiion behavior (Rogers 1996. p. 91). This sparse attention is .surprising, given that consumers' iniiial usage experiences are complex, especially for tectinological products (Mick and Foumier 1998). and are likely to influence product retum rate, future brand loyalty, word-of-m<>uth communications, and desire to purchase other technologies. Journal of Marketing Vbl. 70 (Juiy 2006), 44-57

2006, American Marketing Association ISSN; 0022-2429 (print), 1547-7185 (electronic)

44

Expectation, Emotion, and Evaluation

It has been well established that nev^ product success can be reliably predicted by an evaluation of the product's benefits, costs, and relative advantage over competing alternatives (Moreau, Lehmann, and Markman 2001; Rogers 1996). In Rogers's (1996) classic work on diffusion, he also identifies the innovation's "complexity" (i.e., the perceived magnitude of the learning costs required to achieve its benefits) as an indicator of an innovation's success. Complexity is typically indicative of slower diffusion rates (Rogers 1996) and can even create disutility through "feature fatigue" (Thompson. Hamilton, and Rust 2005). In the formative period of trial and early use of new. complex products, consumers learn how to achieve promised benefits. Rogers calls this "how-to" knowledge and acknowledges that it has been overlooked to date as a key factor in diffusion success. However, interest in use is growing; to wit, Shih and Venkatesh (2004) recently identified "usage diffusion" as an important conceptualization to understand better when consumers will fully integrate a newly adopted innovation into their lives and thus create a strong foundation for the innovation's current profitability and future growth. Simply bringing consumers to the point of trying a new innovation, let alone establishing regular use, is a difficult process (Mcuter et al. 2005). Thus, if firms expend great effort to bring consumers to the point of trial, it seems important that firms then act to encourage a trial environment that is conducive to positive evaluation. Our research demonstrates how consumers' emotions play an influential role in early usage environments. We show that a significant proportion of these emotions results from the disconfirmation of consumer expectations about ease of use, implying that a firm may either predict or influence the consumer's emotional experience. Perhaps the most significant contribution of this research is the demonstration that emotions experienced in early use of the new product are not simply reflective of achieved benefits (or risks) but result directly from leaming efforts. These emotions, both immediately

and over time, significantly influence product evaluations, even after we account for the traditional diffusion indicators.

The E3 Model of Emotional Influence

We present a five-step conceptual model that describes how emotion intluences the evaluation of complex innovations. This model fits well within the traditional models of diffusion but is unique in its focus on the important how-to period when consumers first leam to use a new product. The five-step E-^ model appears in Figure I. The core steps that describe consumers' experience of early u.se and the theoretical process by which emotions may influence diffusion (Steps 2, 3, and 4) are foundational to our theoretical contribution. The model also addresses how firms can either predict or influence the process through relevant and actionable antecedents (Step 1) and whether relatively ephemeral emotions can affect more lasting evaluations (Step 5).

Steps 1 and 2: Complexity Expectations and Ttieir (Dis)confirmation

In general, expectations matter. In marketing, the power of expectations is demonstrated in satisfaction research (e.g., Oliver and Swan 1989); thus, we consider the types of expectations that may be particularly influential in the early stages of new product learning. We posit that complexity expectations (e.g., "How difficult will it be to begin using this?") are critical for predicting both emotions and product evaluations. Defining complexity expectations. We define "complexity expectations" as a priori predictions about ease of use, learning time, and learning difficulty. Theoretically, and akin to gap models of satisfaction, this conceptualization implies that a new product experience may be challenging but not necessarily disappointing if difficulty is expected. This raises two issues that require careful conceptual distinction. First, although the satisfaction literature emphasizes the importance of expectations, these are expectations

FIGURE 1 The E3 Model of the Influence of Complexity Expectations on Innovation Evaluation and Diffusion

Updating Expectations

Step 1: Complexity expectations

Step 2: (Dis)confirmed expectations

H2

Step 3: Experienced emotions H3 Eariy Use of Innovation (Trial)

Step 4: innovation evaluation

H4a. b

Step 5; Usage diffusion

H4C

Awareness

Adoption

Dynamic Innovation Diffusion

Emotion and the Evaiuation and Early Use of Innovations / 45

for the products' promised net benefits (e.g., Cadotte, Woodruff, and Jenkins 1987) and thus are captured in existing diffusion modeis through Rogers*s (1996) long-standing inclusion of nel benefits. This work is not a mere replication of previous satisfaction research. Here, we address consumers' expectations not for the product's ultimate benefits but for its complexity in leaming. ln other words, we focus on what consumers expect during their early usage experience (e.g., the rate at which they can achieve those benefits, the ease of leaming, and their own affective response), and we posii that this is not a trivial construct. Thus, we investigate the potential for complexity expectations to influence product evaluation beyond the strong influence of the expected (and perceived) benefits and costs of the product. As we noted previously, extant diffusion models already successfully predict responses to new products (e.g., Moreau, Lehmann, and Markman 2001; Olshavsky and Spreng 19%), and the demonstration of any additive influence due to consumers' emotional response to initial usage complexity would have broad ramifications for new product research. Second, complexity expectations should be more impactful when products are innovative than when they are new additions to well-understood product categories.^ For some new products (e.g., a new orange juice), there is little uncertainty abtiut the difficulty of use (Phillips and Baumgartner 2002). However, in cases in which the product requires adaptation or leaming (e.g.. Palm Pilots), usage uncertainty abounds (Oliver and Winer 1987). Thus, complexity expectations are likely to be influential only if products are both new and relatively complex. This leads to what is perhaps a marketplace paradox; The emotional experience of initial use is more influential for technological or functional innovations (e.g., a computer program, a Global Positioning System in an automobile) than for simpie experiential or aesthetic products (e.g.. a book, a mu.sic CD, a candle), which are typically associated with (and designed to elicit) an emotional response. However, this paradox is not so counterintuitive in light of the notion that the emotional response we investigate herein is that which arises from leaming, not that which is tied directly to product attributes or benefits. How are complexity expectations formed? For fimis to be able to predict or influence consumers' complexity expectations, they must understand how such expectations are formed. As with attitudes and preferences, we believe that complexity expectations are often constructed rather Ihan retrieved, using both intemal and extemal inputs namely. (1) consumers" relevant prior knowledge. (2) the observable experiences of others, and (3) communications from marketers (Foumier and Mick 1999; Oliver and Winer 1987; Shih and Venkatesh 2004). Although complexity

^II can also be argued Ihat benefit-oriented expectations become more complex with new product categories, and though this topic is still an imponant area for further research, it is addressed in the work of Meyers-Levy and Tybout (1989). Moreau. Lehmann, and Markman (2(X)I). and Hoeffler and Ariely (1999).
46 / Journal of Marketing, July 2006

expectations may arise from several different factors, we focus on two relevant factors: product usage information provided by the retailer (a marketer action) and consumers' prior expertise (a latent consumer characteristic). According to Oliver and Winer (1987), both of these types of information are hypothesized to influence the formation of expectations. Notably, these two factors may also interact so that an expert consumer may be difterentially affected by marketer/ retailer-provided information compared with a novice consumer. Complexity expectations may arise from marketer communications. We have discussed various conceptual ways that expectations can affect product evaluations. In many practical ways, managers can influence complexity expectations, such as when they show a product being used in commercials, describe usage instructions in online formats, or use other high-involvement sales channels (e.g.. sales representatives, infomercials). A tactic that marketers increasingly employ is the in-store product demonstration (RapheJ and Raphel 1994). Product demonstrations reduce the consumer's perceived uncertainty before purchase and have become an important marketing-mix variable in the automobile and computer industries (Heiman and MuIIer 1996) and in business-to-business domains (Gopalakrishna el al. 1995). Product demonstrations can be both benefit oriented (e.g., let consumers gain information about experience attributes) and process oriented (e.g., teach consumers how to set up or use a product feature). Heiman and MulJer (1996) suggest that the benefit-oriented value of demonstrations should be either positive or negative depending on situational characteristics (e.g.. the type of product [or product attribute] and its inherent risk). Although Heiman and MuIIer do not address postpurchase evaluation and satisfaction, they show strong empirical evidence that the efficacy of demonstrations on purchase probabilities (they focus on benefit-oriented demonstrations) varies with consumers' product knowledge. Thus, it is easy to argue that the process-oriented value of demonstrations similarly depends on situational characteristics: in this case, the value of the demonstration depends on whether consumers set high or low expectations for their own difficulty in leaming about and using the product.-^ In line with Heiman and Muller's fmdings, a consumer's expertise may be an important factor in the expectations he or she sets both before and after the demonstration. Thus, we posit that consumers' product category expertise affects both initial complexity expectations and the extent to which these expectations are influenced by product demonstrations. Research in expertise suggests the likely pattem of this influence. Specifically, novices may not be able to imagine what tbey do not know (Kniger and Dunning 1999) and thus may be more positive when there is no demonstration. When a demonstration is provided for an unfamiliar product, the product demonstration may increase both the specificity and the negativity of the novices' expectations (McGill and Iacobucci 1992). Effectively, tbe

thank an anonymous reviewer for articulating this point.

demnnstration may make the technological intricacies required to master the new product highly salient, increasing the novices' expectations (perhaps unrealistically) for difficulty of use. Conversely, expert consumers often approach new product use confidently (Alba and Hutchinson 2000). Because they rely heavily on their own prior experiences, experts often pay little attention to any usage information the firm provides (Wood and Lynch 2002). Formally, we predict the following: H|: Product demonstration (a) increases novice consumers' expectations that the product will be difficult to use but (b) does not affect expert consumers' expectations. We predict that demonstrations will not influence experts' expectations, because experts base their expectations largely on infomiation stored in long-term memory, the ambiguity of which has been discounted (Oliver and Winer 1987). Such discounting diminishes the biasness of the expectation, and thus disconfirmation is less likely to occur Conversely, novices are more likely to hold biased expectations, which can be disconfirmed in one of two ways: (I) Those exposed to a demonstration could experience positive disconfirmation ("better than expected") at first use, and (2) those not exposed to a demonstration could experience negative disconfirmation ("worse than expected"). Formally, H2: (a) Novices who experience a product demonstration report more positive disconfirmation of complexity expectations (better-than-expected ease of use) after first use i)f the innovation than novices who do not experience a demonstration. wherea.s (b) product demonstration does not affect expert consumers' disconfirmation of complexity expectations.

expected progress toward the activated goal is impeded, and positive emotions occur when expected progress toward the activated goal is accelerated or when the goal is attained (Carver and Scheier 1990; Luce, Bettman. and Payne 2001). Although consumption emotions can be driven by actual product performance (Oliver 1993; Phillips and Baumgartner 2002), research also suggests that consumption emotions are a function of disconfirmation of expectations (Westbrook and Oliver 1991). In the introduction of relatively noninnovative new products (e.g., batteries, hotels (Rust et al. 1999; Voss, Parasuraman. and Grewal 19981) or in domains in which consumers have experience (e.g.. current subscribers of an interactive television entertainment service [Bolton and Lemon 19991), consumers may be able to generate accurate usage expectations, and disconfirmation is not likely to occur. Although uncertainty remains about the quality of the consumption experiences in these cases, the consumer has little uncertainty about the difficulty in consuming the product. These situations can be contrasted with those in which consumers are actively learning the new category (e.g., working with new selfservice technologies [Meuter et ai 20051. participating in coproduction |Bendapudi and Leone 2003]) in which real gaps are likely to exist between consumers' expectations of the ease/difficulty of goal attainment and their actual experience. Appraisal of progress, especially given unexpected difficulty or ease, typically generates an emotional response (Smith and Ellsworth 1985). Thus, we propose the following: H3: Positive disconfirmation of complexity expectations is correlated with positive emotion during use. and negative disconfirmation is associated with negative emotion during use.

Step 3: The Influence of Disconfirmation on Experienced Emotion

However, it would be hasty (and possibly ill-advised) to assume that support for H] and Hi suggests that in-store product demonstrations are a bad idea, having no influence on experts and a negative influence on novices. Because of the relationship between disconfirmation of expectations and emotion, the potential negative intluence of product demonstrations on novices may ultimately lead to more positive results. By depressing expectations of an easy experience, novices who receive a demonstration may subsequently be positively surprised by the actual usage experience, seeing it as "not so bad after all," a situation that can create positive etnotions. The E-^ model suggests that consumers who experience different disconfirmation of their usage expectations have different emotional responses. Why do consumers experience emotions in using complex products? V^hen people are confronted with a learning task, they develop goals and then monitor progress toward those goals. Emotions are mechanisms that communicate important infomiation relative to expected progress toward goal achievement (Bagozzi, Gopinath, and Nyer 1999; Luce. Bettman. and Payne 2001); as such, they are likely to occur spontaneously when consumers first use a new. complex product. Specifically, negative emotions occur when

Step 4: Influence of Emotion on the Evaluation of innovative Products

Can emotions affect product evaluations? It could be reasonably argued that emotions that arise from "temporary" learning or trial periods are unlikely to intluence more long-lasting product evaluations. However, recent research in emotion shows that affective influence is often stronger and more far-reaching than previously considered (e.g., Bagozzi, Gopinath, and Nyer 1999). Several recent articles have shown the significant link between consumption emotions and satisfaction (Oliver 1993: Phillips and Baumgartner 2002); most empirical findings indicate that both positive and negative emotions influence satisfaction in their respective directions (Schwarz and Clore 1983). Yet demonstrations of emotional influence on product evaluation have been mixed. In a longitudinal study. Dube and Morgan (1996) demonstrate that retrospective measures of emotion do not influence productor, in their case, service evaluation. Although we posit that the addition of emotional responses benefits traditional models of diffusion, the opposite could also be suggested. A reasonable argument may be that emotions result from the adoption process but that their effect on evaluations Is negligible compared with traditional models that capture both net benefits and product complex-

Emotion and the Evaluation and Early Use of Innovations / 47

ity. As we noted, net-benefit models have a strong track record of effectively predicting innovation evaluations, sometimes accounting for more than 50% of explained evaluative variance (e.g., Moreau. Lehmann, and Markman 2001). We hypothesize that because emotions result from disconflmiaiion of usage expectations, they act as important signals for the consumer regarding his or her ovi'n experience. This hypothesi.s is akin to other examples of "affect as information" in which evaluations of products are influenced by situational mood states (Pham 1998). This emotional influence may operate in addition to other sources of emotion, such as the utility derived from experienced benefits. Because usage expectations are different from consumers' expectations about the product's ultimate benefits, focusing instead on early usage experience (e.g., the rate at which they can achieve those benefits, the ease of learning, and their own affective response), the emotions derived from usage expectations may be distinct from emotions experienced from gaining product benefits. The changing influence of emotions over time. It is important to consider the potentially dynamic influence of emotion at distinct periods in the adoption process. Whereas the dominant satisfaction paradigm assumes that consumers use static standards to make satisfaction judgments. Foumier and Mick (1999) find strong evidence that standards are often malleable. In the early stages of product learning, we also expect to observe dynamic changes in usage difficulty expectations as consumers gain direct experience witb the new product and update their priors accordingly, in particular, Inexperienced consumers, who may have been surprised by the difficulty or ease of their initial usage experiences, are likely to use their newly acquired direct experience to adjust their expectations for their next use. Although the adjustment may not be accurate (e.g., these consumers may either over- or underadjust), basic tenets of learning hold that over time and experience, usage expectations should stabilize. Over time, we expect emotional influence to be the greatest during the early learning stages of product usage, when the disconfirmation gap hetween expectations and actual experience is at its largest; then, as learning occurs, we expect this gap to shrink, thus decreasing the intensity of the emotions and their correlation with evaluations. Formally,
H4: Both positive and negative emotions (a) are significant predictors ot evaluation at the initial introduction of the new product {before first use), (b) are significant predictors of evaluation at the first use ot the new product, and (c) are not significant predictors of evaluation after repeated use of the product.

Study 1
Methods
Choice of innovation stimuli. Three requirements guided our selection of a new product: (1) It had not been used by the participant population. (2) there was natural variation in prior experience with a previous version, and (3) learning was necessary to achieve basic performance goals. We selected a new personal digital assistant (PDA), the Paltii Zirespecifically, the use of the handwritingrecognition software, The Palm Zire offers both a calendar function and an address function, and to use each of these functions, users must learn the proprietary Graffiti handwriting-recognition software."^ Learning to use the Graffiti software was chosen as a task that was btith complex and fundamental to the operation of the new product and that could be viably learned and tested within the parameters of the laboratory setting. Design and procedure. The study was a 2 (product demonstration: no versus yes) x 2 (prior PDA experience: no versus yes) hetween-subjects design. Participants were 175 undergraduate students at a large midwestem university who participated in the hour-long study for course credit. After a brief overview of PDAs and the two required tasks, each participant was given the experimental materials that detailed how to use the Graffiti characters. Next, participants in the randomly assigned "demonstration" condition received the demonstration provided by the same experimenter using the same script.-^ All participants tben read additional details about the Palm Zire and answered the preusage dependent measures. Next, the experimenter delivered a Zire to each workstation. Participants were given step-by-step instructions and asked to complete the first task of entering an appointment. Participants showed completed work to the experimenter and answered the next set of dependent measures. After tbis, the participants proceeded to the second task, namely, entering an address. Participants then answered a final set of dependent measures; before leaving, they were thanked and debriefed.

Independent and Dependent Measures

Prior catesoty experience. Participants indicated their experience level with PDAs on a seven-point scale. Approximately half indicated no experience or a very low level (i.e., a 1); thus, we used a median .split to classify participants as inexperienced versus experienced.

Shih and Venkatesh (2004) call specifically for experimental and longitudinal work to address usage diffusion elements. In Study I, we use a controlled laboratory setting to test our hypotheses. In Study 2, we use a quasi experiment over a three-month period to explore these hypotheses further in a real-world setting.

'There i.s an altemate way to enter information using a simulated keyboard, but the experimental instructions required participants to use the Graffiti software. ^In the demonstration, the experimenier held up the Zire. showed the participants where the stylus was located, and indicated how to tap the main menu buttons to find the appropriate tasks. The experimenter then put a transparency of the sheet detailing the Graffiti characters on an overhead projector and briefly demonstrated how and where to draw the letters on the screen of the Zire.

48 / Journal of Marketing, July 2006

Product-related expectation.^. With the exception of questions on prior PDA usage and demographics, we employed the same set of dependent measures at each of the three points of interest during the experiment. To allow participants to communicate any type of product-related expectation, we employed an open-ended written protocol both before and after product trial ("As you anticipate using the Palm Zire PDA to schedule an appointment, please describe the expectations that you have about your first experience using the Palm Zire PDA"). An examination of the responses indicated that almost all contained information on the anticipation of and reactions to the Zire's ease/difficulty of use. Two coders, who were blind to the condition of the participants, coded the written responses on two sevenpoint scales that captured expectations of difficulty in learning and using. After each subsequent use, the coder rated the written protocols on two additional scales that captured how the experience compared with expectations (much more difficult to learn, much more difficult to use). The coders" reliability was high (correlation on all scales >.74); thus, we averaged their responses. Furthermore, the correlation between each pair of items was also high (each pair >.88): thus, we averaged each pair of items to form a measure of usage expectations for Time I and a measure of disconfirmation of usage expectations for Times 2 and 3. We combined the coded measures for Time I with the traditional complexity measures that Rogers (1996) used (we describe this further in the next section) to form an overall expectation score. Complexity expectations. Following Rogers's (1996) definition of complexity, participants indicated on three seven-point scales how difficult they expected/perceived the Zire to be to use, how long it would take to leam to use, and how much of a challenge it would be to use (at each period, a > .90); higher numbers indicated greater expected difficulty. At Time 1 (preusage), we combined these three selfreport measures with the two items derived from the written protocols to create an overall index of expected usage difficulty (a = .92).

Disconfirmation of complexity expectations. A standard method for assessing disconfirmation is to measure participants' retrospective assessments of whether the product or service was better or worse than expected (e.g., Oliver 1993; Phillips and Baumgartner 2(X)2). Effectively, our coding of the written protocols at Times 2 and 3 captures a similar appraisal of this comparison in a more open-ended manner. In addition, we subtracted usage expectations at time t from usage expectations at time t + I. We combined this difference with the two coded items. Emotions. To assess participants' emotions at each stage in the adoption process, we employed a modified version of Izard's (1977) differential emotions scale (DES-II); we balanced the scale's positive and negative items as Richins (1997) suggests. Similar to previous consumption emotion research (e.g., Phillips and Baumgartner 2O()2), we found a two-factor solution at each point in time using the modified scale, with the two factors representing the positive and negative emotions. We combined the seven positive and seven negative items to create a positive and a negative emotion variable for each point in time (see Figure 2). Product evaluations. We used a five-item scale with seven-point items to assess participants" evaluations of the Zire at each of the three points in time ("As an overall evaluation of this particular PDA, 1 think I will "dislike it very much/like it very much""; "I think this particular PDA and its features are good"': "strongly disagree/strongly agree"; "I am very pleased that I will have to try this particular PDA in this study"; "strongly disagree/strongly agree"; "I would like to use this PDA on a regular basis": "strongly disagree/strongly agree"; "Would you recommend this PDA to your friends?": "definitely would not/definitely would"). At each point in time, all five items loaded on a single factor, and thus we summed the five items to form a single evaluation measure for each point in time (all three a > .89). Covariates. We included compatibility and net benefits as covariates that have been shown to influence new product success. In line with the work of Rogers (1996), we

FIGURE 2 Emotion Response Scale (Modified from Izard's DES II Checklist)

The DES Emotions Checklist Record the emotions you are experiencing now by circling the response that best fits how you truly feel. Not at all happy Not at all frustrated Not at all sad Not af all joyful Not at all angry Not at all confused Not at all motivated Not at all determined Not at all excited Not at all scared Not at all bored Not at all irritated Not accomplished Not at all proud 1 1 1 i 1 1 1 1 1 1 1 1 1 1

2 Z
2 2

2
2

' 2
2 2 2 2

2-1 . &

3 3 3 3 3 3 3 3 3 3 3 3 3 3

4 4 4 4 4 4 4 4 4 4 4 4 4 4

5 5 5 5 5 5 5 5 5 5 5 5 5 5

6 6 6 6 6 6 6 6 6 6 6 6 6 6

7 7 7 7 7 7 7 7 7 7 7 7 7 7

Very happy Very frustrated Very sad Very joyful Very angry Very confused Very motivated Very determined Very excited Very scared Very bored Very irritated Very accomplished Very proud

Emotion and the Evaluation and Early Use of Innovations / 49

measured compatibility with three summed scale items to indicate how significantly participants would need to change the way they currently schedule appointments, keep track of addresses, and manage their information (at each period, a > .74). Two scale items measured net benefits of the Zire (the perceived functional benefits less associated risks), and two other scale items measured participants' evaluations of the Zire's appointment and address functions. The four items loaded on the same factor (a > .78), and we summed them at each of the three points in time.

Results
Tiie factors influencing initial complexity expectations. We used a two-way analysis of variance (ANOVA). with experience and demonstration as the independent factors, to test H|. The results indicate both a main effect of experience (F(l, 172)= 19.62./7<.(X)01) and the predicted interaction between the two factors (F(l, 172) = 5.21. p < .02). In support of H]y, inexperienced participants who received the demonstration had significantly higher expectations of difficulty than those who did not (F(l, 172) = 3.96. p < .05; contrast; M^emo = 20.9 versus M,,^,^^.^,,, = 18.5). However, the demonstration did not significantly influence those with prior experience (F(l, 172) = 1.83, p > .10; M^^^,,^ = 14.8 versus Mn,,t|p,T,,)= 16.6), in support of Hn,. The disconfirmation of complexity expectations at first use. The negative expectations ibat the product demonstration created for inexperienced consumers may not be ultimately detrimental to product evaluations. Witb the measure of disconfirmation as the dependent variable, we used a two-way ANOVA to test Hi. which revealed the predicted interaction (F(L 172)= 14.2'3,/;< .001; see Figure 3, Panel A). Inexperienced participants who received tbe demonstration experienced tbe least negative disconfirmation of expectations (i.e., tbe most positive disconfimiation. finding the Zire easier to use tban expected) conipaied with their counterparts who did not receive the demonstration (F(l. 172) = 22.53,/?<.OOO1: contrast: Mnojc,m, = 4.5 ver'^"s M^teino = -2.6, p < .0001), in support of H^y. Again, the demonstration did not significantly influence tbe degree of disconfirmation for those with prior experience (F(l. 172) = 1.09. p> AO; Mdemo = 4.2 versus Mnoj^n,,, = 2.3), in support of HjbImpact of disconfirmation on experienced emotion and evaluation at first use. If disconfirmation influences emotion (and subsequently product evaluation), we would expect to observe an experience x demonstration interaction for emotion that mirrors the pattern tested in H2. The predicted interaction is significant (F(l. 170) = 4.90, p = .02; Figure 3, Panel B). Experienced participants" emotions were not influenced by the product demonstration, yet inexperienced piirticipants were strongly influenced by the demonstration; those who had a demonstration reported more positive emotion (Mjj.n,,, = 32) than those who did not have a demonstration lM|,,,(Semo= 27). Another way to test H3 is through regression analysis by examining the comparative influence of complexity expectations on positive and negative emotion at all three periods (before use, at first use, and after repeated use). At each

point in time, we ran two regressions with positive and negative emotions as tbe dependent variables. Predictors included tbe two independent factors (experience and demonstration), their interaction, the measure of usage expectations, and the covariates (compatibility and net benefits). At Times 2 and 3, we also included as predictors tbe disconfirmation of usage expectations. Perceptions of tbe Zire's usage difficulty and its net benefits significantly predicted emotions at each point in time. As Table 1 shows, complexity is a significant driver of emotion at Time 1 (B[x,.i,ive = -.31, p < .001; Bn,g,ji,, = .20. p < .05). Before use, net benefits influence only positive emotion (B = .18, /J < .01), not negative emotion. Importantly, this pattern also holds at Time 2 and Time 3, augmenting tbe support for H3. Similar to H3, we test H4a_e with regressions at each period. For each period, we ran a regression using evaluations of the Zire as the dependent measure. Predictors included the two independent factors, their interaction, tbe measures of usage expectations, the covariates (compatibility and net benefits), and positive and negative emotions. At Times 2 and 3. we also included as predictors tbe disconfirmation of complexity expectations. Before first use, positive emotion was a significant driver of evaluation (B = .41. p < .001), as were net benefits (B = .27, p < .001; see Table 2). However, negative emotions were not significant predictors of evaluation (B = -.10, /? > .05). Tbis result is notable, and we discuss it furtber in Study 2. After firsl use, both positive (B = .33, /? < .001) and negative (B = -.22, p < .01) emotions were significant drivers of product evaluations, as were net benefits (B = .25, p < .001). Finally, after repeated usage, positive (B = .47. p < .001) and negative (B =-.24./j<.001) emotions were still both significant drivers of product evaluation, as were net benefits (B = .25, p< .001). Thus, tbe inlluence of emotion can be observed at all points in time, in support of H4a and H^^, but not H4^..

Discussion
Tbe results from Study I demonstrate that a consumer's expectations about tbe difficulty he or she will experience in using a new product can be influenced by both marketerdirected communications and by his or her own prior experience. The.se usage expectations and their (dis)confirmation were important factors that predicted emotions and drove iheir cbanges over tbe course of tbe usage experience. In turn, emotions predicted product evaluations at all points in time, even after we controlled for tbe other diffusion variables. Note that only positive emotions predicted evaluations toward the Zire before its use, whereas both positive and negative emotions influenced evaluations after use. Before use, any threats to expected goal progress were only hypothetical: perhaps the negative emotions generated before use of the Zire were not substantial enough to influence evaluations. Only when the expected progress was actually impeded did negative emotions influence evaluation. This phenomenon might be conceptualized as a "suspension of judgment." in which the consumer holds doubts or negative feelings at bay in the initial assessment of tbe product. In other words, the consumer may experience negative emo-

50 / Journal of Marketing, July 2006

FIGURE 3 Expectation Disconfirmation, Experienced Emotions, and Product Evaluation After First Use
A: Disconfirmation of Complexity Expectations After First Usea B: Positive Emotion After First Use

5 4
M UJ

o o a

s
c o

3 2 1
0 -1 No experien -2.6 Consumer Expertise - - - - - - N o demo Demo No Experience Experience Experience

-2 -3 -4

Consumer Expertise - - - - - - - N o demo Demo

C: Product Evaluation After First Use

Q,

No Experience Experience Consumer Expertise - - - - - - - N o demo Demo

^Larger numbers indicate greater negative disconfirmation (worse-than-expected usage complexity).

tions when considering the leaming/usage difficulties but may not incorporate these initial emotions in evaluating the product. This is a notable and somewhat counterintuitive finding, given that consumers pay disproportionately more attention to negative attribute performance in postuse product evaluations (Mittal, Ross, and Baldasare 1998).

Innovation Continuity and Complexity Expectations

In Study 1, the complexity expectations of the inexperienced consumers were the most influenced by tbe product

demonstration. As such, these consumers were subject to the greatest amount of disconfirmation and, consequently, produced greater emotional reactions to the Zire. In this population, inexperienced participants had no experience with the PDA category. Discontinuous innovations have been defined as those thai do not fit neally into an existing category (Lehmann 1994), and for consumers with no prior PDA experience, there is likely to be no substantial PDA category representation existing in memory. For consumers with PDA experience, however, such a representation exists. As such, the Zire could be considered a discontinuous innovation for those without PDA experience but a con-

Emotion and the Evaluatioti and Early Use of Innovations / 51

TABLE 1 Study 1: Predictors of Emotions (Standardized Estimates)

Emotions Before Use Predictors Prior PDA experience Product demonstration Experience x demo Product-Related Expectations Net benefits Compatibility Usage (difficulty) Disconfirmation of usage expectations
R2

After First Use Positive .08 .01 -.04 .37"* -.11 -.52'" .04 .33 Negative -.10 .01 -.04

After Second Use Positive Negative -.02 -.07

Positive .03 .10 -.07

Negative -.12 -.04 .01 .10 -.11 .20' .12

.07 .18* -^.t2 .16* -.01 .42'** -.17* ' .31

.06 -.04 -.04 .68"' .07 .54

.18" -.08 -.31*" .10

-.08 -.04 .88*" .19* .59

' < .05.

TABLE 2 Studies 1 and 2: Predictors of Evafuations of the innovation (Standardized Estimates)

Evaiuations Before Use Predictors Prior experience Product demonstration Experience x demo Product-Reiated Expectations Net benefits Compatibility Usage (difficulty) Disconfirmation of usage expectations Consumption Emotions Positive Negative
R2

After First Use Study 1 J&l -.01 .03 .25"* .04 -.23* -.16' .33*" -.22" .65 Study 2 .13 .21 -.28 .13 .07 -.51*" .36* .35"* -.24* .51

After Second Use Study 1 -.07 -.01 Study 2 -.01 .01 -.04 .38*** .18 -.21* -.03 .22* -.11 .46

Study 1 -.06 .04 -.04 .27*" .08 -.16-

Study 2 .13 .01 -.04 .36"* .02 -.18*

.25"* -.01 -.26*** .05 .47"* -.24".71

.41*** -.10 .33

.35*" -.08 .52

'p < .05. "p< .01. p< .001.

tinuous innovation for those with it. Because the literature has documented the effect of continuity on responses to new products (e.g., Moreau, Lehmann. and Markman 2001; Rogers 1996), we designed Study 2 to control for the continuity ol" the innovation, white assessing differences in product experience. With a discontinuous innovation, the magnitude of the gap between expected and actual goal progress is likely to be significant because the consumer has no direct category experience on which to base expectations (Oliver and Winer 1987). As a result, the emotions that ensue are likely to be more powertui and predictive of overall evaluations. However, with a continuous innovation, the size of the gap is

likely to be smaller. In these situations, ali consumers are likely to have some direct experience with existing products in the category, bui the level may vary across consumers. Although their existing knowledge may not be perfectly calibrated, it is likely to provide a more accurate prediction of the actual usage experience than if it were unavailable. In Study 2, we test this boundary condition to determine whether the E^ model (H1-H4) holds for a continuous innovation. Study 1 also involved the typical level of artificiality that is characteristic of lab studies, and its controlled nature dictated an accelerated usage schedule. As such, participants had no discretion over the timing of the use of the

52 / Journal of Marketing, July 2006

product. In both academic (e.g.. Rogers 1996) and practitioner-oriented (e.g.. The Economist 2004) literature, there is evidence that the expected usage difficulty of a product causes consumers to delay a purchase. Thus, we might expect differential complexity expectations to affect usage timing even in situations in which the consumer does not make a purchase decision, such as with a forced adoption. Formally. H5: Expectations of complexity are positively correlated with the delay in initial usage. In Study 2, we use a quasi-experimental methodology over an extended period to test H]-H5 in the context of a continuous innovation. Whereas we predict that complexity expectations are less influential for a continuous innovation, we expect (and demonstrate) sitnilarities between Study 1 and Study 2 in the factors that influence overall evaluations. As we observe, the pattern of emotions predicting evaluations of the discontinuous innovation at all three times is almost identical to the pattern predicting evaluations of the continuous innovation.

introduction and the instructional handout for this task. This operationalization of the demonstration differed from Study 1 in that it used a different medium (computer), used a different instructional material for the Web site's use. and was conducted in a group rather than individually. Before use (Time 1). On the first day of class, immediately following the introduction, students reported the emotions they experienced while the professor was introducing PageOut. Students then completed demographic and experience measures, followed by the dependent measures. Finally, students were given an assignment to log on to PageOut and register by the next class. A small extra credit incentive was provided. The administrators' software enabled tbe professor to record students' log-on and registration times. Af^er the first use (Time 2). On the second day of class, students reported the emotions they felt while logging on and registering witb PageOut and responded to the dependent measures. After extended usage (Time 3). Over the course of the semester, students used PageOut to take online quizzes, check exam grades, download class notes, and communicate witb the professor. At a mandatory class meeting shortly before the end of the semester, students reported their current emotions and completed the final set of dependent measures.

Study 2
Method
As in Study 1, several criteria guided our selection of stimuli: (1) The product had to be a continuous innovation for all participants, with natural variation in usage experience among the set, and (2) the product had to require some learning for all participants to achieve its basic performance goals. We selected PageOut, a new course management software program from McGraw-Hill. A pretest confirmed that there was variance in participants' experience in using online course management tools but that all potential (student) participants were highly familiar with different types of software, with Web page interfaces, and with computers. As a cover story, students were told that they had been volunteered to beta-test the software and would be asked to assess PageOut three times throughout the semester.

Dependent Variables
Expectations, emotions, and evaluations. Measures of product-related expectations, emotions, and evaluations were similar to those used in Study 1, witb one exception. In an effort to minimize the time required to collect the data, no open-ended protocols for usage expectations were employed. Log-on time. As we noted previously, the administrators' software enabled the professor to track the time that elapsed from the time each student left class to the time that he or sbe registered on the Web site. Prior e.xperience. Experience witb online course management tools was measured using two items. A median split identified students as experienced or inexperienced.

Design and Protocol

Undergraduates who were enrolled in three sections of an introductory marketing course participated in this 2 (high versus low experience) x 2 (demonstration versus no demonstration) between-subjects study throughout the semester as part of their curriculum. Although 123 students were enrolled in the three sections, 17 failed to complete all aspects of the study and were subsequently deleted from further analyses. The students were not randomly assigned to a section, but the course prerequisites, the professor, the class format, and the course content were held constant across all three sections. On the first day of class, the professor introduced PageOut to tbe students in all three sections using a consistent script and provided all students witb the same handout containing step-by-step instructions. In one randomly selected section, the professor used the computer podium to log on to the Web site and to demonstrate how to create a student profile. The other two sections received just the verbal

Results
Initial usage expectations and timing of initial use. H5 predicted tbat higher expectations of usage difficulty would be associated with delays in product usage, Tbe simple correlation between participants' perceptions of usage difficulty and tbe hours they allowed to elapse before their initial registration suggests that such a mechanism was at work (T = .25, p < .01). The infiuence of experience and demonstration on initial complexity expectations. We tested H| using a two-way ANOVA. As in Study 1, there was a significant main effect of experience (F(K 102) = 7.30, /? < .01). Tbe pattern of data was consistent with the first study; participants with more experience with course management software expected their initial use of the PageOut software to be less

Emotion and the Evaluation and Early Use of Innovations / 53

difficult than did those with les.s experience 4.9 versus Mi^expcriencol = 6.8). The accuracy of complexity expectations for a continuous innovation. After participants logged on and registered, how did their initial usage expectations compare with their actual experience? Given the continuous nature of the innovation, we expected that H2 would be attenuated in this study. Accordingly, in contrast to the findings in Study 1, there were no significant effects of prior experience or demonstration on the degree of disconfirmation of complexity expectations. The means for all four groups were close to zero, indicating that participants were relatively well calibrated in their expectations. Consumption emotions and product evaiuations. As we expected, given participants" greater accuracy in complexity expectations, disconfirmation has a much smaller influence on experienced emotion at each of the three periods. We replicated the experience x demonstration interaction for felt emotion at first use observed in Study I (using standardized measures, B = -.21, p < .05, r = .19). Lessexperienced participants experienced more positive emotion at first use when they received a demonstration (M = .31) than when they did not receive one (M = -.24), whereas more-experienced participants were not affected similarly by the demonstration (Mji^^,,, = -.12 versus Mp,, de^io = .13). However, as we expected with the small degree of" disconfirmation, the experience x demonstration interaction for Time 2 evaluations of the innovation was not significant (F= .bl,p> .2). Tliere are dramatic similarities between tbe Study 1 and the Study 2 findings in the factors tbat influence overall evaluations. Tbe pattern of results that predict evaluations of the Zire across time is almost identical to that of PageOut (Table 2). In botb studies, only positive emotions predict evaiuations before use, which might be interpreted as an initial suspension of judgment during the product introduction such that negative emotions, though they are experienced, are held at bay. However, both positive and negative emotions significantly influenced evaluations after use in both studies, in support of H41,. Finally, H4^ predicted that tbe influence of emotions would wane over time. Unlike Study 1, in which the impact of emotions remained strong during tbe duration of the one-hour session, the impact of emotions declined over the semester for PageOut, in support of H4C. Taken together, these findings suggest that the diminishing influence of emotions can be observed only over longer time horizons. Study 2 demonstrates that the pattern of results from Study 1 holds in a real-world quasi experiment but that emotional influence is not as extreme when the product is less innovative. However, even when emotional influence was smaller as a result of more accurate complexity expectations, the inclusion of emotion still increased the predictive power of the traditional diffusion model.^ Finally, this

Study shows that complexity slows consumers' time to trial (H5).

General Discussion
Theoretical Contribution
The 3 model. By demonstrating how affective respt)nses during new product learning influence innovation evaluations over time, we extend previous work in both satisfaction and diffusion. The E^ model provides a parsimonious description of how emotions are created in tbe early use of complex products and how they may influence product evaluations beyond diffusion's traditional focus on net benefits. These results emphasize the dynamic nature of botb emotion and evaluation in innovation diffusion. Adding emotion to innovation frameworks. Refining models of innovation is easily justified by the importance of new product success to company profitability. Prior research on new product adoption focuses largely on cognitive processes, despite the increasing knowledge of tbe role of emotion in product consumption. We demonstrate tbat emotions resulting from early experience with the product (which therefore might be dismissed as ephemeral) can have a lasting influence on diffusion.^ Exploring and expanding recent innovation research. The use-diffusion model (Shih and Venkatesh 2004) describes four patterns of consumer use of new products based on the rate and variety of consumer use. Shih and Venkatesh {2(KM) theorize that usage diffusion depends on (1) characteristics of the individual or household (e.g., innovativeness, resources, prior experience). (2) characteristics of the innovation (e.g., complexity, technological sophistication, existence of complementary products), and (3) characteristics of the environment (e.g.. communication intensity, technological access, media exposure). The current research expands the knowledge of how these factors may interact by examining how the characteristics of the environment (how the innovation is promoted to the consumer), the product (complexity), and the consumer (prior expertise) combine to create a successful or unsuccessful initial product experience. In addition, the importance of ttie phenomena we report herein is reflected in recent work on feature fatigue. Thompson, Hamilton, and Rust (2005) show tbat consumers assign more weight to product capabilities (addi-

'To assess this contribution, we compared the explanatory power of the model shown in Table 3, Panel B. with that of a reduced model that does not include the two emotion variables (R2 = .38). The resulting F statistic (F(2,95) = 9.13, p < .01)
54 / Journal of Marketing, July 2006

demonstrates that the addition of emoti(tns to the standard model of preference makes a significant contribution. ^Researchers have long advocated the addition of affective elements to primarily utilitarian-based models of product evaluation (Derbaix and Pham 1991). However, many researchers view emotion as another viable end-state benefit similar to utilitarian benefits. For example. Havlena and Holbrook (1986) note that consumers might buy a luxury car lor the pleasure of feeling superior. We do not investigate emotions as independent benefits in ihi.s work but rather as a moderator of product learning and acceptance. Although emotions can color utilitarian attribute evaluation (Smith and Ellsworth 1985), they can also have a more global influence on product experiences (Cohen and Areni 1991).

tional features) in their evaluatiotis before use than after use and le.ss weight to product usability before use than after use. Thus, they show that consutners are often miscalibrated in their expectations about product complexity and usability before use. Our research shows that inaccurate and overconfident expectations about usability (e.g., when consumers seek excessive features in prospect) may create situations in which negative etnotions have a powerful influence on product evaluations and, thus, in retrospect, satisfaction.

Managerial implications
Our results show that the predictive power of traditional models of innovation adoption can be improved by the addition of emotional responses. Importantly, a substantial proportion of consumers' emotional responses appear to be a function of disconfirmed complexity expectations, not simply a result of product benefits. This tells marketers two things. First, emotional influence can be assessed by measuring disconfirmation of usage expectations, and second, emotional responses can be influenced by changing consumers" complexity expectations before use. As we noted previously, marketers can influence complexity expectations when they show product use in commercials, describe usage instructions in online formats, or use other highinvolvement sales channels (e.g., sales reps). We tested product demonstration a.s one way to influence expectations. Product demonstrations are believed to be beneficial to both marketers and consumers but expensive to execute, and marketers do not have a clear idea of how they truly affect consumers (Ailloni-Charas 2000). Our research suggests that training frontline saies personnel is important; the salesperson may be tempted to insist that a technological product is easy to leam to close the sale, yet the mismanagement of usage expectations can lead to high rates of product return in the long run. Promising "easy'* use may promote trial but sabotage product evaluation and diffusion. Thus, marketers must seek a balance in which consumers are encouraged to try new products but not through promises of ease that create unrealistic expectations for early use. For example, we can compare the recent strategic actions of Royal Philips and Best Buy to observe two opposite outcomes of influencing consumers' compiexity expectations. Best Buy's Geek Squad (currently in 642 stores) is a team of professionals, attired in stereotypically nerdy uniforms (black polyester pants, white shirts, and clip-on ties), who make house visits to help consumers learn complicated product tasks, such as how to program an iPod or connect a TiVo system (Deam 2004). Rather than downplaying the complexity of products sold in the store. Best Buy leverages these perceptions by offering (for a fee) a way to shift the learning costs to a team of prototypically smart people. Not only does this generate revenue, but our research would predict that the Geek Squad also sends a signal, albeit humor-

ous, to consumers that product learning is not always easy: with such a signal, consumers tnay subsequently experience little negative disconfirmation in early usage and be positively surprised if they successfully achieve product use on their own. Conversely, Philips's strategy of promoting itself as a provider of user-friendly gadgets (The Economist 2004) may be less optimal for the company. By launching a $100 million ad campaign to promote its technology as "easy to use" (Endt 2004). Philips may lower consutners' preuse complexity expectations and subsequently strengthen the negative emotions that occur in the early learning phase of use. Most learning requires .some setbacks, and reducing consumers' expectations of setbacks may have a negative impact on long-term product and brand evaluations.

Limitations and Further Research

A limitation of the current research is our use of forced adoption in the two studies. Participants in both studies did not choose to adopt the new products, though they had an outlet to report their opinions and evaluations at multiple points in the usage period. Although this situation was chosen for practical reasons, on reflection, we question whether forced adoption has a positive side in offering a better and more conservative test of the influence of complexity expectations and emotion. When consumers choose to try/ buy a product innovation, there are certain emotional and evaluative processes that arise solely from the evaluation of the choice, such as cognitive dissonance. Dissonance may amplify emotions that arise from disconfirmation of usage expectations, Thus, emotional responses may be heightened in voluntary adoption, creating even stronger influence of emotions on innovation evaluations. There are many other diverse avenues for further research suggested by our results. First, the E-^ model advocates the inclusion of both cotnplexity expectations and emotions (positive and negative) in existing frameworks of evaluation, satisfaction, and diffusion. Second, these rcstilts, combined with work on regret (e.g., Tsiros and Mittal 200), begin to build a more complete picture of negative emotion after a choice is made. Luce, Bettman, and Payne (2001) show the importance of negative emotion as consumers contemplate choice. The act of making a choice is not the end of emotional influence. Consumer expectations for the usage or learning process may have a surprisingly strong impact, influencing regret over product choice dissatisfaction not only with product benefits but also with the costs to achieve those benefits in a timely manner. Can complex products make consumers feel smart because of unexpected hurdles overcome or stupid because of unexpected slowness in achievement? More important, can such emotional reactions influence the evaluation of the innovation beyond more concrete evaluative criteria? The current research suggests that they can.

Emotion and the Evaluation and Early Use ot Innovations / 55

REFERENCES
Ailloni-Charas, Dan (2(X)0), "Demos, Yes, But Don't Give Away the Store." Brandweek. (March 13), 30. Albu, Joseph and J. Wesley Hutchinson (2()()0j, "Knowledge Calibration: What Consumers Know and What They Think They Know." Journal of Con.sumer Research. 27 (September), 123-56. Bagozzi, Richard P.. Mahesh Gopinath, and Prashanth U. Nyer (1999), The Role of Emotions in Marketing." Journal of the Academy of Marketing Science, 27 (Spring), 184-206. Bendapudi, Neeli and Robert Leone (2(X)3), "P.sychological Implications of Customer Participation in Co-Production." Journal of Marketing. 67 (January). 14-29. Bolton. Ruth ami Katherine Lemon (1999), "A Dynamic Model of Customers' Usage of Services: Usage as an Antecedent and Cotisequence of Satisfaction." Journal of Marketing Research, 36 (May), 171-87. Cadotte. Emest. Robert Woodruff, and Roger Jenkins (1987). "Expectations and Norms in Models of Consumer Satisfaction." Journal of Markeririfi Research. 24 (August). 305-314. Carver, Charles S. and Michael E. Scheier (1990), "Origins and Functions of Positive and Negative Affect: A Control-Process View." Psychological Review, 97 (January), 19-35. Cohen. Joel B, and Charles S. Areni (199!). '"Affect and Consumer Behavior," in Handbook of Consumer Behavior, Thomas S. Robertson and Harold H. Kassarjian. eds. Englewood Cliffs, NJ: Prentice Hall, 188-240. Deam. Jenny (2004). "Who's Coi:il? It's All Geek to Gadget Users." The Denver Po.st, (November 11), DI. Derbaix, Christian and Michel T. Pham (1991). "Affective Reactions to Consumption Situations: A Pilot Investigation," Journal of Economic Psychology. 12 (June). 325-55. Dube. Laurette and Michael S. Morgan (1996), "Trend Effects and Gender Differences in Retrospective Judgments of Consumption Emotions," Journal of Consumer Research, 23 (September). 156-62. The Economist (2004), "Simplifying Philips," (June 12), 66. Endt, Friso (2(K)4). "It Is Rocket Science," Newsweek. (October 18), E8. Eoumier. Susan and David Mick (1999). "Rediscovering Satisfaction," Jounml of Marketing, 63 (October). 5-24. Golder, Peter N. and Gerard J. Tellis (1998), "Beyond DitTusion: An Affordability Model of the Growth of New Consumer Durables," Journal of Foreca.stinf-. 17 (3-4). 259-80. Gopalakrishna. Srinath. Gary L. Lilien, Jerome D. Williams, and Ian K. Sequeira (1995), "Do Trade Shows Pay Off.'" Journal of Marketing, 59 (July), 75-83. Havlena. W.J. and Monis B. Holbrook (1986), "The Varieties of Consumption Experience: Comparing Two Typologies of Emotion in Consumer Behavior." Journal of Con.sumer Re.'ieanh, 13 (December). 394^04. Heiman. Amir and Eitan Muller (1996), "Using Demonstration to Increase New Product Acceptance: Controlling Demonstration Time," Journal of Marketing Research, 33 (November), 422-31. Hoeffler, Steve and Dan Ariely (1999), "Constructing Stable Preferences: A Look into Dimensions of Experience and Their Impact on Preference Stability," Journal of Consumer PsycholLuce, Mary Frances, James R. Benman. and John W. Payne (2(K)1). "Emotional Decisions: Tradeoff Difficulty and Coping in Consumer Choice," in Monographs of the Journal of Consumer Research. No. I. Deborah Roedder John. ed. Chicago: University of Chicago Press. McGill, Ann and Dawn lacobucci (1992), "The Role of PostExperience Comparison Standards in the Evaluation of Unfamiliar Services," in Advances in Consumer Research: Diversity in Consumer Behavior. Vol. 19. John Sherry and Brian Sternthai, eds. Provo. UT: Association for Consumer Research, 570-78. Meuter, Matthew L., Mary Jo Bitner, Amy L. Ostroni. and Stephen Brown (2005), "Choosing Among Alternative Service Delivery Modes: An Investigation of Customer Trial of Sclf-Service Technologies." Journal of Marketing, 69 (April), 61-83. Meyers-Levy, Joan and Alice M. TVl^""! (1989). "Schema Congruity as a Basis for Prtniuct Evaluation," Journal of Consumer Research, 16 (June), 39-55. Mick. David Glen and Susan Foumier (1998), "Paradoxes of Technology: Consumer Cognizance, Emotions, and Coping Strategies." Jounuil of Consumer Research. 25 (September). 123-44. Mittal. Vikas. William T. Ross Jr.. and Patrick M. Baldasare (1998). "The Asymmetric Impact of Negative and Positive Attribute-Level Performance on Overall Satisfaction and Repurchase Intentions," Journal of Marketing, 62 (January), 33^7. Moreau. C. Page, Donald R. Lehmann, and Arthur B. Markman (2001), "Entrenched Knowledge Structures and Consumer Response to New Products," Journal of Marketini> Research, 38 (Febinary), 14-29. Oliver, Richard L. (1993), "Cognitive, Affective, and Attribute Bases of the Satisfaction Response." Journal of Consumer Research. 20 (December), 418-30. and John E. Swan (1989). "Equity and Disconfirmation Perceptions as Influences on Merchant and Product Satisfaction." Jounml of Consumer Research. 16 (December), 372-84. and Russell S. Winer (1987), "A Framework for the Formation and Structure of Consumer Expectations: Review & Propositions," Journal of Economic Psychology, 8 (4). 469-99. Olshavsky. Richard W. and Richard A. Spreng (1996), "An Explorator)' Study of the Innovation Evaluation Process," 7ornal of Product Innovation Management. 13 (6). 512-29. Pham. Michel T. (1998). "Representativeness. Relevance, and the Use of Feelings in Decision Making." Journal of Consumer Research. 25 (September), 144^59. Phillips, Diane M. and Hans Baumgartner (2002), "Tbe Role of Con.sumption Emotions in the Satisfaction Response." Journal of Consumer Psychology, 12 (3). 243-53. Raphel, Murray and Neil Raphel (1994). "Sampling Sells," Pmgres.sive Grocer, 73 (January), 17. Richins, Marsha L. (1997). "Measuring Emotions in the Consumption Experience." Journal of Consumer Research, 24 (September), 127^6. Rogers. E.M. (1996), Diffusion of lnnovation.s, 3d ed. New York: The Free Press. Rust. Roland T., J. Jeffrey Inman. Jianmin Jia. and Anthony Zahorik (1999). "What You Don't Know About CustomerPerceived Quality: The Role of Customer Expectation Distributions." Marketing Science. 18 (Winter), 77-95. Schwarz. Norbert and Gerald L. Clore (1983). "Mood. Misattribulion. and Judgments of Well Being: Informative and Directive Functions of Affective States." Journal of Personality and Social Psychology, 45 (3), 513-23. Shih. Chuan-Fong and Alladi Venkatesh (2004). "Beyond Adoption: Development and Application of a Use Diffusion Model," Journal of Marketing, 68 (January), 59-72.

ogy, ^ a). 113-39.

lzard. Carroll E. (1977), Human Emotions. New York: Plenum. Kruger. Justin and David Dunning (1999), "Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Asses.sments." Journal of Personality and Social Psychology, 77 (6), 1121-34. Lehmann, Donald (1994), "Characteristics of 'Really' New Products," paper pt^sented at the Marketing Science Institute Conference, London (June 21-22).

56 / Journal of Marketing, July 2006

Smith, Craig A. and Phoebe C. Ellsworth (1985), "Patterns of Cognitive Appraisal in Emotion," Journal of Personality and Social Psychology, 48 (April). 813-38. Thompson, Debora Viana, Rebecca W. Hamilton, and Roland T. Rust (2005), "Feature Fatigue: When Product Capabilities Become Too Much of a Good Thing." Jounuil of Marketing Research, 42 (November), 431-42. Tsiros, Michael and Vikas Mittal (2000). "Regret: A Model of Us Antecedents and Consequences in Decision Making," Journal of Consumer Research, 26 (March). 401-^18.

Voss, Glenn B., A. Parasuraman, and Dhruv Grewal (1998), "The Role of Price. Performance, and Expectations in Determining Satisfaction in Service Exchanges," Journal of Marketing. 62 (October). 46-62. Westbrook, Robert A. and Richard L. Oliver (1991). "The Dimensionality of Consumption Emotion Patterns and Consumer Satisfaction." Journal of Consumer Research, 18 (June), 84-91. Wood, Stacy L. and John G. Lynch Jr. (2002), "Prior Knowledge and Complacency in New Product Leaming," Journal of Consumer Research, 29 (December), 416-26.

B O O K S

F R O M

E M A R K E T I N G

S C I E N C E

S T I T U T E

Essential Readings in Marketing

[ M by Leigh Mdllister, Oniveisity of Jexos at Austin, Ruth W. Boiton. Atizona State llniveisity, ond Ross Rlzley, Marketing Science InstiMe

NEW

Essential Readings in Marketing

For reodefs seeking to understond itie crirical milestones of marketing riiougfit, this colledion will be on invobable resource. Abstracts of over 200 oword-winning popers are organized by leseorch topic, wilfi cbapter introductions providing context and insigbt. MARKETING SCIENCE INSTITUTE, APRIL 2006,196 PAGES, i30

"An excellent and much-needed source of award-winning research fn marketing for doctaral students, scholars, and business libraries." V . "SEENU" SRINIVASAN, Adams Distinguished Professor of Manogement Stanford University Graduote Scbool of Business

Assessing Marketing Strategy Performance

Chrhfc- Mmrnm. sod Oiaald 1. laknin., UUsn

Assessing Marketing Strategy Performance

Edited by Christine Mmman, Duke Univeisity and Donald R. lehmann, Coiutnbio Oniveisity In 13 chapters, authored by leading marketing scbolors, this book offers a pav^erful set of fools to help researchers investigate how marketing influences fitm performance. MAfiKETlNG SOENCE INSTUUTl, AUGUST 2004,292 PAGES, 135

IVH

To order www.msi.org/links/bkoffer2006.cfm

IVH

Buy Both Books for Only $55

Emotion and the Evaluation and Early Use of Innovations / 57