Positive emotional expressions and online
consumer reviews: the effect of consumer
involvement
By
Marc Deckers
University of Groningen
Faculty of Economics and Business
Business Administration, spec. Marketing
Wielewaalplein 174
9713 BR Groningen
0624833836
m.r.deckers@student.rug.nl
Student number 1745492
Supervisor: Dr. Thorsten Wiesel
2
ndSupervisor: Dr. J.A. Voerman
2
Abstract
This study investigates the effect of positive emotional expressions in online consumer reviews on the buying intention and product evaluation towards shampoo and a digital camera. Hereby, also the level of consumer involvement is taken into account. The shampoo and digital camera product were chosen, because a pre-test indicated heterogeneity in involvement for both products. This study houses no significant results to support the hypothesis that positive emotional expressions are positively affecting the buying intention and product evaluation towards a product. In contrast, the opposite effect was found. Positive emotional expression is negatively influencing the buying intention and product evaluation towards the digital camera product. Furthermore, no significant support was found to conclude that the level of consumer involvement was determinant for the effect that emotional expressions have on the buying intention and product evaluation of a product.
3
Content
Abstract ... 2 1. Introduction ... 4 1.1 Problem Statement ... 6 1.2 Structure of thesis ... 8 2. Literature ... 9 2.1 Emotions ... 92.1.1 Emotional contagion and emotional labour in traditional WOM ... 9
2.1.2 Emotional contagion and emotional labour in electronic WOM ... 10
2.2 Involvement ... 11
2.2.1 Low involvement vs. high involvement ... 11
2.2.2 Involvement from the ELM viewpoint ... 12
2.3 Conceptual Model ... 13 3. Methodology ... 14 3.1 Research design ... 14 3.1.1 Variables ... 15 3.1.2 Control variables ... 17 3.2 Pre-test ... 18 3.3 Data Collection ... 19 3.3.1 Sample descriptive ... 19 3.4 Plan of Analysis ... 20
3.4.1 Independent Samples T-Test ... 20
3.4.2 Two-Way ANOVA ... 20 3.4.3 Regression analysis ... 20 3.4.4 Manipulation Check ... 21 4. Results ... 22 4.1 Preliminary analysis ... 22 4.1.1 Buying intention ... 22 4.1.2 Product evaluation ... 24 4.2 Regression analysis... 27 4.2.1 Buying intention ... 27 4.2.2 Product evaluation ... 29 5. Discussion ... 32
5.1 Key findings and managerial implications... 32
5.2 Limitations and future research ... 34
6. Conclusion ... 35
7. References ... 36
Appendix A: Revised Personal Involvement Inventory, Zaichkowsky, 1994... 41
Appendix B: Factor Analyses ... 42
Appendix C: Manipulation Check: Independent-samples t-test ... 46
Appendix D: Independent Samples T-Test ... 47
Appendix E: Two-way ANOVA... 48
Appendix F: Regression Output ... 52
4
1. Introduction
In the pre-purchase process many consumers act to gather relevant product or service related information before proceeding to the actual purchase. The role of recommendations by friends and relatives in this process is largely recognized in the current literature (Dellarocas, 2003; Godes & Mayzlin, 2004). What is interesting, though, is that the channel by which we receive this relevant information is determinative in actually trusting the recommendation. Trend research by Nielsen (2012) shows that trust in traditional ad channels has declined rapidly, with for instance a striking decrease of 20% between 2009 and 2011. On the opposite, the confidence in online ads like search results, social media ads and online banners is still increasing. Together with the measured trust of 92% in word-of-mouth (WOM) ads these trends offer a fruitful future for online word-of-mouth marketing. Bone (1995) defines word-of-mouth as ‘those interpersonal communications in which none of the participants are marketing sources’. In the case of online word-of-mouth one can think of online forums facilitating the exchange of product experiences between consumers, but also the product ratings and user-generated reviews on commercial websites. This online or electronic word-of-mouth (eWOM) can be defined as: “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet” (Hennig-Thurau, Gwinner, Walsh & Gremler, 2004).
5 (Barton, 2006). The emergence of more and more websites to share opinions and write reviews is called one of the main developments of the internet, as seen from a consumer behavior perspective (Khammash & Griffiths, 2011; Chatterjee, 2001).
As a natural consequence, many academic studies are aimed at investigating online user-generated reviews, labeling the phenomenon as a new tool in the marketing communication mix, as can be seen in TABLE 1. In the eWOM-marketing view, the reviewer fulfills the role of ‘free sales assistant’ in providing product evaluations (Chen & Xie, 2008). By clarifying the relationship between online consumer reviews and the resulting consumer behavior, marketing managers become able to indicate the functionality of the reviews on their website. This helps them to subsequently influence the appearance of those reviews in order to direct consumers in their purchasing behavior.
Elements commonly researched in eWOM literature can roughly be divided into three categories. This first one is about review characteristics. Mudambi & Schuff (2010) developed and tested a model of consumer review helpfulness. They analyzed 1587 reviews from Amazon.com across 6 products, and indicated that review extremity (whether the review is positive, negative, or neutral), review depth (extensiveness of the reviewer comments), and product type affect the perceived helpfulness of the review. The effect of review extremity and review depth on relative sales was also investigated by Chevalier & Mayzlin (2006). They for example found that an improvement in a book’s reviews leads to an increase in relative sales. Dellarocas, Xiaoquan Zhang & Awad (2007), investigated the effect of incorporating online movie ratings to box office sales, and found this effect to be positive. In addition, Chintagunta, Gopinath & Venkataraman (2010) found valence of the review (star rating) to be the main driver of box office performance.
6 consumers might be more critical towards information that is available online, due to the varying quality of this information. In addition, De Valck, Van Bruggen & Wierenga (2009) found that consumers with low knowledge were more likely to search and believe online WOM. Other research focussed at the more descriptive variables. Awad & Ragowsky (2008) found that trust plays a more important role for woman than for men in the decision to shop online.
In line with the information search in the pre purchase process, many authors have investigated the effect of the above mentioned characteristics on sales (e.g. Forman, Ghose & Wiesenfeld, 2008; Zhu & Zhang, 2010 and Archak, Ghose & Ipeirotis, 2011), or perceived helpfulness (e.g. Li & Zhan, 2011; Ghose & Iperiotis, and Chen & Xie, 2008) of the review. Another variable that has been investigated already is the buying intention towards a product; however the fact that this variable is very clear and easy to measure is one of the reasons why it is also included in this research. Furthermore, the evaluation towards the product is investigated in this study. One the one hand, because of the connection it has with buying intention. Consumers could for example positively evaluate a product, but are not directly intended to buy the product. On the other hand, because product evaluation is a more vague and less clear concept than buying intention, which could affect the results in a different way.
1.1 Problem Statement
7
TABLE 1: prior research DependentVar IndependentVar
Reference He lpf ul ne ss Sa le s Pr oduc t e va lua tion B uy in g in te nt ion
Reviewer characteristics Reviewer characteristics Reader characteristics
Extr emity/ ra tin g/ va le nc e De pth/ wo rd count /# cha ra cte rs A rgume nt Qua lity Va ria nc e (in r atin g) Emoti on s Sour ce C re d. Invol ve me nt Expe rie nc e (in te rn et) Expe rtise Se gme nt -de sc ript ive
Mudambi, S.M. and Schuff, D. 2010
Li, J. and Zhan, L. 2011.
Pavlou, P., and Dimoka, A. 2006.
Forman, C. et al. 2008.
Zhu, F. and Zhang, X. 2010.
Chevalier, J. A. and Mayzlin, D. 2006
Dellarocas et al. 2007 Archak et al. 2011 Chintagunta et al. 2010 Liu, Y. 2006 Duan et al. 2008 Cui et al. 2012 Clemons et al. 2006 Zhang, X.M. 2006 Sun, M. J. 2012
Chen, Y. and Xie, J. 2008
Resnick et al. 2006
Ghose, A. and Ipeirotis, P.G. 2011
Godes, D. and Mayzlin, D. 2009
Hu et al. 2010
Park, D.H. and Kim, S. 2008
Zou et al. 2010
De Valck et al. 2009
Palmquist, R.A. and Kim, K.S. 2000
Klein, L.R. and Ford, G.T. 2003
Kim, J and Gupta, P. 2012
8 Therefore, the aim of this study is to investigate whether positive emotions in an online customer review contribute positively to the product evaluation or buying intention towards a product. Product evaluation is treated as: ‘The evaluation of a consumer towards a product’. The consumer could evaluate the product in a positive manner; by for example give a high grade to the product. Or for example by indicating how he/she thinks about the product and what kind of feelings the consumer has with a certain product. Next, buying intention is defined as: ‘The probability that a consumer will purchase the product’.
As can be seen in TABLE 1, this study focuses on a unique combination of variables: buying intention, product evaluation and emotional expressions in reviews. In addition, the actual influence of information that contains emotional expression is most likely to vary with the degree of product-involvement (Petty, Cacioppo, Strathman & Priester, 2005; Park & Kim, 2008; Fennis & Stroebe, 2010; Chan, 2011). Therefore, the consumer product-involvement is also incorporated in this study. As is explained later on in the literature section, the emotions and involvement concept are intertwined via the Elaboration Likelihood Model (ELM). From the managerial perspective, the findings have implications for both marketing managers and designers of e-commerce websites on how to provide online consumer reviews.
The central research question in this study is:
“What is the effect of positive emotional expressions in online consumer reviews on buying intention and product evaluation for differing consumer involvement situations?”
This question can be divided in two sub questions:
- “What is the effect of positive emotional expressions in online consumer reviews on buying intention and product evaluation?”
- “Is the effect that was indicated in the first sub question different for differing consumer involvement levels?”
1.2 Structure of thesis
9
2. Literature
In this chapter existing literature on online consumer reviews and the expression of emotions is discussed, together with an analysis of the concept consumer involvement. This theoretical analysis is aimed at conceptualizing the relationships between the variables (Van Aken, Berends and Van der Bij, 2007), resulting in several hypotheses that will be tested in the results section.
2.1 Emotions
The role of emotions in traditional WOM- or face-to-face situations has been researched extensively. As mentioned before, proponents of emotional branding state that positive emotional product-information can lead to a high degree of consumer passion to the product, which is seldom created when cultivated through the rational, argument-based appeal (Gobe, 2001). This consumer passion can improve the legitimacy of the product (Roberts, 2004). Two extensively investigated concepts explaining the role of emotional expressions are emotional contagion and emotional labour, which are used as guidance for this literature part. Furthermore, compared to the use of emotions in WOM, in which it is copious, emotional expressions are even more penetrative in eWOM, since lower thresholds for expressing oneself are applicable in anonymous communications (Kim & Gupta, 2012). Therefore, potential differences for those forms of word-of-mouth are also discussed.
2.1.1 Emotional contagion and emotional labour in traditional WOM
A phenomenon called emotional contagion was found in 1994 by Hatfield, Cacioppo & Rapson. This widely studied concept can be explained as: one persons’ “catching” of the emotions another person displays in an interaction (Hatfield et al., 1994). A closely related topic is emotional labour, which refers to the "effort, planning, and control needed to express organizationally desired emotions during interpersonal transactions" (Morris and Feldman 1996). Both concepts are particularly used in service organizations, like airlines or theme parks, in which the role of the customer is of great importance. The power of emotional contagion and emotional labour is clearly noticeable in for example a famous theme park like Disneyland. Disney, often also called “the smile factory”, makes his personnel smile all day long. Those ever-smiling Disney theme park employees have become a stereotype of modern culture (Bryman, 2004). The employees are conveying the impression that they’re having fun too and therefore not engage in real work. It was found that when service providers maintain an expression signalling positive affect (service with a smile), consumers report higher incidences of positive affect themselves (Babin & Harris, 2009).
10 and evaluations of the service quality during service encounters (Pugh, 2001). To recapitulate, emotions and emotional contagion are being increasingly recognized as crucial variables influencing individual behaviour and organizational functioning (Vijayalakshmi & Bhattacharyya, 2012).
2.1.2 Emotional contagion and emotional labour in electronic WOM
Unfortunately, emotions in online communication typically lack necessary elements such as nonverbal cues like facial expressions, tone of voice, physical proximity and personal ties between senders and receivers (Kim & Gupta, 2012; Hatfield et al., 1994). To compensate for the absence of nonverbal cues in online communication, emoticons :-) (character strings) and smileys (graphical pictograms) are widely used as substitutes (Ganster, Eimler & Krämer, 2012). Hereby, participants seem to use the emoticons and smileys in a way similar to facial behaviour in face-to-face communication with respect to social context and interaction partner (Derk, Bos, Von Grumbkow, 2008). Furthermore, emoticons and smileys were not only fun to use, but also a valuable addition to communication methods (Huang, Yen & Zhang, 2008). However, as a recent study by Ganster et al., 2012 indicated there are no differences between both forms with regard to message interpretation; it was found that smiling smileys have a stronger impact on personal mood than smiling emoticons. In other words, smileys elicit a stronger impact than emoticons (Ganster et al., 2012).
To conclude, for this research smileys, but also bold capital letters and exclamation marks, are used to compensate for the absence of nonverbal cues (Kim & Gupta, 2012). Hereby, it is expected that the effect of using smileys is similar to the effects of emotions in traditional WOM. Fennis and Stroebe (2010) also investigated the appearance of emotions in marketing, in the area of advertising. They found companies make use of so called transformational appeals, also described as emotional or affect-based appeals, in their advertising campaigns (Roberts, 2004; Gobe, 2001; Fennis & Stroebe, 2010). These types of appeals were defined as ‘the use of affect and emotion in advertising to appeal to consumers’ feelings about a product in order to persuade’ (Fennis & Stroebe, 2010) and aim to focus more on consumers’ feelings and emotions rather than consumers’ thoughts (Johar & Sirgy, 1991; Fennis & Stroebe, 2010; Cutler et al., 2000).
The following hypotheses can be formulated:
H1: The presence of positive emotions in online consumer reviews is positively affecting the buying intention towards a certain product.
11
2.2 Involvement
‘Product involvement reflects the recognition that a certain product category may be more or less central to people’s life, their sense of identity, and their relationship with the rest of the world’ as was stated by Traylor in 1981. In other words, product involvement refers to a consumer’s perceived importance of, and interest in, a product (Gu, Park & Konana, 2012). Furthermore, as was indicated by Guthrie and Kim (2009), consumer involvement is a motivational state that can be used to understand consumer attitudes towards products.
Broadly it can be stated that consumers are dividable into three categories, on the basis of their information search and sharing of information. The first group consists of consumers who are very willing to obtain product information, for example in online consumer reviews, before making a product purchase. The second group are consumers who are not only searching for, and comparing product specific information, but are also motivated to write reviews themselves. It consists of consumers who are willing to spend the time and efforts to share their experiences with others (Lee, Cheung, Lim & Sia, 2006). The last group, which is central to this study, consists of consumers whose purchase decisions are affected by the information they hear from others or for example find in reviews, however do not initiate a proactive search for this information. These consumers could make their (purchase) decisions, or change their attitude, based on a psychological model, called the Elaboration Likelihood Model (ELM) (Petty, Cacioppo & Schumann, 1983, Fennis and Stroebe, 2010).
2.2.1 Low involvement vs. high involvement
To discover the effects of involvement a distinction can be made between differing states. On the one hand a product can be a low-involvement product, and on the other hand also high-involvement products can be found. Important to know is that consumers always aim at limiting the risks associated with purchase decision. However, a general rule is that the perceived benefits should outweigh the perceived costs (Kankanhalli, 2005). Characterising for low-involvement products, like groceries and books (Kannan, Chang & Whinston, 2001), is that the risks of purchase are limited, for example because of low costs or the availability of comparable alternatives (Gu, Park & Kunana, 2012). This is one of the main reasons why a consumer rarely engages in an ‘extensive search for information or a comprehensive evaluation of the choice alternatives’ (Zaichkowsky, 1985), for low-involvement products.
12 information) and make more product comparisons to reduce risk (Clarke and Belk, 1979, Sun, 2010).
2.2.2 Involvement from the ELM viewpoint
The Elaboration Likelihood Model is a dual process theory of persuasion. Dual process models acknowledge that recipients not always thoughtfully consider message arguments; they may sometimes take short cuts in accepting or rejecting the position recommended by the communicator (Fennis and Stroebe, 2010). Furthermore, these dual process theories distinguish two routes to persuasion, which form the end points of an elaboration continuum (Fennis and Stroebe, 2010; Hoyer and MacInnis, 2008). On the one hand there is the central route to persuasion, in which involvement of the consumers is high. Due to the high personal relevance the consumer puts a lot of effort in the information search, and has many thoughts about the information source. In this situation attitude can only be changed by offering strong and quality arguments. On the other hand the peripheral route to persuasion can be seen; in this route the consumer doesn’t pay too much attention to the information source, this could be due to time-pressure or distraction. So, it is a route in which effort in information search of the consumer is low, and in which the consumer has very less thoughts about the information source because of the low personal relevance. The attitude of the consumer can be changed by heuristic (peripheral) cues, repetition and internal consistency (Petty, Cacioppo and Schumann, 1983; Fennis and Stroebe, 2010; Hoyer and MacInnis, 2008). At last needs to be mentioned that consumers’ route to persuasion is determined by the motivation and ability to think, so it is not determined by the amount of cues or arguments available in the information source (Hoyer and MacInnis, 2008).
13 H3: The presence of positive emotions in online consumer reviews is expected to have a more positive effect on the buying intention towards a certain product for lower-involved consumers compared to higher-involved consumers.
H4: The presence of positive emotions in online consumer reviews is expected to have a more positive effect on the product evaluation towards a certain product for lower-involved consumers compared to higher-involved consumers.
A conceptual framework in which the expected relationships between the independent variable and the dependent variables are displayed is provided in FIGURE 1.
2.3 Conceptual Model
FIGURE 1: CONCEPTUAL MODEL
14
3. Methodology
This chapter covers the discussion of the research design, for which the results of the pre-test functioned as input. Next, the results of the pre-test are shown. Thereafter, the way of data collection is presented, followed by a description of the sample. Furthermore, the statistical tests that are used in this study are explained in the plan of analysis.
3.1 Research design
To test the central research question: “What is the effect of positive emotional expressions in online consumer reviews on buying intention and product evaluation for differing consumer involvement situations?” a 2 (emotions: present or not) x 3 (involvement: high/medium/low) design was implemented for two product categories. The choice for these product categories was based on the outcomes of the pre-test. Results of the pre-test can be seen in paragraph 3.2. The product reviews that were used in this study were gathered from totalbeauty.com (shampoo) and from toptenreviews.com (digital camera). These reviews were adapted and manipulated in the same way. For both product categories one review was constructed in which positive emotions were expressed, for the other review emotions were not included. Emotions were added using emoticons, smileys, and bold capital letters, according to conventions of expressing written emotions (Kim & Gupta, 2012). All of these fictitious reviews contained a positive tone.
Each respondent was confronted to only one condition, either with positive emotional expressions in the product review either without emotions in the review. A respondent that was confronted with a non-emotional shampoo review was also confronted with a non-emotional digital camera review.
The reviews that were used in this study are as follows:
Shampoo X – Non-emotional Review
This shampoo works well. It makes my hair shine after shampooing. This shampoo is good for all hair types, especially for those who heat-style their hair regularly. I usually stick to budget shampoos but I will be purchasing this because it is worth the money." – Marri_Anne, TotalBeauty.com member
Shampoo X – Emotional Review
This shampoo works SO WELL!!! I have never seen my hair SO SHINY AFTER SHAMPOOING!!! This shampoo is great for all hair types, especially those who heat-style their hair regularly. I usually stick to budget shampoos but I WILL BE PURCHASING THIS because it is WORTH THE MONEY!!!" –
15
Digital Camera Y – Non-emotional Review
Y's latest compact digital camera model, the FinePix Y550, has useful features not yet seen in many other digital cameras (GPS, long-lasting battery). The Y550 takes the highest-quality photos I've ever seen a digital camera take and processes them in the blink of an eye. – Chu_Randy TopTenReviews.com member
Digital Camera Y – Emotional Review
Y's latest compact digital camera model, the FINEPIX Y550, has USEFUL FEATURES not yet seen in many other digital cameras (GPS, LONG-LASTING BATTERY)!!!. The Y550 takes the HIGHEST-QUALITY PHOTOS I'VE EVER SEEN a digital camera take and processes them IN THE BLINK OF AN EYE!!! –
Chu_Randy TopTenReviews.com member
As was mentioned already the two non-emotional reviews are the original reviews from respectively totalbeauty.com and toptenreviews.com, although they concern the fictional products X and Y in this study to eliminate brand preferences. Furthermore, it needs to be mentioned that both non-emotional reviews are manipulated the same way, so both with equal smileys and punctuation.
3.1.1 Variables
The dependent variables in this study are buying intention and product evaluation. The buying intention measures were based on research from Singh and Cole (1991) and Juster (1966). Research of Troye and Supphellen (2012), Zhao et al. (2011) and Kim and Gupta (2012) was used to construct measures for product evaluation.
First of all, an internal reliability and consistency analysis through Chronbach’s alpha was performed. This was done for both products separately. See TABLE 4.
TABLE 4: RELIABILITY AND CONSISTENCY ANALYSIS (CHRONBACH’S ALPHA)
Measures Questions Measure Reliability (Chronbach’s α) Buying intention
(Singh & Cole, 1991 and Juster, 1966)
-I will consider buying this product
-I will definitely buy this product
-I would recommend this product to my relatives -Assuming that you are planning to make a purchase in the product category, how likely would it be for you to consider this specific product?
All on a 7-point Likert-scale (1=strongly disagree/ very unlikely, 7= strongly agree/very likely)
16 Product Evaluation (Troye & Supphellen, 2012; Zhao et al., 2011 and Kim & Gupta, 2012) - like/dislike, useful/useless, negative/positive, satisfied/dissatisfied) - On a scale from 1-10, I would rate this product with a:
- I think the described product is an excellent product.
- 7 point sematic differential scale
- scale from 1 to 10 - 7 point Likert-scale (1-Totally disagree – 7-Totally agree)
Rating on scale from 1-10 was excluded. Shampoo: 0.936 Digital Camera: 0.932 Involvement (Zaichkowsky, 1994) Revised scale of Zaichkowsky, 1994. See Appendix A.
Opposites on a 7 point scale. Shampoo: 0.926 Digital Camera: 0.905 Manipulation
Check How do you perceive the verbal tone in this review Positive - negative
Buying intention
The four questions intending to measure buying intention for the shampoo product show a Chronbach’s alpha of 0.901, which means that internal consistency is high. The high Chronbach’s alpha value indicates that a set of items (the four questions) measures a single one-dimensional construct (buying intention). Similar findings were done for the digital camera product where the internal reliability was 0.909.
Factor analysis was used to reduce the four items measuring buying intention, this was appropriate since for both products Bartlett’s Test of Sphericity was significant (p = 0.000) and KMO was 0.780 and 0.745 for the shampoo product and the digital camera respectively (APPENDIX B). Furthermore, all items ranked high on one component in the component matrix. For the shampoo product the variable was called “FAC1_BuyingintentionShamp” and for the digital camera product the item was called “FAC2_BuyingintentionDC”. Factor scores were saved in SPSS as input for a regression analysis.
Product evaluation
To determine internal reliability for the product evaluation questions some adaptions had to be made. First of all in the semantic differential scale section, the four scales: like/dislike, useful/useless, negative/positive and satisfied/dissatisfied are not measuring in the same direction. Therefore, the like/dislike, useful/useless and satisfied/dissatisfied scales were reversed; now also measuring in the same direction as the buying intention variables. Secondly, the “I think the described product is an excellent product” was already measuring in the right direction. The rating question on a scale from 1 to 10 is held separately.
Internal reliability of the product evaluation measure for the shampoo and the digital camera product were 0.936 and 0.932 respectively.
17 since p was 0.000 (APPENDIX B). The analysis indicated that the five items measuring product evaluation could be reduced to only one item. Factors were renamed in “FAC3_ProductEvalShamp” and “FAC4_ProductEvalDC” for the shampoo product and the digital camera product respectively. Finally, factor scores were saved as input for regression analysis.
Involvement
Scale of Zaichkowsky (1994) was used to determine involvement of the respondents. Also here, some items (important – unimportant, means a lot to me – means nothing to me, valuable – worthless, appealing – unappealing and involving – uninvolving) had to be reversed in a way that all items were measuring in the same direction, consistent with buying intention and product evaluation. Internal reliability was 0.926 and 0.905, for the shampoo and digital camera product respectively.
Furthermore, involvement dummy variables were created for the shampoo product as well as for the digital camera product, which function as input for regression analysis.
3.1.2 Control variables
Attitude towards product reviews
Initially three questions were developed measuring the attitude towards product reviews of a certain product. These questions were: “Before buying shampoo, I’ll always look for online reviews on the shampoo”, “When I buy shampoo, online reviews are helpful in my decision making”, and “When I buy shampoo, the online presented reviews make me feel confident in buying shampoo”.
For the shampoo product the internal reliability of the three items measuring the attitude towards shampoo reviews is high (Chronbach’s α = 0.931). The same findings were done for the digital camera product, where Chronbach’s alpha is 0.921.
Next step was to perform factor analysis to reduce the amount of items measuring the attitude towards product reviews. For both products, Bartlett’s Test of Sphericity was significant (p = 0.000), and KMO was 0.704 and 0.698 for the shampoo product and the digital camera product respectively. Factor analysis indicated that for both products the three items measuring attitude towards the review of the specific product could be reduced to only one item. Factor scores were saved as new variables in SPSS.
Commonly used controls in research on people
18 control variables are used a lot in marketing research; this enables the researcher to come up with a better marketing plan.
3.2 Pre-test
In a pre-test, 30 consumers were asked to indicate how involved they were with certain product categories. The test included 5 different product categories: Shampoo, Digital Camera, Car, Tablet PC and Audio CD. In 1985, Zaichkowsky developed a context-free 20 item scale called the Personal Involvement Inventory (PII), which measures the motivational state of involvement. To determine the respondents’ involvement, the revised scale of Zaichkowsky, 1994, was used. This scale consists of 10 items and the scores range from 10 to 70, where scores from 10-29 indicate low involvement, 30-50 indicates medium involvement and respondents who score 51-70 are highly involved with the given product (Zaichkowsky, 1994). The revised Personal Involvement Inventory is listed in Appendix A.
The following descriptives can be derived from the pre-test. First of all it should be noticed that the sample consisted of slightly more males compared to females (M = 1.37; SD= .49). Furthermore, most respondents were aged between 20-30 and 51-60. The education level of most respondents was HBO/WO-bachelor or WO-master. Income levels were widespread with a large group earning less than €15.000 a year, and a large group earning €30.000-€50.000 and €50.000-€100.000 a year.
To be able to compute the involvement scores of a single respondent in a certain product category, several steps were important. First of all, for each of the 10 contrasting items there were 7 options. Items on the left are scored (1) low involvement to (7) high involvement on the right. The second important step is that some of the items are reversed scored, which is indicated with an asterisk. The ten item scores of a single respondent were totalled to a final product category score; hereby this score could be low (10-29), medium (30-50) or high (51-70). To be able to make such calculations new variables were created in SPSS. First of all the reversed scores had to be converted into normal scores, which means that a score of 1 had to be converted in 7, and a score of 2 was converted in 6, etc. When this was done for all reversed variables, total scores had to be computed. TABLE 2 harbours the results of the pre-test.
TABLE 2: RESULTS PRE-TEST- RESPONDENT INVOLVEMENT
Involvement level Shampoo Digital Camera Car Tablet PC Audio CD
Low (10-29) 7 2 0 0 9
Medium (30-50) 16 9 9 9 14 High (51-70) 7 19 21 21 7
19 The two product categories that will be used in this study are shampoo and digital camera. Both product categories showed most heterogeneity in consumer involvement. The audio CD category also showed a lot of heterogeneity in involvement, although this category is a bit vague and is extremely driven by taste (personal preferences) of the individual (Brown, 2012). The use of shampoo and digital camera in this research is therefore more appropriate. The shampoo product has a lot of middle involved consumers, and slightly less low and high involved consumers. The digital camera category shows a low amount of low involved consumers, and more medium and high involved consumers.
3.3 Data Collection
Data was gathered by sending out a questionnaire via thesistools.nl, the link to the questionnaire was made available via email and Facebook. Subsequently, the data was transferred to SPSS. All missing values were replaced by the mean, this is appropriate as the number of missing values was limited. A dummy variable was created, to determine whether the respondent was confronted with the emotional questionnaire or with the non-emotional questionnaire. The dummy variable was named “emotional expression”.
3.3.1 Sample descriptive
TABLE 3: SAMPLE DESCRIPTIVE
Gender Age Male Female <20 20-30 31-40 41-50 51-60 >60 Pre-test (%) 19 (63.3) 11 (36.7) (0.0) 0 (36.7) 11 (16.7) 5 (10.0) 3 (26.7) 8 3 (10.0) Main research (%) 56 (54.9) 46 (45.1) 1 (1.0) (26.5) 27 (19.6) 20 (25.5) 26 (23.5) 24 (3.9) 4 Income Education >€15.000 €15.000 -€3 0.000 €30.000 -€5 0.000 €50.000 -€1 00.000 >€100.000 Don ’t wa nt to an swe r B el ow M B O MBO HA VO -VW O HB O/W O -b ac he lor WO -ma ste r Pre- Test (%) (40.0) 12 (3.3) 1 (23.3) 7 (20.0) 6 (6.7) 2 (6.7) 2 0 (0.0) (6.7) 2 (3.3) 1 (30.0) 9 (60.0) 18 Main research (%) 17 (16.7) (20.6) 21 (30.4) 31 (14.7) 15 (1.0) 1 (16.7) 17 2 (2.0) (2.0) 2 (3.9) 4 (51.0) 52 (41.2) 42
20 age categories of 20-30 and 51-60 for both the pre-test and the main research. A side note here is that in the main research age is very well distributed over all categories, except for the under 20 and above 60 category. In the main research also the income level is well divided over all categories, only respondents who earn more than €100.000 a year were in the minority. Finally, for the pre-test as well as for the main research education levels show similarities, most of the respondents were educated at HBO-WO-bachelor or WO-master level.
3.4 Plan of Analysis
In this section is explained which statistical tests are used to come to an answer to the central research question. ANOVA analyses and regression analyses were performed, below is described why it was chosen to use these tests.
3.4.1 Independent Samples T-Test
The first step in the analysis is to conduct an Independent Samples T-Test. Via this method can be analysed whether there are differences in the dependent variables (buying intention and product evaluation) between the group of consumers that was confronted with emotional expression in the product reviews versus the group of consumers that was not confronted with emotional expressions in the product reviews.
3.4.2 Two-Way ANOVA
The Independent Samples T-test analysis is followed by the execution of a Two-Way ANOVA, which offers the opportunity to include two independent variables, in this case emotional expressions as well as the level of involvement. This categorization is based on the article of Zaichkowsky (1994). In other words, a two-way ANOVA analysis was performed to check whether there were interaction effects between emotions and involvement. The first step in this analysis was to check whether the dependent variables were normally distributed. Results of Skewness and Kurtosis were within the margin of -1 and +1, which indicates that the variables are approximately normally distributed. After running a two-way ANOVA, Syntax was required to check the mean difference in buying intention (or product evaluation) of respondents who were confronted with emotions compared to consumers who were not confronted with reviews, at each involvement level.
3.4.3 Regression analysis
21 To investigate this relationship, first new variables had to be created centring involvement, which was done by subtracting the mean of each from every single data point. The next step was to create a new variable called interaction effect. This was done by multiplying the emotional expression scores with the centralized involvement scores. These two steps were performed for the shampoo product as well as for the digital camera product. Final step is to conduct a regression analysis.
The regression analysis will be performed in three steps. In step one the interaction between the dependent variable and emotions is investigated. In the following step the same interaction is investigated, however now also the level of involvement of the consumer is included. In the final step, also the interaction effect, of emotional expression and the level of involvement is measured (APPENDIX E).
The formula of the multiple regression analysis concerns the following:
i D D i
E
I
E
I
Y
0
1
2
.
In which; iY
= DV, in this case buying intention or product evaluation
= constant, unstandardized B coefficient of base case0
= unstandardized B coefficient, corresponding toE
DD
E
= Emotional expression, dummy variable1
= unstandardized B coefficient, corresponding toI I= Involvement of the consumer2
= unstandardized B coefficient, corresponding toEDIi
= error term3.4.4 Manipulation Check
22
4. Results
The first step in the results section is to look at the correlation matrix. All variables used in this study were included. Several variables showed multicollinearity, although these variables were also intended to measure the same effect. Hence, the correlation matrix indicated no reason for exclusion. The next step is the performance of the primary analyses; this section presents the results of the Independent Samples T-Tests and ANOVA analyses. In the main part of the results, the results of the regression analysis are presented.
4.1 Preliminary analysis
First of all, the results of the Independent Samples T-Test are presented. An Independent Samples T-Test was performed to verify whether there are differences in the dependent variables (buying intention and product evaluation) between the group of consumers that was confronted with emotional expression in the product reviews versus the group of consumers that was not confronted with emotional expressions in the product reviews. Results of this analysis are presented in TABLE 4.
TABLE 4: RESULTS OF T-TEST; DEPENDENT VARIABLES VS EMOTIONAL EXPRESSION
F p
Buying Intention SHAMP 1.850 0.177 Buying Intention DC 0.342 0.560 Product Evaluation SHAMP 5.268 0.024** Product Evaluation DC 3.318 0.072** * p < 0.10
** p < 0.05
As can be seen in TABLE 4 there were only significant differences between those groups in the product evaluation of both the shampoo product as well as the digital camera product.
Secondly, the Two-Way ANOVA analyses were performed. As stated before, this Two-Way ANOVA offers the opportunity to include two independent variables, in this case emotional expressions as well as the level of involvement. As was mentioned in the methodology section, involvement is categorized on the basis of the article of Zaichkowsky (1994). The results concerning the buying intention are displayed first, starting with the results of the shampoo product (SEE TABLE 5), followed by the results of the digital camera product (TABLE 6). The results are shown in a table and in a plot, to give an illustration of the results.
4.1.1 Buying intention
23 TABLE 5: BUYING INTENTION - SHAMPOO
Involvement Emotional expression
Low Medium High
No emotions expressed (n=51) Mean = 0.024 (n=12) Mean = 0.085 (n=32) Mean = 0.548 (n=7) Emotions expressed (n=51) Mean = -0.589 (n=10) Mean = -0.071 (n=35) Mean = 0.254 (n=6) F = 2.080, p = 0.152 F = 0.411, p = 0.523 F = 0.283, p = 0.596 * p < 0.10
** p < 0.05
FIGURE 2: PLOT BUYING INTENTION - SHAMPOO
As can be seen in TABLE 5 and FIGURE 2, the higher the level of involvement of the consumer for the shampoo product, the higher the buying intention towards the shampoo product is. Unfortunately, none of the results were significant. The table and the plot also show that when emotions are expressed in a shampoo review, this negatively affects the buying intention towards the shampoo product, although these results were also not significant.
24 TABLE 6: BUYING INTENTION – DIGITAL CAMERA
Involvement Emotional expression
Low Medium High
No emotions expressed (n=51) Mean = -0.535 (n=1) Mean = -0.082 (n=16) Mean = 0.141 (n=34) Emotions expressed (n=51) Mean = -0.398 (n=14) Mean = 0.071 (n=37) F = 0.742, p = 0.391 F = 0.088, p = 0.767 * p < 0.10
** p < 0.05
The table indicates that only one consumer was low involved with the digital camera product, hereby this consumer was confronted with a non-emotional review.
FIGURE 3: PLOT BUYING INTENTION – DIGITAL CAMERA
Also here, both the table and the plot give an indication for a negative effect of emotional expression towards the buying intention of, in this case, the digital camera product. However, none of these results was significant. Furthermore, the same effect of the level of consumer involvement is noticeable. The higher the level of involvement of the consumer the more consumers are intended to buy the digital camera product. Unfortunately, also these results were not significant.
4.1.2 Product evaluation
25 The dependent variable product evaluation towards the shampoo product was approximately normally distributed. Furthermore, there was homogeneity of variance between the groups as assessed by Levene’s test (p = 0.762).
TABLE 7: PRODUCT EVALUATION – SHAMPOO
Involvement Emotional expression
Low Medium High
No emotions expressed (n=51) Mean = -0.135 (n=12) Mean = 0.201 (n=32) Mean = 0.932 (n=7) Emotions expressed (n=51) Mean = -0.730 (n=10) Mean = -0.182 (n=35) Mean = 0.388 (n=6) F = 2.157, p = 0.145 F = 2.747, p = 0.101 F = 1.069, p = 0.304 * p < 0.10
** p < 0.05
FIGURE 4: PLOT PRODUCT EVALUATION – SHAMPOO
As is indicated by FIGURE 4 and TABLE 7 the product evaluation towards the shampoo product becomes lower when respondents are confronted with emotions in a review. This effect can be seen for all three involvement levels. However, none of the results was significant. But, as the multiple comparisons table indicates (APPENDIX E3), there are significant differences in the product evaluation of the shampoo product between low and high (p = 0.004) and medium and high (p = 0.051) involved consumers.
26 TABLE 8: PRODUCT EVALUATION – DIGITAL CAMERA
Involvement Emotional expression
Low Medium High
No emotions expressed (n=50) Mean = -2.929 (n=1) Mean = -0.246 (n=15) Mean = 0.461 (n=34) Emotions expressed (n=51) Mean = -0.356 (n=14) Mean = -0.110 (n=37)
F = 0.103, p = 0.749 F = 6.836, p = 0.010** * p < 0.10
** p < 0.05
FIGURE 5: PLOT PRODUCT EVALUATION – DIGITAL CAMERA
The plot indicates that the product evaluation towards the digital camera product becomes lower when respondents are confronted with emotions in a review. This effect can be seen for both involvement levels. However, results of medium involved consumers were not significant (p = 0.749). But the results of the highly involved consumers were (p = 0.010). This shows that when respondents are highly involved with the digital camera product, there are significant differences in product evaluation between consumers who were confronted with emotions compared to consumers who were not confronted with emotional expression in their review.
27
4.2 Regression analysis
The first step in this regression analysis is to check whether the control variables (Gender, Education, Age, Income, and Attitude towards the ad) show significance with the dependent variables. Hereby, age and income show to be significant with the dependent variable product evaluation however only for the digital camera product. Furthermore, the effect of the control variables is difficult to investigate in a regression analysis, this because of the categorical character of most of these control variables. Therefore, it is chosen to eliminate the control variables from the regression analysis.
Also in this second part of the results section the buying intention results are presented first, starting with the results of the shampoo product, followed by the results of the digital camera product.
4.2.1 Buying intention
The results of the regression analysis of the buying intention towards the shampoo product are displayed in TABLE 9 (more extended in APPENDIX F – MODEL A). The R-square of 0.018, in the first model of model A, explains the proportion of variability in the data set that is accounted for by the statistical model. This value indicates that only 1.8% is explained, which is quite low. The corresponding adjusted R-square, which adjusts for the number of explanatory variables in a model, is also quite low, namely 0.008. These numbers indicate that a large amount of other variables are influencing buying intention towards the shampoo product.
TABLE 9: BUYING INTENTION – SHAMPOO
Model A: Buying intention shampoo B Sig. F Sig.
1 Constant 0.134 0.338 1.850 0.177 Emotional expression -0.268 0.177 2 Constant -0.739 0.033** 4.858 0.010** Emotional expression -0.282 0.142 Involvement Shampoo 0.023 0.006** 3 Constant -0.594 0.179 3.309 0.023** Emotional expression -0.283 0.143 Involvement Shampoo 0.019 0.084* EmotionsXinvolvement shampoo 0.009 0.596 * p < 0.10 ** p < 0.05
28 regression model has become significant now (F = 4.858, p = 0.010). Involvement is slightly positively influencing the buying intention towards the shampoo product (B = 0.023, p = 0.006 in step 2; and B = 0.019, p = 0.084 in step 3). In the third step, also the interaction term is included. The one-way ANOVA analysis shows that the regression model is still significant (F = 3.309, p = 0.023). Unfortunately, the interaction term that was created, to investigate the moderating effect of involvement, appears to be insignificant (p = 0.596).
Same effects were noticeable for the digital camera product (see TABLE 10, and APPENDIX F – MODEL B); however none of the results was significant.
TABLE 10: BUYING INTENTION – DIGITAL CAMERA
Model B: Buying intention digital camera B Sig. F Sig.
1 Constant 0.058 0.680 0.342 0.560 Emotional expression -0.116 0.560 2 Constant -0.825 0.229 1.049 0.354 Emotional expression -0.159 0.431 Involvement DC 0.017 0.188 3 Constant -0.127 0.888 1.163 0.328 Emotional expression -0.163 0.417 Involvement DC 0.004 0.836 EmotionsXinvolvement DC 0.030 0.243 * p < 0.10 ** p < 0.05
29 term that was created, to investigate the moderating effect of involvement, appears to be insignificant (p = 0.243).
Finally, taking the results of the preliminary analysis and the results of the regression analysis towards buying intention into account, it can be stated that hypothesis one: “The presence of positive emotions in online consumer reviews is positively affecting the buying intention towards a certain product” is not supported. And also hypothesis three: “The presence of positive emotions in online consumer reviews is expected to have a more positive effect on the buying intention towards a certain product for lower-involved consumers compared to higher-involved consumers” cannot be supported based on the results.
4.2.2 Product evaluation
Also in this final results part, the results of the shampoo product are presented first, followed by the results of the digital camera product.
The results of the regression analysis of the product evaluation towards the shampoo product are displayed in TABLE 11 (more extended in APPENDIX F – MODEL C). The R-square of 0.050, in the first model of model C, explains the proportion of variability in the data set that is accounted for by the statistical model. This value indicates that only 5.0% is explained, which is quite low. The corresponding adjusted R-square, which adjusts for the number of explanatory variables in a model, is also quite low, namely 0.041. These numbers indicate that a large amount of other variables are influencing product evaluation towards the shampoo product. In step 2 of model C, R-square becomes 0.195 with an adjusted R-square of 0.178. In step 3 those numbers are very equal, 0.196 and 0.171 for R-square and adjusted R-square respectively.
TABLE 11: PRODUCT EVALUATION – SHAMPOO
Model C: Product evaluation shampoo B Sig. F Sig.
1 Constant 0.223 0.108 5.268 0.024** Emotional expression -0.445 0.024** 2 Constant -1.022 0.002** 11.966 0.000** Emotional expression -0.465 0.011** Involvement Shampoo 0.033 0.000** 3 Constant -0.981 0.028** 7.964 0.000** Emotional expression -0.466 0.011** Involvement Shampoo 0.030 0.005** EmotionsXinvolvement shampoo 0.006 0.668 * p < 0.10 ** p < 0.05
30 emotional expression also indicates to have a negative effect (B = -0.445) on product evaluation of the shampoo product (p = 0.024).
In the second step of the regression analysis involvement is included in the model. The one-way ANOVA analysis shows that the regression model remains significant (F = 11.966, p = 0.000). Involvement is slightly positively influencing the product evaluation towards the shampoo product (B = 0.033, p = 0.000 in step 2; and B = 0.030, p = 0.005 in step 3). Furthermore, emotions are still negatively influencing the product evaluation of the shampoo product (B = -0.465, p = 0.011).
In the third step, also the interaction term is included. The one-way ANOVA analysis shows that the regression model is still significant (F = 7.964, p = 0.000). Emotional expression in reviews is still negatively influencing the product evaluation towards the shampoo product (B= -0.466, p = 0.011). Unfortunately, the interaction term that was created, to investigate the moderating effect of involvement, appears to be insignificant (p = 0.668).
Same effects were noticeable for the digital camera product (see TABLE 12, and APPENDIX F – MODEL D).
TABLE 12: PRODUCT EVALUATION – DIGITAL CAMERA
Model D: Product evaluation digital camera B Sig. F Sig.
1 Constant 0.181 0.199 3.318 0.072* Emotional expression -0.358 0.072* 2 Constant -2.476 0.000 11.199 0.000** Emotional expression -0.474 0.011** Involvement DC 0.051 0.000** 3 Constant -3.331 0.000** 8.346 0.000** Emotional expression -0.466 0.012** Involvement DC 0.068 0.000** EmotionsXinvolvement DC -0.036 0.130 * p < 0.10 ** p < 0.05
First of all, the results of the regression model, in which only emotional expression and the product evaluation towards shampoo are included, are presented. The analysis is significant since the ANOVA analysis indicated a p level of 0.072 and an F-value of 3.318. Results indicated that emotional expression appears to have a negative effect (B = -0.358) on product evaluation of the digital camera product (p = 0.072).
31 emotions are still negatively influencing the product evaluation of the digital camera product (B = -0.474, p = 0.011).
In the third step, also the interaction term is included. The one-way ANOVA analysis shows that the regression model is still significant (F = 8.346, p = 0.000). Emotional expression in reviews is still negatively influencing the product evaluation towards the digital camera product (B= -0.466, p = 0.012). Unfortunately, the interaction term that was created, to investigate the moderating effect of involvement, appears to be insignificant (p = 0.130).
32
5. Discussion
5.1 Key findings and managerial implications
The results of this study, which are displayed in the former section, are discussed below.
Positive emotional expression in online product reviews is not positively influencing buying intention or product evaluation towards a product
The results give no support for hypothesis 1 and 2. This means that, according to this dataset, positive emotional expressions are not positively influencing either the buying intention towards a product or the evaluation of a product. In other words, the concept of emotional contagion, which is known from traditional WOM- or face-to-face situations, is clearly not having the same effect in eWOM situations. This was opposite to what was expected beforehand. This could be caused by the fact that in eWOM situations the receiver of message is dealing with information that is spread on the internet by a stranger (Kim & Gupta, 2012). Or as stated by Hatfield, in 1994, emotions in online user reviews are unlikely to evoke affective reactions in receivers because there are no personal ties or physical proximity between senders and receivers. As a consequence it could be that the receiver sees the sender of the message as less credible, and therefore takes the message less seriously and blocks the possibility of emotion transfer. However, when visiting Disneyland, for example, Disney’s staffs are also unknown to the visitors, but emotion transfer does take place in these situations.
Another possibility could be that emoticons :-) (character strings) and smileys (graphical pictograms) which are widely used as substitutes to compensate for the absence of nonverbal cues in online communication (Ganster, Eimler & Krämer, 2012), do not entail the same effect as emotions do in face to face situations. So, in contrast to studies of Derk, Bos & Von Grumbkow, 2008; Huang, Yen & Zhang, 2008 and Ganster et al., 2012, this study does not support the finding that the use of emoticons and smileys in eWOM works in a similar way to the use of emotions in face-to-face communication. Hence, offering the possibility to include emotions in online consumer reviews is not of concern for managers in their mission to increase review effectiveness.
Positive emotional expression in online product reviews has a significant negative effect on product evaluation towards a digital camera for high involved consumers
33 support was found to conclude that positive emotional expressions have a larger negative effect on the buying intention towards product for high involved consumers compared to low involved consumers. One explanation for this could be that persuasion of the consumer via the peripheral route is not similar for online consumer reviews and normal WOM, when investigating buying intention towards a product. Reviews are information sources with regard to a certain product, which require active search behaviour of consumers. In other words, it is information that is more suitable for high involved consumers, because those are willing to put effort in their information search for a certain product. In this situation consumers were confronted with product reviews and hence the active search factor is not present in this situation.
Another reason could be that emotions, in this study displayed as smileys, are regarded as childish, especially in a situation where the buying intention towards a product is investigated. The concept of buying intention is more concrete than the concept of product evaluation, for which support was found that positive emotional expressions have a negative influence, more for high involved consumers than for low involved consumers. The difference between both dependent variables is therefore also the reason why both are included in this research. Consumers could for example have a positive attitude towards a certain product, and therefore evaluate the product in a positive manner. However this does not directly imply that the consumer is also intending to buy the product. This is because the consumers’ ability to buy a product, or the possession of purchasing power, plays a major role in the buying intention towards a product (Budiman, 2012).
Finally, the results of this research indicated that hypothesis 4 is not supported. This means that no support was found to conclude that positive emotional expressions have a larger positive effect on the product evaluation towards product for low involved consumers compared to high involved consumers. Although it was found that positive emotional expressions are negatively affecting the product evaluation of a digital camera, no support was found for the second part of the hypothesis. The role of involvement was opposite to what was expected. Results, of all two-way ANOVA’s and regression analyses, showed that the higher the level of involvement, the higher the buying intention or product evaluation towards the product. However, only significant support was found for regression models C and D, and the two-way ANOVA of product evaluation of the digital camera product.
34
5.2 Limitations and future research
The amount of respondents is one factor that limits the generalizability of this study. For a 2 x 3 design at least around the 150 respondents is needed to come to a sufficient analysis. In this study 102 participants were involved, which makes the results less reliable. Furthermore, age and gender are more or less evenly distributed, but the education level is not. Around 92% of the respondents have the minimum education level of HBO/WO-bachelor, which means that around 8% of the respondents is lower educated. The database that was used is therefore maybe not a good reflection of society; however it showed to be very similar to the data that was used for the pre-test. Next to this, most of the respondents were either friends of mine, or either colleagues of my father, so most of them come from or work in the same environment. In other words, the background of the respondents used in this study is either student or teacher. Conclusively can be stated that the results of this study could be a little biased, which also not positively contributes to the generalizability of this research. For future research a more extended and evenly distributed database of respondents could be useful.
Secondly, the research is also limited by the fact the only positive emotions were tested. In future research also the effect of negative emotional expressions on buying intention and product evaluations should be taken into account.
Thirdly, the effect of emotional expressions on other dependent variables could be investigated. For example helpfulness of the review, the sales of a product, or the willingness to pay for a product could have been added to this research. As indicated in TABLE 1 of the introduction of this research these variables are widely used in the area of consumer reviews, however the influence of emotions in consumer reviews remains an under investigated area.
35
6. Conclusion
The purpose of this study was to find out what the effect of emotional expressions in online consumer reviews is on the buying intention or product evaluation towards a product. The products used in this study were, the shampoo and the digital camera product. Those products were chosen because a pre-test indicated heterogeneity in the level of consumer involvement for both of them. Involvement was expected to have a moderating effect on the relationship between emotional expressions and buying intention or product evaluation. A database was created by sending out two different questionnaires (one with, and another without emotional expressions in the reviews), which were completed by 102 respondents.
Overall it can be stated that formulating an answer to the central question has to be done with caution, due to the insignificance of most results. Therefore, this conclusion is based on the central sub questions of this research. The first one: “What is the effect of positive emotional expressions in online consumer reviews on buying intention and product evaluation?” can be answered with a negative effect. Although not all results were significant, the overall negative effect of emotions appeared to be constant in most models. The regression output and the output of the two-way ANOVA’s both showed that buying intention or product evaluation were higher when no emotions were expressed in the product review, compared to a situation in which emotions were expressed.
36
7. References
Aken, J. E. van, Berends, H. & Van der Bij, H. 2007. Problem Solving in Organizations. Cambridge, UK: University Press
Ante, S. E. 2009. Amazon: turning consumer opinions into gold. Available at: http://www.businessweek.com/magazine/content/09_43/b4152047039565.htm (accessed 8 August 2012).
Archak, N., Ghose, A. and Ipeirotis, P. G. 2011. Deriving the Pricing Power of Product Features by Mining Consumer Reviews. Management Science. 57(8): 1485 – 1509.
Awad, N. F. and Ragowsky, A. 2008. Establishing trust in electronic commerce through online word-of-mouth: an examination across genders. Journal of Management Information Systems. 24(4): 101 – 121.
Babin, B. J. and Harris, E. G. 2009. CB2. Student Edition, 2nd Edition. South Western Educational
Publishing.
Barton, B. 2006. Ratings, Reviews & ROI: How Leading Retailers Use Customer Word of Mouth in Marketing and Merchandising. Journal of Interactive Advertising. 7(1): 1 – 7.
Bickart, B. and Schindler, R. 2001. Internet forums as influential sources of consumer information. Journal of Interactive Marketing. 15(3): 31 – 40.
Bone, P. F. 1995. Word-of-Mouth Effects on Short-term and Long-term Product Judgments. Journal of Business Research, 32(3): 213 – 223.
Brown, R. A. 2012. Music preferences and personality among Japanese university students. International Journal of Psychology. 47(4): 259 – 268.
Bryman, A. 2004. The Disneyization of Society. 1st edition Sage Publications Ltd.
Budiman, S. 2012. Analysis of Consumer Attitudes to Purchase Intentions of Counterfeiting Bag Product in Indonesia. International Journal of Management, Economics and Social Sciences. 1(1): 1 – 12.
Chatterjee, P. 2001. Online reviews: Do consumers use them? Advances in Consumer Research. 28(1): 129 – 134.
Chan, Y. Y. 2011. Interplay of message framing, keyword insertion and levels of product involvement in click-through of keyword search ads. International Journal of Advertising. 30(3): 399 – 424.
Chen, Y. and Xie, J. 2008. Online Consumer Review: Word-of-Mouth as a New Element of Marketing Communication Mix. Management Science. 54(3): 477 – 491.
Chevalier, J. A. and Mayzlin, D. 2006. The Effect of Word of Mouth on Sales: Online Book Reviews. Journal of Marketing Research. 43(3): 345 – 354.
