When does electronic word-of-mouth matter?A study of consumer product reviews ☆
Jason Q.Zhang a ,⁎,Georgiana Craciun a ,1,Dongwoo Shin b ,⁎
a Sellinger School of Business,Loyola University Maryland,Baltimore,MD 21210,USA
b
School of Business Administration,University of Seoul,13Siripdae-gil (Jeonnong-Dong 90),Dongdaemun-Gu,Seoul,130-743,Republic of Korea
a b s t r a c t
a r t i c l e i n f o Article history:
Received 15May 2008
Accepted 19December 2009Keywords:
Consumer product reviews eWOM
Self-regulatory focus Persuasiveness
Online consumer product reviews,a form of electronic word-of-mouth (eWOM),have attracted incread attention from rearchers.This paper examines the persuasiveness of eWOM.Drawing on regulatory focus theory,the authors propo that the consumption goals that consumers associate with the reviewed product moderate the effect of review valence on persuasiveness.Data from lab experiments and actual online retailers suggest that consumers who evaluate products associated with promotion consumption goals perceive positive reviews to be more persuasive than negative ones (i.e.,a positivity bias).Converly,consumers who evaluate products associated with prevention consumption goals perceive negative reviews to be more persuasive than positive ones (i.e.,a negativity bias).
©2009Elvier Inc.All rights rerved.
1.Introduction
In recent years,a growing number of consumers publish product/rvice reviews on the Internet.This new form of electronic word-of-mouth (eWOM)has received incread attention from rearchers.Prior studies examine what leads to eWOM (e.g.,Hennig-Thurau et al.,2004;Ho and Dempy,2010)and how eWOM affects the business bottom line,including product sales (e.g.,Cheval
ier and Mayzlin,2006;Liu,2006),customer value and loyalty (Gruen et al.,2005),and the success of new product introductions (e.g.,Clemons et al.,2006).Yet few studies,with two noted exceptions (i.e.,Park and Lee,2009;Sen and Lerman,2007),focus explicitly on how urs of eWOM evaluate its ufulness.
The objective of this paper is to examine the effects of eWOM valence on eWOM persuasiveness.Drawing on regulatory focus theory (Higgins,1997),which distinguishes between promotion and prevention goals,the authors propo that a contextual variable –the consumption goal that consumers associate with the reviewed product –moderates the effect of review valence on persuasiveness.Speci fically,in evaluating products cloly associated with a promo-
tion consumption goal (e.g.,photo-editing software ud to create ideal pictures),consumers perceive positive reviews to be more persuasive than negative ones.In contrast,in evaluating products cloly associated with a prevention consumption goal (e.g.,anti-virus software ud to avoid a computer crash),consumers perceive negative reviews to be more persuasive than positive ones.Data from both lab experiments (Study 1)and actual online retailers (Study 2)strongly support the propod relationships.
This paper intends to make veral contributions to marketing lite-rature.First,this paper extends WOM rearch to a virtual environment.Traditional WOM literature often relies on social cues (e.g.,relationship to WOM communicator)to explain WOM persuasiveness (Knapp and Daly,2002).Yet,in a virtual environment,the contextual cues are unavailable (Gupta and Harris,2010).The lack of social cues in eWOM forces consumers to evaluate eWOM persuasiveness solely bad on content characteristics (Walther,1996).This rearch adds to the literature by examining eWOM evaluation process in the abnce of social cues.
Second,the findings of this paper clarify,in part,the equivocal findings regarding the effects of WOM valence on persuasiveness.Some studies find that consumers perceive negative messages,in general,to be more persuasive than positive ones (e.g.,Arndt,1967;Laczniak et al.,2001;Mizerski,1982;Yang and Mai,2010).Other studies,however,find rather the opposite (e.g.,Gershoff et al.,2003;Skowronski and Carlston,1987,1989).In light of the mixed findings,rearchers have called for the examination of additional variables (Gershoff et al.,2003).Drawing on regulatory focus theory,this paper introduces a new contextual variable and aims to bridge the gap in prior literature.2.Conceptual framework
Typically,consumers evaluate product information (e.g.,product reviews)in order to help them ful fill t
heir consumption goals.In this
枙子Journal of Business Rearch 63(2010)1336–1341
☆All three authors contributed equally to the paper.The authors thank Gerard Athaide,Rick Bagozzi,Claudiu Dimofte,Rick Klink,Jim Leigh,and George Zinkhan for their comments on earlier drafts of this paper.The authors also gratefully acknowledge the helpful comments made by the editor and anonymous reviewers.The third author acknowledges the support by the University of Seoul 2008rearch fund.
⁎Corresponding authors.J.Q.Zhang is to be contacted at Marketing Department,Loyola University Maryland,4501N Charles Street,Baltimore,MD 21210,USA.Tel.:+14106175837;fax:+14106172117.D.Shin,School of Business Administration,University of Seoul,13Siripdae-gil (Jeonnong-Dong 90),Dongdaemun-Gu,Seoul,130-743,Republic of Korea.Tel.:+82222105745;fax:+82222460570.
E-mail address:jzhang1@loyola.edu (J.Q.Zhang),gcraciun@loyola.edu (G.Craciun),dshin@uos.ac.kr (D.Shin).1
Tel.:+14106175727;fax:+1410617
2117.0148-2963/$–e front matter ©2009Elvier Inc.All rights rerved.doi:
10.1016/j.jbusres.2009.12.011
Contents lists available at ScienceDirect
Journal of Business Rearch
process,lf-regulation is likely to affect how consumers evaluate information.Self-regulation refers to the process through which people t their goals,choo behavioral strategies to achieve the goals,and asss progress toward the goals(Carver and Scheier, 1998).According to regulatory fo
cus theory(Higgins,1997),people strive to achieve their goals through two distinct modes of lf-regulatory system:promotion and prevention.
When people focus on their“ideal goals”(e.g.,aspirations),they develop the promotion system and rely on eagerness behavioral strategies to move clor toward positive end states.In contrast,when people focus on their“ought goals”(e.g.,obligations),they develop the prevention system and rely on vigilance strategies to stay away from negative end states.Although people usually posss both reg-ulatory systems,they tend to have predispositions such that one regulatory system is dominant in directing behaviors.In certain situations,however,people can override their predispositions,and activate a regulatory system that betterfits with the contextual goal. The contextual change of lf-regulatory system providesflexibility in lf-control strategies.In fact,rearch suggests that regulatory focus is more of a contextual motivational state as oppod to a strict motivational trait characterizing an individual's personality(Pham and Avnet,2004).
Prior rearch suggests that consumers tend to compartmentalize products associated with promotion or prevention consumption goals into parate mental categories(Zhou and Pham,2004).Creating p-arate categories,each with its own goals,provides a system that helps consumers allocate ,time,attention)to achieve con-flicting goals.Thus,the compartment
alization makes the control of consumption behaviors more effective(Thaler,1999).For example, with respect tofinancial decisions,Zhou and Pham(2004)suggest that consumers rely on the promotion system to regulate the achieve-ment offinancial gains and the prevention system to regulate the avoidance offinancial loss.Over time,through repeated product usage and/or exposure to product usage information,consumers learn to mentally associate products with distinct promotion or prevention goals. For instance,common stocks are more reprentative of promotion, whereas certificates of deposit are more reprentative of prevention.
In the context of product review evaluation,consumers may acti-vate the regulatory system that is congruent with the consumption goal.More specifically,consumers may develop two parate systems to process information:one that calls on the promotion system to identify uful information for achieving desirable outcomes and the other that calls on the prevention system to identify uful informa-tion for avoiding undesirable outcomes.In effect,the consumption goal that consumers associate with a product operates as a contextual variable to activate consumers'regulatory foci.
Further,the activated regulatory foci affect the way that consumers perceive product related information.Numerous prior studies show that different lf-regulatory systems can generate bias in consumers' ,Aaker and Lee,2001;Chernev,2004;Kim,2006;Lee and Aaker,2000,20
04;Pham and Avnet,2004;Poels and Dewitte,2007; Yeo and Park,2006).For example,Aaker and Lee(2001)find that consumers with promotion(prevention)foci perceive benefit-framed ads as more(less)persuasive than risk-framed ads.Pham and Avnet (2004)reveal that people with promotion foci perceive affective ,feelings experienced during an ad exposure)as more diagnostic than substantive ,the strength of ad claims).
In this rearch,the authors argue that consumers'regulatory foci, activated by a specific consumption goal,motivate the consumers to give different weights to atively valenced messages. Consumers with promotion foci are more concerned with advance-ment and achievement through product consumption.Positive product reviews provide information about satisfactory experiences with the product,and thus reprent opportunities to attain positive outcomes. The reviews are more congruent with consumers'promotion foci and, therefore,are likely to be more persuasive than negative ,a positivity bias).On the other hand,consumers with prevention foci are more concerned with the avoidance of negative outcomes.Negative product reviews provide information about dissatisfactory experiences with the product,and thus reprent opportunities to avoid negative outcomes.The reviews are more congruent with consumers'pre-vention foci and are likely to be more persuasive than positive , a negativity bias).Thus,the authors hypothesize:
H1a.For products associated with promotion consumption goals, positive reviews are more persuasive than negative ones.
H1b.For products associated with prevention consumption goals, negative reviews are more persuasive than positive reviews.邀请函结婚
3.Pretests
Thefirst pretest aims to identify products associated with promo-tion and prevention consumption goals.Twenty-three college students evaluate35different product stimuli and rate each product on two 7-point scales,anchored by:not enhancing/very enhancing and not protecting/very protecting.In the pretest,the authors define products with enhancing characteristics as“products that increa fun in life;the are things you like to have in order to feel good/happy”,and products with protecting characteristics as“products that increa safety in life; the are things you need to have in order to avoid negative con-quences.”The goal is to identify stimuli with high discriminating scores on the promotion/prevention dimensions.The product stimuli lected for the main study are:1)photo-editing software which is associated primarily with promotion goals(M enhancing=5.7,M protecting=2.3, p b.001)and2)anti-virus software which is mainly associated with prevention goals(M protecting=6.4,M enhancing=3.0,p b.001).沾亲带故什么意思
The cond pretest examines the assumption that products asso-ciated with different consumption goals can activate different regu-latory foci.Seventy-five college students evaluate a new software product and then answer a short questionnaire.The authors randomly assign participants to one of two conditions.Half of the participants examine a photo-editing software ,promotion consump-tion goals condition)and the other half examines an anti-virus soft-ware ,prevention consumption goals condition).
The questionnaire includes10items measuring individuals'regu-latory tendencies in a shopping context(adapted from Higgins et al., 1997;Yeo and Park,2006).Five items measure participants'promotion ,“In evaluating this product,I am more concerned about achieving success rather than avoiding failure.”“When I evaluate this product,Ifirst consider what is good about the product.”“When eval-uating this product,I consider achieving positive conquences from using it.”“If I buy this product,I will feel excited about the purcha.”“When evaluating this product,Ifirst consider aspects of this product that I like.”Cronbach's alpha=.79).The otherfive items measure participants'prevention ,“In evaluating this product, I am more concerned about avoiding failure rather than achieving success.”“When I evaluate this product,Ifirst consider what is bad about the product.”“When evaluating this product,I consider preventing negative conq
uences from using it.”“If I buy this product,I will feel safe about the purcha.”“When evaluating this product,Ifirst consider aspects of this product that I dislike.”Cronbach's alpha=.76).The dependent variable is the difference between the average of promotion items and the average of prevention items(Higgins et al.,2000).This variable indicates an individual's overall tendency toward promotion orientation.A one-way ANOVA reveals a significant difference in the regulatory focus of participants who examine the anti-virus software product(M=−0.5)vs.tho who examine the photo-editing soft-ware product(M=0.5;F(1,71)=6.45,p=.01).The results provide evidence that consumers experience momentary states of promotion (prevention)focus after they examine products associated with pro-motion(prevention)consumption goals.
1337
J.Q.Zhang et al./Journal of Business Rearch63(2010)1336–1341
4.Study1
This study tests hypothes H1a and H1b.The authors expect that consumers will perceive positive reviews as more persuasive than negative ones for a product associated with promotion consumption goals.Converly,for a product associated with prevention goals, consumers will percei
ve negative reviews as more persuasive than positive ones.
4.1.Method
The study us a2(consumption goals:promotion vs.prevention)×2(review valence:ative)between-subject designs with a sample size of150undergraduate students.All study ssions take place in computer labs where participants can brow an website that the authors build for this study.文明礼仪顺口溜
Participantsfirst read an online shopping scenario involving the purcha of a software program.The product stimuli are twofictional brands:Digital Studio Professional2007(photo-editing program)in the promotion condition and E-Secure Professional2007(anti-virus program)in the prevention condition.After reading the scenario, participants click on a link to an page and examine the product and its is appropriate for this study for two reasons.First,this website offers a more realistic environment for the experiment compared to afictional website becau most participants are very familiar with this online retailer. Second,as a conquence of the retailer's popularity,participants may examine products displayed more naturally and may perceive related reviews as more credible than if examining the same products and reviews offered by afictional online retailer.
Following the standard format ,the authors u four different Web pages for each of the four conditions.Across the pages,key attributes,including the product price,number of product features,Web page length,and number of graphics and links,remain constant.In each condition,participants e a picture of the product together with the price and shipping information,followed by a short product description,product features,and three reviews:twofiller reviews and the focal review.Each page contains three reviews rather than one in order to simulate a more realistic online shopping envi-ronment.Thefiller reviews rate the product with three stars out of five and include a very broad product ,“This product does what it is suppod to do”).In a pretest,respondents rate the filler reviews for the two products as moderately positive on a7-point scale anchored by negative/positive(M anti-virus=4.5,M photo-editing=4.8, p N.10).
The positive(negative)focal review gives the product a rating of five(one)and describes six positive(negative)product , ea of installation and u,effectiveness at protecting against virus or editing pictures,compatibility with other software).All four focal reviews are similar in length,the number of facts,and the reviewer's lf-rated experti.After examining the product and consumer reviews displayed ,participants complete a short questionnaire in which they evaluate one ostensibly randomly lected , the focal review)on veral dimensions.Finally,participants answer a few demographic questions.
4.1.1.Manipulation check
In order to asss the effectiveness of the review valence manip-ulation,participants rate the focal review on a single item scale from1(“negative”)to7(“positive”).As expected,the results from a2×2ANOVA with review valence and product category as inde-pendent variables show only a significant main effect of review valence, F(1,148)=1265.76,p b.001.That is,participants perceive the positive reviews to be more positive(M anti-virus=6.5,M photo=6.5)than the negative reviews(M anti-virus=1.8,M photo=1.9)across both products.4.1.2.Dependent variable
To measure participants'perception of the review persuasiveness, the authors u four7-point mantic differential scales anchored by persuasive/not persuasive,convincing/unconvincing,important to me/not important to me,and helpful/not helpful(adapted from Burton and Lichtenstein,1988;Mitchell and Olson,1981;Pham and Avnet,2004),and then average the four items to form a composite measure(Cronbach's alpha=.83).Higher scores reflect higher eval-uations of perceived persuasiveness.The correlation matrix appears in Table1.
4.2.Results
The results from a2×2ANOVA support the hypothesized inter-action between review valence and consumption goals on perceived review persuasiveness,F(1,146)=13.70,p b.001,η2=8.49%.None of the main effects are statistically significant at the10%level.Table2 shows the cell means for perceived review persuasiveness.Significant simple effects show that participants perceive the positive review to be more persuasive(M=5.8)than the negative review(M=5.1, F(1,146)=9.52,p b.05,η2=5.9%)when they evaluate a product associated with promotion consumption ,the photo-editing software).On the other hand,participants perceive the negative review to be more persuasive(M=5.5)than the positive review(M=5.0, F(1,146)=4.62,p b.05,η2=2.9%)when they examine a product associated with prevention consumption ,anti-virus software). Thefindings support both H1a and H1b.In sum,when evaluating product reviews,consumers show a positivity bias for products as-sociated with promotion consumption goals and a negativity bias for products associated with prevention goals.Thefindings extend prior literature on the congruence between regulatory focus and biad information ,Aaker and Lee,2001;Keller,2006).
5.Study2刘德华的老歌
Study2examines consumer product reviews to replicate thefindings of Study1.Study2extends Study1in two meaningful ways.First,unlike Study1in which the dependent var-
iable is consumer perception,Study2us a behavioral dependent variable—consumers'votes on review helpfulness.Second,Study1 is bad on the data from controlled experiments,whereas Study2 is bad on the data from an actual retail website.
5.1.Data
For the purpo of this appears to be a good source to collect hosts one of the most popular forums to post consumer reviews.To review a product on Amazon. com,a consumerfirst gives an overall asssment using star ratings (from one tofive),and then,in the content of the review,the con-sumer elaborates on the reasons for the assigned stars.After a review is posted,readers can rate the helpfulness of this review by answering the question“Is this review helpful to you?—Yes or No.”Over time, has accumulated a large number of product reviews as well as the helpfulness ratings of the posted reviews.Both types of data are instrumental to this rearch.
Table1
Means,standard deviations and correlation matrix.
M SD Convincing Important to me Helpful
Persuasive 5.6 1.280.570.370.53 Convincing 5.5 1.17 1.000.560.70 Important to me 4.9 1.45 1.000.60 Helpful 5.6 1.23 1.00
1338J.Q.Zhang et al./Journal of Business Rearch63(2010)1336–1341
In addition,product reviews do not appear to be heavily censored.Among popular retail ,, ,,,),policies regarding review publishing are purpoly vague about censorship.For example, Dell Inc.'s policy states that the company“rerves the right to remove or to refu to post any(review)submission for any reason.”One author,however,has extensive experience of writing reviews online. has published and maintained all of this author's reviews (including very negative ones).Bad on this author's judgment, appears to interfere with review publishing at a low level.
Consistent with Study1,Study2examines two product categories: photo-editing software and anti-virus software.Within each product category,Study2includes all the product reviews of25best llers. There are two considerations in using the best llers,as oppod to a random sample of products.First,products chon at random are likely to have a small number of product reviews and even a smaller number of helpfulness ratings,becau product reviews concentrate on a small numb
er of popular products.Second,collecting reviews from randomly lected products is,in effect,stratified random sampling,which gives more reprentation to products but less reprentation to product reviews.Drawing a simple random sample of all the published reviews is,however,extremely difficult.Given the considerations,product reviews of best llers reasonably reprent the population of all the reviews in a product category.
For each product review,the authors gather information for ven variables.1)Sales rank:product ranks according to dollar sales in the category.2)The number of available reviews of each product:the total number of product reviews to date since the introduction of the product.3)Lifetime of a review:the longevity(in days)between the date of review publication and the date of data collection.4)Con-sumption goals:a dummy indicator that turns to1for photo-editing ,promotion consumption goals)and0for anti-virus ,prevention consumption goals).5)Star ratings:the number of stars a consumer us as the overall asssment of a product. In this rearch,without analyzing review content,the authors u the number of stars as a surrogate measure for the valence of review content.Prior rearch using similar data indicates that measures obtained from coding the review content are very noisy (Chevalier and Mayzlin,2006).6)Review length:the number of typed characters in a product review.7)The percentag
e of consumers who have rated a product review as helpful:the ratio defined by dividing a)the number of consumers who have rated a review as helpful by b)the total number of consumers who have rated the review.In Study2,review helpfulness is the surrogate measure for review persuasiveness.
Thefinal datat includes27,985review helpfulness ratings with similar sample size across the two product ,14,586for photo-editing software and13,399for anti-virus software).Thefinal analysis excludes a fraction of the product reviews that have no helpfulness ratings(7%for photo-editing software and5%for anti-virus software)and thefirst2reviews of each product.It is possible that people with an interest in the success of a , marketers)may post thefirst few reviews.If so,such reviews may artificially inflate the rating of the product and/or the helpfulness rating of a review.Table3prents the summary statistics of key variables.
5.2.Model
The dependent measure in this study is a dichotomous variable (consumers voted either“Yes”or“No”to the question“Is this review helpful to you?”).A binary logit model is appropriate in this context to model the likelihood that a consumerfinds a review helpful. Specifically,in the modelin
g exerci,the authors suggest that three types of effect,namely product specific ,the consumption goals associated with the product),review specific ,review valence),and the interaction between review valence and product consumption goals,may determine review helpfulness.
In the process of model development,the initial model includes 7independent variables(Sales rank,Number of available reviews of each product,Lifetime of review,Consumption goals,Star ratings,Con-sumption goals×Star ratings,and Review Length).Then,the authors gradually reduce the number of independent variables in order to identify the most parsimonious yet adequate model specification.The model lection process relies on veral criteria:likelihood ratio test, AIC and SC statistics,and the percentage of concordant classification. Thefinal model includes only four independent variables.Table4 prents the estimates of this model.
The emphasis of the modeling exerci is the interaction term Consumption goals×Star ratings.Across different model specifications in the model development,the coefficient estimates for this inter-action term are consistent(both in direction and effect size),sug-gesting the robustness of the results.In addition,model estimates obtained from different ,top and bottom5products among the25best llers)are,in general,consistent with estimates obtained from the
entire sample.The authors,therefore,ba their conclusions on the estimates of thefinal model using the entire sample.
5.3.Results
According to Table4,the coefficient estimate for the interaction term(Consumption goals×Star ratings)is0.42(p b.001).This result suggests a strong moderating effect of consumption goals on the relationship between review valence and persuasiveness.Given the dummy variable coding(1for promotion consumption goals and0for prevention consumption goals),the authors decompo the interaction effect into two simple effects:a)the effects of Star ratings for products associated with promotion consumption goals(−0.26+0.42=0.16), and b)the effect of Star ratings for products associated with prevention consumption goals(−0.26).The positive effect of review valence(0.16) suggests that,for products associated with promotion goals,an increa in Star ratings leads to a greater probability that consumersfind a review as ,a positivity bias).Converly,the negative effect of review valence(−0.26)indicates that,for products associated with prevention goals,an increa in Star ratings leads to a smaller probability that consumersfind a review as ,a negativity bias).The results support H1a and H1b.Fig.1depicts this interaction effect with ln (odds ratio)as the dependent measure and Review length at its mean
Table2
奇人奇事
Results of Study1.
Promotion consumption goal Prevention consumption goal
Negative review Positive
review
Positive
vs.
negative火高
review
Negative
review
Positive
review
Positive
vs.
negative
review
(n=38)(n=37)(n=40)(n=35)
Review persuasiveness 5.1 5.80.7** 5.5 5.0−0.5* (1.19)(.94)(.23)(.87)(.97)(.23)
Note:Standard deviations are reported in parenthes;*p b.05,**p b.01.
Table3
Data description.
Anti-virus software Photo-editing software
Average stars 2.6 3.6 Average number of helpfulness ratings10.619.4 Average number of typed characters851.8795.8 Average review lifetime(in days)279.3402.0 %of reviews without helpfulness ratings 5.2%7.5% Total number of product reviews12657521339
J.Q.Zhang et al./Journal of Business Rearch63(2010)1336–1341
level (830characters).Different values in Review length move both lines in Fig.1in the same direction but will not change their interaction.
In terms of effect size,the results suggest that,for products asso-ciated with promotion goals,a one-star increa in review rating increas the odds ratio by about 17%(exp(0.16)–1=0.17),whereas for products associated with prevention goals,the same change de-creas the odds ratio by 23%(exp(−0.26–1)=−0.23).In compar-ison,Review length ,as a control variable,has a positive effect on review persuasiveness.A one-hundred-character addition in Review length increas the odds ratio by 5.13%.
In sum,the above results suggest that the consumption goals associated with a product moderate th
e effect of review valence on persuasiveness.Together,the results of Studies 1and 2offer strong empirical evidence from both inside and outside the lab tting to support H1a and H1b .Next,the authors discuss the implications of the findings.
6.General discussion 6.1.Conclusions and discussion
From the perspective of eWOM urs,this paper explains how consumers evaluate valenced product reviews and how this evalua-tion relates to review persuasiveness.The results of this rearch show that consumers do not give equal weights to positive and neg-ative product reviews.Rather,the consumption goals that consumers associate with the reviewed product trigger consumers'regulatory foci,which,in turn,bias consumers'evaluations of positively and negatively valenced product reviews.For products associated with promotion consumption goals,consumers show a positivity bias,whereby they rate positive reviews as more persuasive than negative ones.Converly,consumers show a negativity bias for products asso-ciated with prevention consumption goals.
Word-of-mouth,online or off-line,is a form of interpersonal inter-action.Prior rearch is inconclusive regarding the effects of message valence on persuasiveness with both positivity and negativity bia
s reported (e.g.,Herr et al.,1991;Skowronski and Carlston,1987).Some recent studies examine boundary condition variables (e.g.,consumers'prior impression,review extremity,product category)in order to shed light on the equivocal findings (e.g.,Gershoff et al.,2003;Park and Lee,2009;Sen and Lerman,2007).The findings of this paper add to this growing body of literature by showing that consumer regulatory focus also matters.In the domain of product reviews,this rearch demonstrates that the moderation effect of product consumption goals partly explain the mixed effects of mes-sage valence on persuasiveness.
狮城新加坡
Throughout the paper,the authors are careful in describing the notion of consumption goals associated with a product.The goals are in the eyes of the consumer.Although the two types of con-sumption goals (i.e.,promotion and prevention)can sometimes over-lap in a product,this rearch focus on situations where one of the two goals is dominant.Future rearch can examine the situation where one product cloly associates with both promotion and pre-vention consumption goals.For example,the usage of a laptop computer may include both personal entertainment (i.e.,a promotion goal)and backing up critical data (i.e.,a prevention goal).
The authors draw conclusions bad on the findings from two software products.Future rearch may replicate the findings of this paper using data from other product categories.This rearch does
not examine the effects of review valence on consumers'product attitude (e.g.,Laczniak et al.,2001),product choice (e.g.,Gupta and Harris,in press ),or post-purcha attitude (e.g.,Bone,1995).Instead,this rearch examines a product review's persuasiveness or uful-ness for eWOM urs.Lastly,in Study 2,the large sample size of collected product reviews increas the statistical power to reject the null hypothesis,which may rai the possibility of statistical signi ficance but practical insigni ficance.6.2.Managerial implications
The Internet has greatly empowered consumers in their ability to gather and disminate product related information.Today,con-sumers can easily access peer-generated product information around the globe and can also in fluence numerous consumers by voicing their own experiences (Ward and Ostrom,2003).Rearchers who are aware of this new phenomenon call for new knowledge to understand consumer behavior in virtual communities and,more importantly,how firms can u this knowledge (Laroche,2010).This rearch attempts to explain eWOM ufulness in the eyes of consumers.The findings of this rearch offer implications to help managers embrace the notion of “consumer advocacy ”in that firms should strive to provide uful and complete information to consumers in order to earn their trust and purchas (Urban,2005).
In the context of consumer product reviews,the authors suggest that retail companies should not cen
sor negative reviews.The reviews may be very helpful and persuasive to consumers,especially for products associated with prevention consumption goals.In addition,companies should,in fact,make a conscious effort to organize and prent the most persuasive reviews to consumers,particularly when numerous reviews are available.For made veral important changes in the prentation of product reviews during 2007–2009.The firm initially prented reviews according to their recency,then valence,then importance in Amazon's eyes (called “Spotlight Reviews ”by Amazon),and finally their helpfulness as rated by review readers.This paper explains why some reviews are more persuasive than others and thus provides theoretical underpinnings to companies'intuitive efforts in providing relevant information.When marketers apply the findings of this rearch to other product cat-egories,veral generalizability issues warrant further discussion.
Table 4
Results of Study 2.Independent Variables
Model Estimate p -value Intercept
1.29(0.04)b .001Consumption goals −0.93(0.06)b .001Star ratings
−0.26(0.01)b .001Consumption goals ×Star ratings 0.42(0.02)b .001Review length
0.0005
(0.00002)
b .001
Note:Standard deviations are reported in
parenthes.
Fig.1.Interaction effect between consumption goals and Star ratings (Study 2).
1340
J.Q.Zhang et al./Journal of Business Rearch 63(2010)1336–1341
