Exploring the influence of information from customer engagement behavior on customer satisfaction and regret

(1)

Exploring the influence of information from customer

engagement behavior on customer satisfaction and regret

Master thesis

University of Groningen - Faculty of Economics and Business

Msc - Marketing Management

June 2013

Lanxin Hu (2160641) Aduarderstraat 15 9725CL Groningen Email: l.hu.6@student.rug.nl Tel: 0626714908

(2)

1

1. Introduction

The internet has significantly affected the information search behavior of consumers. A lot of consumers look internet as an information source on product categories, brands, manufacturers, and retailers, particularly when making a purchase decision (Kulkarni et al., 2012). Especially nowadays, social media offer a good online platform for consumers to share information and experience, which is helpful to their buying decision. So consumers can not only seek online information, but also can join social media to share information (Richard and Kelli, 2013). According to Naylor et al. (2012), consumers are increasingly relying on social media to discuss and learn about unfamiliar products or brands. Thus, consumers are becoming increasingly active online besides the purchase itself. Firms are aware of this and hence using social media to engage consumers. By 2011, about 83% of Fortune 500 companies were using some form of social media to connect with consumers (Naylor et al., 2012). Social media is just one example of online information provider. Consumers can also check and participate in online recommendation, online reviewing and online rating, etc. Therefore, customer engagement or customer engagement behavior nowadays is getting critical.

In fact, customer engagement is an emerging and hot topic in the academic world and in the management field. However, the concept is still new in market research and has been dealt with by far in widely different and sometimes contradictory ways in both the academic and professional literature, so understanding the concept of customer engagement is now both timely and necessary (Gambetti and Graffigna, 2010).

(5)

purchases, can be both positive (i.e., posting a positive brand message on a blog) and negative (i.e., organizing public actions against a firm) (van Doorn, et al., 2010).

Thus, CEB is the customer’s behavior, and from their behavior, we should be able to get some insightful information and judgments towards a product, a service, a company, etc. Nonetheless, this paper will focus only on the online field of CEB (OCEB) since the increasing importance of consumers’ online behavior, not as the scope mentioned above. So here it mainly includes online review, online rating, online recommendations, online WOM, etc. The information from online CEB is non-commercial since it is initiated by consumers. It is more trustworthy than commercials.

Hence, OCEB is influential to consumer’s buying decision making (White, Harrison and Turner, 2010). Particularly, literature in economics and marketing has suggested that consumers depend on online product reviews and ratings to make buying decisions (Chatterjee, 2001; Reinstein and Snyder, 2005; Chevalier and Mayzlin, 2006). Furthermore, the literature has also demonstrated the positive relationships between the average review score and product sales and between the volume of reviews and sales (Dellarocas et al., 2005; Godes and Mayzlin, 2004; Duan et al., 2008; Liu, 2006). Thus, the non-commercial information from OCEB, especially from online reviews and ratings could facilitate consumers to make purchase decisions.

Admittedly, online reviewing and online rating can be manipulated (Hu et al, 2011). But this paper does not cover that.

(6)

5 First of all, the increased online reviewing and rating might bring too much information to possibly generate more choice, which may cause extra costs to decrease consumer’s decision satisfaction. Recent evidence from the psychology and economics literature indicates that even though more choice does provide additional benefits, it also imposes extra costs. Indeed, these costs, likely search costs, can even prevent people from making a decision altogether, suggesting that more choice might reduce satisfaction on the decision (Irons and Hepburn, 2007).

Second, although consumers enjoyed the choice process more when they had more choices, they also found that the process of cognitive processing is much harder, complex and, crucially, subsequently reported lower satisfaction and regret over their choices they made. So consumers could prefer to have less choice to help make a purchase decision (Iyengar and Lepper, 2000). In addition, if consumers spend a considerable amount of time on making a choice, they will feel less satisfied on the decision and the regret will be likely occurring (Su, Comer and Lee, 2008).

Third, the consistency of online review and rating are essential. When information provided by online reviewing and rating on a product or a service is not consistent, ambiguity and confusion will occur (Walsh, Thurau and Mitchell, 2007). So consumers are uncertain about their buying decision, the confidence on this purchase decision is low. Under the circumstances, customer dissatisfaction will be likely induced when customers make this buying decision (Foxman, Muehling and Berger, 1990). Regret on this decision will be likely brought about, too.

However, consumer expertise on products can decrease the uncertainty and confusion (Kramer 2007) and might reduce the search cost and complexity of processing information (Boyle, Saad, 2011).

(7)

Therefore, the main research questions are:

1. How do information quantity and consistency affect customer satisfaction and regret on buying decision?

2. How do information quantity and consistency influence consumer confidence on buying decision?

3. How do information quantity and consistency affect search cost and search complexity?

4. Does customer expertise on products play a role on question 2 and 3? And if so, what is the role of customer expertise on products on question 2 and 3?

The rest of this paper will be structured in the following order: First, the literature review will be presented. In this part, the different variables will be introduced in order to provide background information about the topics; and the hypotheses will be developed, followed by conceptual model. Second, the research design will be given. It consists of data collection method, the sample, measurement of the included variables and research methodology. Third, analysis and result will be followed. Eventually, the research results will be discussed, managerial implications and limitations will be presented and suggestions for further research will be given.

(8)

7

2. Literature review

In an increasingly networked society where customers can interact easily with other customers and companies via social networks and other new media, non-transactional customer behavior is likely to become more important in the near future. Under this condition, Customer Engagement (CE) is becoming increasingly critical (Verhoef, Reinartz and Krafft, 2010). CE consists of multiple behaviors such as WOM, blogging, providing customer reviews and ratings, and so on (van Doorn et al., 2010). Nonetheless, customers seem to be more active in engagement behavior online, since internet brings convenience and easiness of interaction for customers (Sawhney, Verona and Prandelli, 2005). So it is interesting and pivotal to look at Online Customer Engagement Behavior (OECB), which mainly includes online review, online rating, online recommendations, online WOM, etc.

However, this paper particularly focuses on online reviews and ratings as they facilitate the buying decision making, which is the topic of this research as pointed out earlier.

2.1 Information load and information consistency

With the rise of e-commerce, online consumer reviews and ratings have increasingly become vital sources of information that help consumers to make their purchase decisions. Nevertheless, the large inflow of online consumer reviews and ratings have caused information overload, making it difficult for consumers to select reliable reviews (Baek, Ahn and Choi, 2012). Besides, online consumer reviews and ratings can lead to inconsistent information due to different opinions of people (Katsis et al., 2010).

(9)

capability of human beings to assimilate and process information during any given unit of time (Jacoby, 1977).

Information consistency refers to text, images, and other content remaining the same regardless of how and where they are presented (Costello, 2008).

Online environment allows everyone to provide online reviewing and rating about different or similar products and services, so the amount of product information is huge (Hoffman and Novak 1996) and the inconsistent information likely shows up.

2.2 Search cost and search complexity

Search refers to: “the motivated activation of knowledge stored in memory or acquisition of information from the environment” by Engel, Blackwell and Miniard (1995, p. 182). It is conducted in order to reduce risk during the decision-making process and helps to ensure making an optimum choice (Evans, Moutinho and Van Raaij, 1996). Search is a process that takes place both internally within the consumer as well as by the consumer acting externally in the environment. Under the online condition, it is interesting to understand both internal and external factors. First, internal psychological aspect of the consumer is cognitive processing. Second, one of the important external factors is the cost of online search (Grant et al., 2007).

Since most online search is free of charge, user’s search cost can be mainly defined as the amount of time spent on searching the Internet (Bakos, 1997; Brynjolfsson and Smith, 2000). The literature suggests that the cognitive processing is related to complexity (Bystom and Jarvelin, 1995). With more different information increases, that is, more products’ information are considered, analyzed and processed by consumers who look for products, the search complexity rises accordingly. Therefore, consumers must invest more cognitive effort to gather and process the information really needed to choose the most suitable alternative (Hu et al., 2007).

(10)

9

2.3 Consumer confidence on decision

Prior researches show quite some definitions about consumer confidence. Howard (1989, p. 34) defines confidence as the buyer’s extent of certainty that his evaluative judgment of a brand is correct.

According to Adelman (1987), consumer confidence is defined as the degree to which an individual feels capable and assured with regard to his or her marketplace decisions and behaviors. As such, consumer confidence reflects subjective evaluations of one's capability to generate positive experiences as a consumer in the marketplace.

Moreover, confidence may refer to the buyer's overall confidence in the brand. To be specific, the degree of certainty the buyer perceives toward a brand (Howard and Sheth, 1969). Alternatively, it may refer to the buyer's confidence in his ability to judge or evaluate attributes of the brands (Bennett and Harrell, 1975).

More importantly, consumer confidence is an economic indicator, which measures the degree of optimism that consumers feel about the overall state of the economy and their personal financial situation in the economic field (Ludvigson, 2004).

Thus, the definition of consumer confidence is differing in different contexts. While this paper mainly adopts the definition from Wendler (1983). Confidence refers to a consumer's subjective certainty that he or she has made the decision that is best for him or her (Wendler, 1983). So its meaning here is consumer confidence on buying decision.

2.4 Consumer expertise

(11)

Sujan, 1987; Johnson, 1998). In this paper, these are crucial and this paper focuses on consumer expertise towards products.

2.5 Customer satisfaction

Customer satisfaction refers to a customer's overall evaluation of the performance of an offering to date (Johnson and Fornell, 1991). In this case, it usually directs customer satisfaction to products or services.

Nevertheless, this definition is not applied in this paper.

Cadotte et al. (1987) define customer satisfaction as an affective state and satisfaction is a pleasurable contentment. Oliver (1999) explains it as the customer senses which are pleasurable. Therefore, customer satisfaction is a kind of pleasurable and content sense or emotion. This definition is applied here and this sense or emotion is towards customer’s purchase decision.

2.6 Customer regret

Regret has been defined as the negative cognitively-based emotion when recognizing or imagining that our current situation would have been better if we acted differently (Zeelenberg, 1999). Regret is linked with thoughts, feelings, and desired actions such as thinking about a mistake that has made and a lost opportunity. Simply, just doing something differently, and wishing to have a second opportunity to enhance one's performance (Zeelenberg and Pieters, 1999).

2.7 Hypotheses

2.71 Effect of a large amount of information and consistent information on customer satisfaction and regret on decision

(12)

11 complexity and makes information processing for purchasing tasks confusing.

Moreover, the information can be inconsistent, which can even cause more confusion (Snider, 1993). Confusing decisions will likely lead to customer dissatisfaction and regret on buying decision (Foxman, Muehling and Berger, 1990). While if the information were consistent, it would be helpful to bring down the confusion (Arkes et al. 1986; Oskamp 1965). Therefore, the results will be opposite.

Wherefore, hypothesis 1 and 2 can be set up.

Hypothesis 1a: A large amount of information reduces customer satisfaction on buying decision.

Hypothesis 1b: A large amount of information increases customer regret on buying decision.

Hypothesis 2a: Consistent information increases customer satisfaction on buying decision.

Hypothesis 2b: Consistent information decreases customer regret on buying decision.

On top of that, some other factors need to be considered in the relationships between a large quantity of information and customer satisfaction and between a large quantity of information and customer regret on buying decision. These factors also should be taken into account in the relationships between consistent information and customer satisfaction and between consistent information and customer regret on buying decision.

(13)

Second, the abundance of information increases search time, in other words, search cost and makes information processing complex thereby increasing search complexity. This has an impact on customer satisfaction and regret on purchase decisions. While consistent information plays a reverse role by comparison with that of a large quantity of information, as consistent information increases decision confidence (Gill et al., 1998) and decreases search cost and search complexity (Costello, 2008), thereby affecting customer decision satisfaction and regret.

Therefore, it can be believed that there are three mediators in the relationships as pointed out above, they are consumer confidence towards buying decision, search cost and search complexity. Their mediating roles will be discussed in the next section.

2.72 Mediating role of consumer confidence on customer satisfaction and regret

Online reviews and ratings can bring a huge amount of information about products. When consumers get a large amount of information, consumers will feel overwhelmed. In this situation, consumers will likely feel confused (Huffman and Kahn, 1998). Thus, consumers are likely uncertain about their buying decisions, the confidence is then lowered. This likely causes customer dissatisfaction and regret on buying decision (Walsh and Mitchell, 2008).

While Gill et al. (1998) argue that if information is consistent, greater confidence towards the decision will be produced. When the consumer’s confidence towards the decision is high, customer purchase decision on satisfaction will be significantly influenced (Shafir, Simonson and Tversky, 2000), accompanied with lower regret.

Hence, hypothesis 3a, 3b and 3c can be formed:

Hypothesis 3a: A large amount of information reduces consumer confidence on buying decision.

Hypothesis 3b: Consistent information increases consumer confidence on buying decision.

(14)

13

2.73 Mediating roles of search cost and search complexity on customer satisfaction and customer regret

When consumers gain a huge quantity of information from online reviewing and rating, they will more likely dissatisfy with their purchase decision (Scammon, 1977 and Malhotra, 1982). As the increase of information quant, consumers must invest more time to gather and process the information really needed to choose the most suitable choice (Hu, et al., 2007). This means that the search cost rises. In addition, the large quantity of information available on the Internet is often not adequately organized for consumer search (Alba et al., 1997), and usually the information is unstructured. Consequently it can be difficult for consumers to find the information they seek or search and hard to analyze and process (Jepsen, 2007). So search complexity increases as well. The increase of search cost and search complexity results in much more time and effort people spend, which can lead to the delay of purchase. Maybe the outcome of delay can over the benefits of a superior choice itself although people could have made a right decision for a superior option based on the online information (Irons and Hepburn, 2007). The consequence is likely to be the decrease of customer satisfaction.

While provided that the information is consistent, the most content is somewhat duplicated. In this way, the load of handling the information becomes light, time and effort consumers invest become less (Costello, 2008). Consequently, search cost and search complexity are diminished. Delay does not likely happen. Therefore, customer satisfaction on purchase decision will increase.

(15)

literature, the more people seriously consider other options; the more likely they are to have regret (Keaveney et al., 2007). If so, customer regret increases with the increase of serious considering other options. Serious considering needs the investment of time and effort, so search cost and search complexity increase.

While the consistent information of products is more or less the same, it hence reduces the load of thinking or considering. Therefore, consumers do not have to consider that seriously for all the similar information. Search cost and search complexity are lessen and customer regret decreases accordingly.

Thus, hypothesis 4a, 4b, 4c and 5a, 5b, 5c can be developed. Hypothesis 4a: A large amount of information increases search cost. Hypothesis 4b: Consistent information reduces search cost.

Hypothesis 4c: Mediator search cost decreases customer satisfaction and increases customer regret on buying decision.

Hypothesis 5a: A large amount of information increases search complexity. Hypothesis 5b: Consistent information reduces search complexity.

Hypothesis 5c: Mediator search complexity decreases customer satisfaction and increases customer regret on buying decision.

2.74 Moderating role of consumer expertise on products

(16)

15 (Johnson, 1998), which lowers their time to search, and are able to better obtain information in an unstructured environment (Brucks, 1985). In other words, no matter how a vast amount of information and inconsistent information are, expert consumers are able to only choose the most relevant information. In this way, search cost decreases as the search becomes more efficient (Bruks, 1985).

With a high extent of product expertise, it is easier for expert consumers to process new information and helpful to process important product-related information easily (Johnson and Russo, 1984). One reason is that expert consumers are able to truncate information based on their expertise on products because they think that expending the effort to process all information does not seem worthwhile (Moorthy, Ratchford, and Talukdar, 1977). So again, they only select the most relevant information no matter how a large amount of information and inconsistent information are. The other reason is that expert consumers have the capability to analyze information, distinguishing the important and relevant from the unimportant and irrelevant and better capacity to easily process and generate accurate inferences from a vast amount of information (Alba and Hutchinson, 1987) and inconsistent information. In this way, search complexity decreases.

Hence, it is possible to believe the existence of a moderator to H3ab, H4ab, and H5ab. That is the consumer expertise on products.

So hypotheses 6a, 6b, 7a, 7b and 8a, 8b can be set up.

Hypothesis 6a: Consumer expertise towards products weakens the negative relationship between a large amount of information and consumer confidence on buying decision.

Hypothesis 6b: Consumer expertise towards products strengthens the positive

(17)

Hypothesis 7a: Consumer expertise towards products weakens the positive

relationship between a large amount of information and search cost. Hypothesis 7b: Consumer expertise towards products strengthens the negative

relationship between consistent information and search cost.

Hypothesis 8a: Consumer expertise towards products weakens the positive relationship between a large amount of information and search complexity.

Hypothesis 8b: Consumer expertise towards products strengthens the negative relationship between consistent information and search complexity.

In addition to consumer expertise towards products, consumer expertise on online search also brings about the similar moderating effect with what consumer expertise towards products do. Therefore, consumer expertise on online search will be taken into account and tested in the research as well. But this paper mainly focuses on consumer expertise towards products.

(18)

17

2.8 Conceptual model

Figure 1: Conceptual model

(19)

3. Research design

3.1 Data collection method

The aim of this research is to find out the relationships between a large quantity of information and customer satisfaction and between a large quantity of information and customer regret on purchase decision; the relationships between information consistency and customer satisfaction and between information consistency and customer regret on purchase decision. In order to conduct this research, a quantitative research method will be used. To be specific, mail survey will be used to collect primary data in order to save the cost and reach people easily. This method involves sending questionnaires to the respondents with a request to complete them and return them (Malholtra 2010).

3.2 Sample

In order to easily and quickly reach people, college or university students will be mainly selected as the sample. So the age range is approximately from 18 to 35 years old. The sample size will be 100 and simple random sampling will be used.

3.3 Stimuli

The electronic product category is used as a stimulus for asking respondents since this product category requires a relatively complex buying process, not like purchase a simple fast moving consumer good. It is then likely that consumers use online reviewing and rating as a helpful tool for making a purchase decision on a electronic product.

3.4 Measurement

3.41 Independent variables

(20)

19 information quantity is the same with that of information consistency, but the specific measures are different, as shown in appendix 9.1.

3.42 Mediators

First mediator, consumer confidence towards purchase decision is measured by two statements, as illustrated in table 1. The same measuring scale with that of information consistency is applied.

Second, in the light of search cost, it basically means time, which is unambiguous and the scope of this construct is narrow, so a single item with the same measuring scale is applied here, as displayed in table 1 (Sackett and Larson,1990).

Third, in terms of search complexity, it is highly related to cognitive process with effort spent as described earlier. Thus, search complexity is measured mainly by cognitive effort, measuring this, two statements with the same scale as applied before and one question with a scale ranging from 1 (Very little effort) to 7 (A great deal of effort) are posed, as demonstrated in table 1 (Martin, 1994).

3.43 Moderators

(21)

search is measured in the same way via assessing the self-report knowledge, except for the second measure as it does not really fit in the online search context. The items are all displayed in table 1.

3.44 Dependent variables

Adapted from previously validated scales measuring satisfaction levels (Crosby and Stephens, 1987; Eroglu and Machleit, 1990; Spreng et al., 1996), customer satisfaction is measured by three items (see table 1). The three items are measured through a seven-point likert scale ranging from 1 (minimally satisfied) to 7 (highly satisfied).

Customer regret can also be measured on the basis of prior validated scales (Creyer and Ross, 1999; Bui, et al., 2009). It is measured by means of three items too (see table 1). The scale is from 1 (Strongly disagree) to 7 (Strongly agree).

Table 1

Construct Measured Item Information

Consistency

The reviews provided consistent information about the product you purchased.

The reviews differed a lot with respect to their evaluations of the product you purchased.

The reviews were very similar.

Consumer Confidence I am sure I made a good decision when buying this product. I have strong doubts about my decision to buy this product.

Search Cost It is time-consuming to make this purchase decision.

Search Complexity I thought very hard about making this buying decision. It was difficult for me to make this decision.

How much effort did you put into making this decision?

Consumer Expertise on products

How familiar are you with the product?

How clear do you have about which characteristics of the product are important in providing your maximum usage satisfaction? I know a lot about this product.

How would you rate you knowledge about this product relative to the rest of population?

Consumer Expertise on online research

How familiar are you with online search? I know a lot about online search.

How would you rate you knowledge about online search relative to the rest of population?

Customer Satisfaction To what extent would you feel satisfied with this purchase decision?

To what level would you feel content with this purchase decision? To what degree would you feel pleased with this purchase decision?

Customer Regret I really feel that I was making an error when I made this decision. I feel sorry for my decision.

(22)

21 In the next section, a multiple regression analysis will be operated to test the main effect. Baron and Kenny test will be used to test the mediating effect. The interaction test will be conducted to test the moderating effect.

Last but not least, the coefficient alphas, factor analysis for each construct and factor analysis for the whole model are going to be applied for assessing the internal consistency or reliability (Malhotra, et al., 2012).

4. Analysis and result

In this section, the data analyses and results will be discussed. In total, 124 people have responded to the mail survey. Nevertheless, only 104 responses can be used. The reason is that these 20 respondents only filled the first question out and left the rest empty since they did not use online reviewing and rating for the most recent electronic product purchasing, which is the first compulsory question.

4.1 Descriptive analysis 4.11 Demographics

The respondent’s genders are almost equally distributed with 52% female respondents and 48% male respondents. Besides, most respondents are in the age range from 21 to 30. Specifically, respondents between 21 and 25 occupy 74% and respondents from 26 to 30 represent 19%, as demonstrated in table 2. As shown in table 4, the majority of respondents are students with a distribution of 77%. Regarding the nationalities, most respondents are non-Dutch, specifically, 73%. In terms of education, 91% of the respondents have a higher education. Only 9% of the respondents have a high school degree, as illustrated in table 5.

Table 2: Age

Age range Frequency Percent

16-20 2 0.02

21-25 77 0.74

26-30 20 0.19

(23)

Table 3: Gender Table 4: Student/non-student Table 5: Education Table 6: Nationality 4.2 Reliability test

In order to judge the conceptual model, most variables (constructs) have been assessed by two or three questions or statements (items). The coefficient alphas of all constructs were computed to test reliability. As indicated in table 7, the majority of coefficient alphas of the multi-item scales used in this research are higher than 0.70. The constructs, for the dependent variables, customer satisfaction and customer regret even scored 0.937 and 0.939. These results demonstrate a rather high reliability. Only “Information Consistency” scored below 0.70 with a score of 0.429. The coefficient alphas of “Consumer Confidence” and “Information Load” are extremely close to 0.70, namely 0.698 and 0.624. Therefore, these results show a high degree of

reliability of the constructs used in this research. A detailed list of the coefficient alphas can be found in table 7. In addition, table 7 also illustrates the means and

Gender Frequency Percent

(24)

23 standard deviations of all variables.

Moreover, in order to further test the reliability of the scales, an exploratory factor analysis was conducted as an additional test. Before using a factor analysis, the Kaiser-Meyer-Olkin (KMO) score has been computed first. The majority of the Kaiser-Meyer-Olkin (KMO) scores are over 0.5, as shown in appendix 9.22, which means that using factor analysis is appropriate. However, for “Information Consistency” and “Consumer Confidence”, their KMO scores are only equal to 0.5. Nevertheless, factor analysis is still appropriate for these two constructs. The results demonstrate sufficiently high loadings per item per construct. The majority of items belong to the constructs accordingly.

However, the result of the factor analysis for all items illustrates that “Information Load” has three dimensions. The first dimension has just the recoded item 11. The second dimension consists of item 2 and the recoded item 4. The last dimension consists of item 3 and item 5. Only the last dimension is chosen since they are most close to the construct. The result also demonstrates that “Information Consistency” has two dimensions. Thus, only one dimension with item 1 and item 3 are selected due to the same reason as mentioned above. In this way, the new result of the factor analysis based on only the chosen items is nicely presented and item 1 and the recoded item 2 under the construct of “Consumer Confidence” are grouped together, as shown at the end in appendix 9.22.

Nonetheless, the new result shows that two pairs of constructs are in the same factor. The first pair is “Information Consistency” and “Consumer Confidence”. But the reason for this is likely their just mediocre Kaiser-Meyer-Olkin(KMO) scores as stated previously. So they are still considered as two different constructs throughout this paper. The second pair is “Product Expertise” and “Online Search Expertise”. Both of them are concerned with “Expertise”. Hence, the solution is to combine these two constructs into “Expertise” and treat them as one variable.

1_{Some scales of the items under the same construct points out in an opposite way, so those scales of the items are}

(25)

The coefficient alphas of “Information Load” and “Information Consistency” in table 7 are shown based on the only selected items.

Table 7: Mean, Standard Deviation and Coefficient Alpha Scores of all Variables Mean Std. Deviation Coefficient Alpha

Information Load 3.38 0.853 0.624

Information Consistency 4.70 1.190 0.429

Consumer Confidence 5.53 1.268 0.698

Search Cost 4.48 1.822 Not Applicable

Search Complexity 4.12 1.435 0.827

Product Expertise 4.75 1.175 0.802

Online Search Expertise 5.26 1.021 0.704

Customer Satisfaction 5.46 1.060 0.937

Customer Regret 1.93 1.175 0.939

Table 8: Pearson correlation table between constructs

(26)

25 The table 8 illustrates the correlations between the nine constructs. The majority of correlation coefficients are below 0.5, which indicates the insignificant correlations between the constructs. Only between “Search Complexity” and “Search Cost”, between “Customer Satisfaction” and “Consumer Confidence”, and between “Customer Regret” and “Customer Satisfaction”, there is a slight sign of multicollinearity. Their Pearson correlation coefficients are above 0.5, specifically, 0.661, 0.557, -0.528. Therefore, the collinearity diagnostics will be determined later on.

4.3 Hypothesis testing

4.31 Antecedents of Customer Satisfaction and Regret-hypothesis 1 and 2

Hypothesis 1 and 2 are tested by using a multi-regression analysis. The outcomes of the regression analysis show a weakly significant effect of information load to customer satisfaction (β=-0.15, p=0.08) but a strongly significant effect of information load to customer regret (β=0.23, p=0.02). In addition, a strongly significant effect of information consistency to customer satisfaction ((β=0.46, p=0.00), but an insignificant effect to customer regret (p=0.31) can be found. These results are displayed below in table 9a and table 9b:

Table 9a: Antecedents of Customer Satisfaction Dependent Variable

-Customer Satisfaction α Estimate

Standardized Estimate VIF F-value p-value R 2 Main effects 3.79 17.54 0.00 0.26 Information Load -0.14 -0.15 1.04 0.08 Information Consistency 0.44 0.46 1.04 0.00

Table 9b: Antecedents of Customer Regret Dependent Variable

-Customer Regret α Estimate

(27)

Hypothesis 1a: A large amount of information reduces customer satisfaction on buying decision.

Hypothesis 1b: A large amount of information increases customer regret on buying decision.

Therefore, hypothesis 1 can be approved. Specifically, hypothesis 1a can only be weakly supported, while hypothesis 1b can be strongly supported.

Hypothesis 2a: Consistent information increases customer satisfaction on buying decision.

Hypothesis 2b: Consistent information decreases customer regret on buying decision.

Thus, hypothesis 2a can be strongly supported, but hypothesis 2b cannot be proved.

4.32 Mediation testing-hypothesis 3 to hypothesis 5

According to Baron and Kenny (1986), there are four steps for testing mediation. The first step is to use a regression test between the independent variable(s) and the dependent variable. The second step is to use a regression test between the independent variable(s) and the mediator. The third step is to use a regression test between the mediator and the dependent variable. Ultimately, a regression test between the independent variable(s), the mediator and the dependent variable is conducted. The basic requirement is that the relationships must be significant from step 1 to step 3. If one or more of these relationships are insignificant, it can be concluded that mediation is not possible or likely

In this 4-Step model, mediation is supported if the effect of the Mediator (M) remains significant after controlling the independent variable(s) (X) in step 4. If the independent variable(s) is no longer significant when the mediator is controlled, the finding supports full mediation. If the independent variable(s) is still significant (i.e., both X and M both significantly predict Y), the finding supports partial mediation.

(28)

27

Step 1: Information Load, Information Consistency→ Customer Satisfaction, Customer Regret

As shown earlier in table 9a and table 9b

Step 2: Information Load, Information Consistency→ Consumer Confidence, Search Cost, Search Complexity

Table 10: Antecedents of Consumer Confidence Dependent Variable – Consumer Confidence α Estimate Standardized Estimate VIF F-value p-value R 2 Main effects 3.85 13.81 0.00 0.22 Information Load -0.17 -0.16 1.04 0.07 Information Consistency 0.47 0.41 1.04 0.00

Table 11: Antecedents of Search Cost Dependent Variable

-Search Cost α Estimate

Standardized Estimate VIF F-value p-value R 2 Main effects 2.38 2.32 0.10 0.04 Information Load 0.24 0.16 1.04 0.11 Information Consistency -0.28 -0.17 1.04 0.09

Table 12: Antecedents of Search Complexity Dependent Variable

-Search Complexity α Estimate

Standardized Estimate VIF F-value p-value R 2 Main effects 3.77 0.11 0.90 0.02 Information Load 0.05 0.04 1.04 0.68 Information Consistency -0.04 -0.03 1.04 0.77

(29)

Table 13: Proposed mediators to Customer Satisfaction Dependent Variable

Standardized Estimate VIF F-value p-value R 2 Main effects 2.36 16.93 0.00 0.34 Customer Confidence 0.47 0.56 1.00 0.00 Search Cost -0.02 -0.04 1.78 0.70 Search Complexity -0.10 -0.13 1.78 0.22

Table 14: Proposed mediators to Customer Regret Dependent Variable

Standardized Estimate VIF F-value p-value R 2 Main effects 3.36 7.05 0.00 0.18 Customer Confidence -0.33 -0.36 1.00 0.00 Search Cost 0.15 0.24 1.78 0.06 Search Complexity 0.06 0.08 1.78 0.50

Step 4: Information Load, Information Consistency, Consumer Confidence, Search Cost, Search Complexity→ Customer Satisfaction, Customer Regret

Table 15: Final mediation test to Customer Satisfaction Dependent Variable

Standardized Estimate VIF F-value p-value R 2 Main effects 2.03 13.85 0.00 0.41 Information Load -0.08 -0.09 1.1 0.27 Information Consistency 0.28 0.29 1.31 0.00 Customer Confidence 0.34 0.41 1.29 0.00 Search Cost -0.01 -0.02 1.91 0.88 Search Complexity -0.12 -0.17 1.8 0.11

Table 16: Final mediation test to Customer Regret Dependent Variable

Standardized Estimate VIF F-value p-value R 2 Main effects 2.77 4.72 0.00 0.19 Information Load 0.14 0.15 1.10 0.13 Information Consistency 0.00 0.00 1.31 0.99 Customer Confidence -0.30 -0.33 1.29 0.00 Search Cost 0.14 0.21 1.91 0.10 Search Complexity 0.05 0.06 1.81 0.59

(30)

29

Hypothesis 3

Table 17: Mediator-Consumer Confidence

Step 1 Step 2 Step 3 Step 4 Conclusion Mediator –

Consumer Confidence

IL,IC→CS IL,IC→CC CC→CS IL,IC,CC→CS IL→CC→CS

IC→CC→CS

Hypothesis 3c1 √√ √√ √ ×√√ √√

IL,IC→CR IL,IC→CC CC→CR IL,IC,CC→CR IL→CC→CR

IC→CC→CR

Hypothesis 3c2 √× √√ √ ××√ √×

Note: IL=Information Load; IC=Information Consistency; CC=Consumer Confidence; CS=Customer Satisfaction; CR=Customer Regret.

Hypothesis 3a: A large amount of information reduces consumer confidence on buying decision.

Hypothesis 3b: Consistent information increases consumer confidence on buying decision.

Hypothesis 3a is weakly supported since the relationship between information load and consumer confidence is significant (β=-0.16, p=0.07). However, hypothesis 3b is strongly supported as the relationship between information consistency and consumer confidence is significant (β=0.41, p=0.00). The results are displayed in table 10, regarding step 2.

Hypothesis 3c: Mediator consumer confidence on decision increases customer satisfaction and decreases customer regret on buying decision.

Consumer confidence is a mediator on the relationship between information load and customer satisfaction, and on the relationship between information consistency and customer satisfaction.

(31)

The first relationship is fully mediated since the effect of consumer confidence is significant (p=0.00) while the effect of information load remains no longer significant (p=0.27) in step 4. The second relationship is partially mediated as the effect of information consistency remains significant (p=0.00) in step 4. They are demonstrated in table 15.

Consumer confidence is a mediator on the relationship between information load and customer regret as the relationships in the first three steps are significant. Also, the relationship between consumer confidence and customer regret is significant after information load is controlled in step 4 (see table 17).

However, it is not a mediator on the relationship between information consistency and customer regret because the relationship between information consistency and customer regret is not significant (p=0.31) in step 1, as shown in table 9b.

The relationship between information load and customer regret is fully mediated since the effect of consumer confidence is significant (p=0.00) while the effect of information load remains no longer significant in step 4 (p=0.13), as illustrated in table 16.

Therefore, hypothesis 3c can be supported to a high extent. Consumer confidence plays a mediating role in most relationships. This concerns the relationships between information load and customer satisfaction, information load and customer regret and the relationship between information consistency and customer satisfaction. It only does not play a mediating role on the relationship between information consistency and customer regret. They are graphically displayed in the conclusion part of table 17.

(32)

31

Hypothesis 4

Table 18: Mediator-Search Cost

Step 1 Step 2 Step 3 Step 4 Conclusion Mediator -

Search Cost

IL,IC→CR IL,IC→SCOST SCOST→CS IL,IC,SCOST→CS IL→SCOST→CS

IC→SCOST→CS Hypothesis

4c1 √√ ×√ × ×√× ××

IL,IC→CR IL,IC→SCOST SCOST→CR IL,IC,SCOST→CR IL→SCOST→CR

IC→SCOST→CR Hypothesis

4c2 √× ×√ √ ××× ××

Note: IL=Information Load; IC=Information Consistency; SCOST=Search Cost; CS=Customer Satisfaction; CR=Customer Regret.

Hypothesis 4a: A large amount of information increases search cost. Hypothesis 4b: Consistent information reduces search cost.

Hypothesis 4a cannot be supported as the relationship between information load and search cost is non-significant (p=0.11). However, hypothesis 4b can be weakly supported as the relationship between information consistency and search cost is significant (β=-0.17, p=0.09). They are shown in table 11, regarding step 2.

Hypothesis 4c: Mediator search cost decreases customer satisfaction and increases customer regret on buying decision.

Search cost does not play a mediating role on the relationship between information load and customer satisfaction, and on the relationship between information consistency and customer satisfaction. The reason is that the relationship between search cost and customer satisfaction is not significant (p=0.70) in step 3, as displayed in table 13. Besides, the effect of search cost to customer satisfaction is not significant (p=0.88) after information load and information consistency are controlled in step 4, as demonstrated in table 15.

(33)

the effect between information consistency and customer regret is not significant (p=0.31) in step 1, as displayed in table 9b. Apart from these, the effect of search cost to customer regret is insignificant (p=0.10) after controlling information load and information consistency in step 4, as shown in table 16.

Thus, search cost is not a mediator on the relationships between information load and customer satisfaction, and between information load and customer regret. Also, it has no mediating effect on the relationships between information consistency and customer satisfaction, and between information consistency and customer regret (see the conclusion in table 18).

Hence, hypothesis 4c cannot be accepted.

Interestingly, search cost itself has a significant relationship with customer regret (β=0.24, p=0.06), as displayed in table 14.

Hypothesis 5

Table 19: Mediator-Search Complexity

Step 1 Step 2 Step 3 Step 4 Conclusion Mediator –

Search Complexity

IL,IC→CS IL,IC→SCOM SCOM→CS IL,IC,SCOM→CS IL→SCOM→CS

IC→SCOM→CS Hypothesis

5c1 √√ ×× × ×√× ××

Hypothesis 5c2

IL,IC→CR IL,IC→SCOM SCOM→CR IL,IC,SCOM→CR IL→SCOM→CR

IC→SCOM→CR

√× ×× × ××× ××

Note: IL=Information Load; IC=Information Consistency; SCOM=Search Complexity; CS=Customer Satisfaction; CR=Customer Regret.

Hypothesis 5a: A large amount of information increases search complexity. Hypothesis 5b: Consistent information reduces search complexity.

Hypothesis 5a cannot be confirmed as the relationship between information load

(34)

33

Hypothesis 5c: Mediator search complexity decreases customer satisfaction and increases customer regret on buying decision.

Search complexity does not play a mediating role on the relationships between information load and customer satisfaction and between information consistency and customer satisfaction. Also, it is not a mediator on the relationships between information load and customer regret and between information consistency and customer regret.

The reason is that the effects of information load (p=0.68) and information consistency (p=0.77) to search complexity are not significant in step 2, as illustrated in table 12. Moreover, the effects of search complexity on customer satisfaction (p=0.22) and customer regret (p=0.50) are not significant either in step 3, as demonstrated in table 13 and table 14. Last, the relationship between search complexity and customer satisfaction (p=0.11) and the relationship between search complexity and customer regret (p=0.59) are not significant after controlling information load and information consistency in step 4, as shown in table 15 and table 16.

Thus, hypothesis 5c cannot be supported (see the conclusion in table 19).

4.33 Hypothesis 6 testing

In order to test the moderation effect, the first step is to use a simple regression test between the independent variable(s) and a dependent variable, which is the main effect. The second step is to add an additional term, which is the interaction between an independent variable and a proposed moderating variable in the regression model. This is the interaction effect (Cohen et al., 2003). According to this theory, all the moderation analysis needed for this paper can be tested in this way. For the proposed moderators in the test, “Product Expertise” and “Online Search Expertise” are combined into “Expertise” as the result of factor analysis implies that these two constructs are in the same factor, as shown at the end in appendix 9.22.

(35)

effect between information consistency and consumer confidence is strongly significant (β=0.41, p=0.00). However, there is no any significant interaction in the model, which indicates that expertise is not a moderator on the relationships between information load and consumer confidence, and between information consistency and consumer confidence (see table 20).

Hypothesis 6a: Consumer expertise towards products weakens the negative relationship between a large amount of information and consumer confidence on buying decision.

Hypothesis 6b: Consumer expertise towards products strengthens the positive

relationship between consistent information and consumer confidence on buying decision.

Therefore, hypothesis 6 (6a and 6b) cannot be approved.

Nevertheless, one thing is worth mentioning, which is the significant effect of expertise on consumer confidence (p=0.03) in moderation effect 1. The relationship between expertise and consumer confidence is indeed significant (β=0.44, p=0.00), as indicated in table 21. So this can be a further research topic, since this paper does not cover it.

Table 20: Moderation regression model 1

Dependent Variable -

Consumer Confidence α Estimate

(36)

35

Table 21: Main effect of Expertise on Consumer Confidence

Dependent Variable - ConsumerConfidence α Estimate Standardized Estimate VIF F-value p-value R 2 Main effects 2.60 24.15 0.00 0.19 Expertise 0.59 0.44 1.00 0.00 4.34 Hypothesis 7 testing

In the moderation regression model 2, the main effect between information load and search cost is insignificant (p=0.11) but another main effect between information consistency and search cost is weakly significant (β=-0.17, p=0.09). Nonetheless, none of any interaction effect is significant, as displayed in table 22.

Dependent Variable -

Search Cost α Estimate

Standardized Estimate F-value p-value R 2 Main effects 2.38 2.32 0.10 0.04 Information Load 0.24 0.16 0.11 Information Consistency -0.28 -0.17 0.09 Moderating effect1 5.25 1.65 0.18 0.05 Information Load 0.73 0.47 0.41 Expertise -0.26 -0.14 0.62 InfoLoad*Expertise 0.18 0.65 0.27 Moderating effect2 2.53 0.96 0.42 0.03 Information Consistency -0.23 -0.13 0.78 Expertise -0.20 -0.10 0.79 InfoCon*Expertise -0.01 -0.02 0.98

Hence, expertise does not play a moderating role on the relationships between information load and search cost, and between information consistency and search cost.

Hypothesis 7a: Consumer expertise towards products weakens the positive

relationship between a large amount of information and search cost. Hypothesis 7b: Consumer expertise towards products strengthens the negative

relationship between consistent information and search cost.

(37)

4.35 Hypothesis 8 testing

In the moderation regression model 3, the main effects of information load (p= 0.68) and information consistency (p=0.77) on search complexity are insignificant and none of any interaction effects are significant as well, as shown in table 23.

Dependent Variable - Search

Complexity α Estimate Standardized Estimate F-value p-value R 2 Main effects 3.77 0.11 0.90 0.02 Information Load 0.05 0.04 0.68 Information Consistency -0.04 -0.03 0.77 Moderating effect1 2.48 1.68 0.18 0.05 Information Load 0.01 0.01 0.99 Expertise -0.28 -0.19 0.49 InfoLoad*Expertise 0.02 0.08 0.89 Moderating effect2 2.77 1.53 0.21 0.04 Information Consistency -0.06 -0.04 0.93 Expertise -0.30 -0.20 0.61 InfoCon*Expertise 0.01 0.03 0.97

Thus, expertise does not play a moderating role on the relationships between information load and search complexity, and between information consistency and search complexity.

Hypothesis 8a: Consumer expertise towards products weakens the positive relationship between a large amount of information and search complexity.

Hypothesis 8b: Consumer expertise towards products strengthens the negative relationship between consistent information and search complexity.

Therefore, hypotheses 8a and 8b cannot be confirmed.

4.36 Summary

Since all the Variance Inflation Factors (VIF) scored far below 4, multicollinearity is not an issue.

(38)

37

Table 24: Summary

The goal of this paper is to find out the impact of non-commercial information from Online Customer Engagement Behavior (OCEB) specifically, from online reviewing and rating, on customer satisfaction and regret towards their purchase decision. Information here refers to information quantity and consistency.

Hypothesis Outcome

Hypothesis 1a: A large amount of information reduces customer

satisfaction on buying decision. Weakly supported

Hypothesis 1b: A large amount of information increases customer regret

on buying decision. Strongly supported

Hypothesis 2a: Consistent information increases customer satisfaction

Hypothesis 2b: Consistent information decreases customer regret on

buying decision. Not supported

Hypothesis 3a: A large amount of information reduces consumer

confidence on buying decision. Weakly supported

Hypothesis 3b: Consistent information increases consumer confidence

Hypothesis 3c: Mediator consumer confidence on decision increases customer satisfaction and decreases customer regret on buying decision.

Supported to a high extent

Hypothesis 4a: A large amount of information increases search cost. Not supported Hypothesis 4b: Consistent information reduces search cost. Weakly supported Hypothesis 4c: Mediator search cost decreases customer satisfaction and

increases customer regret on buying decision. Not supported Hypothesis 5a: A large amount of information increases search

complexity. Not supported

Hypothesis 5b: Consistent information reduces search complexity. Not supported Hypothesis 5c: Mediator search complexity decreases customer

satisfaction and increases customer regret on buying decision.

Not supported Hypothesis 6a: Consumer expertise towards products weakens the

negative relationship between a large amount of information and consumer confidence on buying decision.

Not supported Hypothesis 6b: Consumer expertise towards products strengthens the

positive relationship between consistent information and consumer confidence on buying decision.

positive relationship between a large amount of information and search cost.

negative relationship between consistent information and search cost.

positive relationship between a large amount of information and search complexity.

negative relationship between consistent information and search complexity.

(39)

So far, the impact is clear.

Customer satisfaction on buying decision is decreased if consumers are confronted with too much information from online reviewing and rating, but the result demonstrates that the effect is not rather significant. Customer satisfaction is increased if consumers encounter consistent information from online reviewing and rating. Customer regret on buying decision rises if consumers are confronted with a large amount of information. But consistent information does not have any influence on customer regret on buying decision.

After exploring the goal, the answers of the four main research questions as proposed in the beginning can be discovered.

1. How do information quantity and consistency affect customer satisfaction and regret on buying decision?

The conclusion was just previously drawn.

2. How do information quantity and consistency influence consumer confidence on buying decision?

Consumer confidence on purchase decision is negatively affected by a large amount of information, but the result indicates that the influence is not rather significant. However, it is indeed positively influenced by consistent information.

Another notable finding related to this question is that consumer confidence on buying decision is a mediator on the relationship between information load and customer satisfaction, on the relationship between information load and customer regret and on the relationship between information consistency and customer satisfaction.

3. How do information quantity and consistency affect search cost and search complexity?

(40)

39 Nonetheless, the weak evidence implies that consistent information indeed decreases search cost.

Another striving finding related to this question is that search cost and search complexity both do not play a mediating role on the effects between information load and customer satisfaction on buying decision and between information load and customer regret on buying decision. Also, they are not the mediators on the effects between information consistency and customer satisfaction on buying decision and between information consistency and customer regret on buying decision.

4. Does customer expertise on products play a role on question 2 and 3? And if so, what is the role of customer expertise on products on question 2 and 3?

(41)

5. Discussion

The most important contribution of this paper is the demonstration of customer’s feelings, specifically, customer satisfaction and customer regret after their purchase decision making by using non-commercial information from online reviewing and rating, because no prior literature explicitly discussed this. This paper implies that customers feel less satisfied if they look at too much online review and rating as references for their buying decision. But if all the information is consistent, they feel more satisfied. Another insightful finding is that customers feel regretful if they check too much information from online reviewing and rating. On top of this, customer confidence towards their purchase decision influences those effects as described before as a middleman.

In addition, the other two mediation analyses bring in a critical implication. Checking a vast amount of online reviewing and rating does not necessarily lead to a lot of time and effort involved in by customers. Besides, consistent online reviewing and rating do not necessarily turn out less effort invested, but indeed cause less time spent by customers. Thus, these findings are surprisingly different with the prior literature and only consistent with the previous literature on the negative relationship between consistent information and search cost (time spent). Furthermore, the investment of time and effort does not have an impact on the relationships between information and customer satisfaction and regret.

From moderation analysis, it can be concluded that expertise cannot strengthen or weaken any relationships between information and customer confidence on buying decision. It can apply the same to the relationships between information and search cost and search complexity.

(42)

41

6. Managerial implication

The results of our study have a few important implications in the managerial field.

First, in order to enhance customer satisfaction on buying decision, product online ranking websites should make a distinction or two groups between superior online review and rating and inferior online review and rating rather than just letting mixed comments or scores together or making an order from high score to low score on products.

Second, in order to reduce customer regret on purchase decision, product online ranking websites can classify the similar comments in one category and endow a meaningful summary to this category. In this way, the information becomes less. This also helps to heighten customer satisfaction.

Third, lessening quantity of information and increasing consistency of information are not necessarily the best method to make consumers quickly and easily check online reviewing and rating.

Fourth, in order to increase consumer confidence on buying decision making, one of the approaches companies can take is to take measures to increase consumer’s expertise towards their products. Consequently, it can increase customer satisfaction and decrease customer regret on buying decision as well.

(43)

7. Limitation and further research

Even though our study produced interesting and meaningful findings, several limitations need to be acknowledged.

First and foremost, the sample size is not quite large. The minimum criteria of n=30 is met, but the useful sample size around 100 could be not that representative. The majority of data is collected through schoolmates, friends and family members. The sample thus cannot be really called random selection and the education level in the sample is rather high.

Second, the research based on the electronic product category as only one stimulus might be somewhat restricted. The results might be different if the stimulus extends to a car category or even a real estate category.

Third, this paper only studies the moderating role of expertise. The significant relationship between expertise and consumer confidence on buying decision can be also interesting to have a further exploration.

(44)

43

8. References

Adelman, P. K. (1987). Occupational Complexity, Control, and Personal Income: Their Relation to Psychological Well-Being in Men and Women. Journal of Applied Psychology, 72, 529-537.

Alba, J, W., & Hutchinson, J. W. (1987). Dimensions of Consumer Expertise. Jounal of Consumer Research, 13 (4), 411-454.

Arkes, H., Dawes, R., & Christensen, C. (1986). Factors Influencing the Use of a Decision Rule in a Probabilistic Task. Organizational Behavior and Human Decision Processes, 37, 93–110.

Baek, H., Ahn, J., & Choi, Y. (2012). Helpfulness of Online Consumer Reviews: Readers' Objectives and Review Cues. International Journal of Electronic Commerce, 17(2), 99-126.

Bakos, J. Y. (1997). Reducing Buyer Search Costs: Implications for Electronic Marketplaces. Management Science, 43, 1676–1692.

Baron, R. M., & Kenny, D. A. (1986). The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations. Journal of Personality and Social Psychology, 51(6), 1173– 1182.

Bearden, W. M., Hardesty, D. M., & Rose, R. L. (2001). Consumer Self-Confidence: Refinements in Conceptualization and Measurement. Journal of Consumer Research, 28, 121-134.

Beattie, J., Baron, J., Hershey, J.C., & Spranca, M. D. (1994). Psychological Determinants of Decision Attitude. Journal of Behavioral Decision Making, 7, 129-44.

Bell, D. E. (1982). Regret in Decision Making under Uncertainty. Operations Research, 30, 961–981.

Exploring the influence of information from customer engagement behavior on customer satisfaction and regret