Getting personal: The effect of different managerial response
types and the level of personalization on perceived credibility
Denise van der Meer s2214288
June 26, 2017
Supervisor: dr. J.A. Voerman
Getting personal: The effect of different managerial response types
and the level of personalization on perceived credibility
University of Groningen
Faculty of Economics & Business MSc Marketing Management
Author: Denise van der Meer
Date: June 26, 2017
Adress: Rustenburgerstraat 291-3
1073 GE Amsterdam
Phone number: 0031 628413631
E-mail address: dirvandermeer@gmail.com
Student number: 2214288
First supervisor: dr. J.A. Voerman
Management summary
This study examines the relationship between different type of managerial responses and the level of personalization of these response on the perceived credibility of that response, given a negative online customer review. Consumers may write negative online customer reviews on review website, to share a dissatisfying experience with a service provider. Based on this, management can provide a response towards this review. Several response strategies can be chosen, which in turn affect the perceptions of readers of this review. This research provides valuable insights in how different type of managerial responses and the level of
personalization enhance the perceived credibility of that response.
The research was conducted by a 3x2 between-subjects design experiment, using an online survey. A total of 297 respondents took part in this experiment in which their perceived credibility of a managerial response was measured, based on different type of managerial responses, which vary along their level of personalization.
The type of managerial response and level of personalization itself did not have an effect on the perceived credibility of a response. A significant effect was found for personalized apology, in which offering a personalized apologetic response was perceived as significantly more credible compared to a personalized refutable response. Similar effects were not found for offering a compensational response. Thereby, respondents perceived a personalized apologetic response as significantly different from a standardized apologetic response. The results support that responding with a personalized apology results in higher credibility. Within this research, the attribution of responsibility is the extent to which the service
TABLE OF CONTENTS INTRODUCTION 1.1 Introduction 6 1.2 Managerial response 7 1.3 Credibility 9 1.4 Level of personalization 10
1.5 Research questions and problem statements 10
1.6 Theoretical and managerial relevance 11
1.7 Outline of thesis 12
THEORETICAL FRAMEWORK
2.1 Type of managerial response 13
2.1.1 Apology 13
2.1.2 Refutation 14
2.1.3 Compensation 15
2.1.4 Effect of type of managerial response on credibility 16
2.2 Level of personalization 17
2.3 Covariates 18
2.3.1 Skepticism towards organizations 19
2.3.2 Need for cognition and credibility 20
2.3.3 Attribution of responsibility 20
2.4 Conceptual Model and hypotheses 21
METHODOLOGY 3.1 Research design 23 3.2 Survey 24 3.3 Procedure 24 3.4 Operationalizations 3.4.1 Manipulations 25 3.5 Operationalized
3.5.2 Variables and scales 27
3.6 Sample 31
3.7 Randomization check 33
3.8 Plan of analysis 33
3.8.1 Data preparation 33
3.8.2 Analyses
3.9 Results manipulation checks
3.9.1 Type of managerial response 35
3.9.2 Level of personalization 35
RESULTS
4.1 Analysis of variance
4.1.1 Main effects 36
4.1.2 Interaction effect 37
4.1.3 Test of Simple Effects 38
4.2.2 Regression model with scepticism towards org 40 4.2.3 Regression model with attribution of responsibility 40
4.2.4 Full regression model 41
4.3 Multicollinearity
4.4 Discussion of hypotheses 43
4.4.1 Main effect of type of managerial response on credibility 43 4.4.2 Effect of level of personalization on credibility 44 4.4.3 Effect of scepticism towards organizations on credibility 44 4.4.4 Effect of need for cognition on level of personalization 45 4.4.5 Main effect attribution of responsibility 45
DISCUSSION
5.1 Theoretical implications 48
5.2 Managerial implications 49
5.3 Limitations and future research 49
5.4 Overall conclusion 50
References 53
Appendix A: Survey 62
Appendix B: Reliability analysis 70
Appendix C: Factor Analysis 72
Appendix D: Randomization checks 76
Appendix E: Manipulation checks 78
Appendix F: Analysis of Variance 81
INTRODUCTION
Within the Netherlands, a total of 35.5 million holidays were booked in 2016 (Rabobank, 2016), of which an estimated 75 to 80 percent of the consumers were using the internet to research, and optionally book, their holidays. Hotel products are highly experiential in nature, therefore customers experience uncertainty when booking online (Wei et al., 2016). Online review websites are perceived as a valuable information source, offering a wide variety of online customer reviews (OCR), which assist consumers identify products or services that best match their needs to reduce this uncertainty. OCRs are a type of electronic word-of-mouth (eWOM). 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). Due to the fact that the information contained in OCRs does not originate from the company itself, it is generally considered highly credible and influential (Bickart & Schindler 2010).
Online travel reviews have been in existence for over a decade with TripAdvisor being the first, founded in 2000 (Lee, Law & Murphy, 2011). A significant amount of these reviews are based on hotel visits: over one third of travel businesses reviewed on TripAdvisor are accommodations (Levy, Duan & Boo, 2013). A differentiation can be made between two categories of internet-based companies providing hotel reviews. First of all, online travel agencies (OTAs), can be specified as electronic booking channels that aggregate lodging options and provide consumer-generated reviews (Levy et al., 2013). Booking and Expedia are websites that can be classified in the category of OTAs. On the other hand, online review specialists provide consumers with trip and hotel information (Levy et al., 2013).
TripAdvisor is currently dominating this market, with more than 500 million reviews and opinions on their website (TripAdvisor, 2017).
Consumers can easily exercise their voice via these platforms to share their opinions, by making use of the online customer review, (Black & Kelley, 2009). eWOM
communication can take place in many ways, e.g. web-based opinion platforms, discussion forums, boycott Web sites and news groups (Hennig-Thurau et al., 2004). Within this
provide an incentive for customers to share either a positive, neutral, or negative customer review (Yacouel & Fleischer, 2012). Subsequently, eWOM can have positive or negative implications for firms. Positive eWOM can result in free advertising for a firm, growing brand recognition and increased sales (Longart, 2010). However, negative eWOM can be costly for firms and cause irreparable damage (Kietzmann & Canhoto, 2013). This also influences prospective customers, as the contemporary consumer rely heavily on online reviews to inform their purchase decision (Hennig-Thurau et al., 2010). According to Chan et al. (2011) the importance of consumers’ interactions as a market force increased. As a result, managing customer feedback should be highly important to managers.
Online review website (e.g. TripAdvisor) added an important feature to their
platforms, in which companies are enabled to provide a managerial response towards online customer reviews. TripAdvisor (2014) conducted a survey, which highlights the importance of responding to online reviews. Of all users, 84 percent agreed that an appropriate
management response to a negative review improved their impression of a hotel, while 64 percent agreed that an aggressive or defensive management response to a bad review makes them less likely to book that specific hotel. Especially within the hospitality industry, companies have to consider the powerful impact of these platforms. First and foremost, they are seen as valuable sources of information. However, customer complaints are unavoidable from time to time, which can result in negative eWOM. Especially after a failure, dissatisfied customers might be encouraged to share their negative experiences and write negative online reviews (Kau & Loh, 2006). As a result, this can lower readers’ attitudes towards the service provider. According to Chan et al. (2011), negative word-of-mouth is more influential than positive WOM. Due to the transparency of online platforms, online customer reviews are available to a large number of customers. Therefore, firms should consider responding to negative reviews. Unfortunately, many companies lack the knowledge or ability to handle customers’ complaints in an appropriate manner (Davidow, 2003).
1.2 Managerial response
itself in a disadvantageous position by being unresponsive to negative comments, which might result in a loss of customers’ business in the future. By responding to negative
comments in its initial stage, firms immediately tackle the problem (Schmallegger & Carson, 2008), and avoid further damage to the firm. Wei et al. (2013) recommend hospitality
businesses to actively respond to consumer comments. Therefore, firms should not
underestimate the impact of (negative) eWOM, which could result in serious damage to firms or organizations.
In the case of a negative online customer review, Ullrich & Brunner (2015) state that a response from a brand or customer is always better than a non-response. These responses should be immediate and authentic (Chan & Guillet, 2011) in which each complaint should receive a response. Handling both positive and negative consumer feedback is an essential part of customer service, but managing attention selectively is not yet a conscious process. Attention is the key to managing customer reviews in which it should be prioritized to what, where, and when they should respond and how much time should be devoted to any
individual post.
1.3 Credibility
Several studies have investigated the effects of eWOM on current and future customers. Although eWOM may be less personal than traditional WOM, it is seen as more powerful because it is immediate, has a significant reach, is credible, and is publicly available (Hennig-Thurau et al., 2004). Research shows that customers tend to perceive online commentaries as credible and trustworthy (Flanagin & Metzger, 2013), and tend to shift their opinions and purchasing behaviors accordingly (Sparks & Browning, 2010). eWOM is perceived to have higher credibility for consumers than marketer-generated information, as their authors are fellow consumers, in which they have no intention to manipulate the reader (Bickart & Schindler, 2010). Contradictory, Eysenbach (2000) state that the credibility of these
unfiltered online reviews raises concerns and even suspicion of online information seekers. Cox, Burgess, Sellito & Buultjens (2009) indicate that user-generated content on websites may not be considered as credible as other sources of travel information, such as government-sponsored travel websites. Eventhough research shows that eWOM is perceived as credible (Hennig-Thurau et al., 2004; Flanagin & Metzger, 2013; Bickart & Schindler, 2010), it remains unclear to what extent managerial responses are perceived as credible.
The complaint process and related credibility perceptions continue with complaint response perceptions and third party observers incorporating an entire dialogue into their credibility assessment set (Breitsohl, Khammash & Griffiths, 2010). This contradicts with findings of online complaint literature, in which they state that the complaint process and related credibility perceptions end with the act of complaining. Breitsohl et al. (2010) presumes that third-party observers will perceive the company response to have higher
complaint utility and credibility compared to the complainer, as its identity is known and they possess expertise. Metzger et al. (2003) states that company interference with consumer opinion statements will result in credibility benefits. However, Dellarocas (2006) states that company interference might result in skepticism and loss of credibility for eWOM in general.
Perceived credibility is not a single construction according to literature (Breitsohl et al., 2010; Cheung, Lee, & Rabjohn, 2008). A differentiation can be made between content and source credibility. Source credibility is defined as the combined perception of
trustworthiness and expertise (Cheung et al., 2008). It describes attributes of the
enhance the believability of the transmitted information for message recipients (Ohanian, 1990). This implies that the judgement about the credibility is directed towards the writer of the response, which is the source. Message credibility relates to argument and information quality, which can be defined by its comprehensiveness, relevance and accuracy. According to Racherla & Friske (2012) source credibility is among the most important factors that affect online information adoption. DiStaso, Vafeiadis & Amaral (2014) state that it is crucial for organizations to have appropriate response strategy to protect their credibility. This raises the question to what extent managerial responses towards negative OCRs are perceived as credible by observers of a complaint dialogue.
1.4 Level of personalization
A differentiation can be made between a personalized or standardized response. Dens, De Pelsmacker & Purnawirawan (2015) emphasize that more research is needed to untangle the complexity of consumer evaluation processes. They state that it remains unclear what the effect of a standardized response is, in which companies offer the same response to every review. Although providing a standardized response indicates some interest by the service provider to respond towards customers, it could also be perceived as a lack of effort. Wei et al. (2013) examined whether customers’ perceived effectiveness of hotel firms’ responses vary along the generic-specific dichotomy, in which a distinction can be made between positive and negative customer engagement behaviors (CEB). They found that the type of management response to positive CEBs did not affect potential customers’ evaluation of these responses. However, specific management responses to negative CEBs gained more trust and delivered higher perceived communication quality than generic ones. Within these findings, they account for the different perceptions towards negative and positive CEB, however, future studies might explore other dimensions of management responses towards CEBs, such as the tone of the responses.
1.5 Research questions and problem statements
the level of personalization influences how different types of managerial responses are perceived as credible by observers.
The following main research question is addressed:
What is the effect of different type of managerial responses and the level of personalization of these managerial responses on the perceived credibility, given a negative online customer review?
The main question can be answered by researching the following questions: 1. What type of managerial responses can be identified?
2. To what extent does the type of managerial response influence the credibility of a response?
3. To what extent does the level of personalization influence the credibility of a response?
4. To what extent does level of personalization enhance the effect of type of managerial response on perceived credibility?
5. To what extent do certain consumer characteristics influence the tendency to perceive a managerial response as credible?
1.6 Theoretical and managerial relevance
Research found evidence that being unresponsive to negative online customer reviews results in negative effects (Bitner et al., 1990; Chan & Guillet, 2010; Wei et al., 2013; Ullrich & Brunner, 2015). Whereas most research has focused on whether management should respond to online customer reviews and the effect on current customers, it remains relatively vague which managerial responses are proven to be perceived as credible by subsequent customers. Thereby, more research is needed to obtain insights in the influence of personalization (Dens et al., 2015; Wei et al., 2013) on different types of managerial responses and how this
of OCR’s responses for managers. Hence, from an applied perspective, the findings attest to the importance of the way of responding as related to credibility of OCR’s. This provides organizations with valuable diagnostic information, as it relates to the potential utility of different responses. Emphasis by managers on enhancing their level of personalized
responses has the potential of, consequently, signalling better service quality toward potential customers. This can result in organizations obtaining greater benefit from online review websites, to potentially increase their bookings.
1.7 Outline of the thesis
The rest of this research is organized as follows. In the following section, an extensive
THEORETICAL FRAMEWORK
This chapter starts by providing an overview of several response strategies that companies can adopt to respond towards negative online customer reviews. As explained in chapter one, firms are recommended to provide a response towards negative reviews. However, it remains unclear which managerial responses are appropriate. The first independent variable, type of managerial response, will be discussed in depth by using relevant academic literature, as well as the second independent variable level of personalization. Insights will be given into their relationship with the dependent variable, credibility. Thereby, skepticism towards
organizations and attribution of responsibility will be discussed, as this possibly influences the perceived credibility of a managerial response. Lastly, need for cognition will be considered, as this consumer characteristic might influence the effect of level of
personalization on the perceived credibility of a managerial response. Each section will be concluded with subsequent hypotheses. The conceptual model is shown at the end of this chapter, which represents a graphical depiction of the relationship between the variables.
2.1 Type of managerial response
A distinction can be made between accommodative responses, which are any form of
apology, compensation and/or corrective action, and defensive response, which are any form of denial, attack or shifting the blame to others (Van Noort & Willemsen, 2012). Based on offline service recovery literature (Benoit, 1995; Conlon & Murray, Davidow, 2000; Kim, Ferrin, Cooper & Dirks, 2004; Mattila, 2001, Mueller, Palmer, Mack & McMullan, 2003) we distinguish three type of response strategies: an apology, refutation and compensation. The expected effects of these types of managerial responses towards a negative online customer review on observers and perceived credibility will be discussed in the following section.
2.1.1 Apology
repeating the same wrongful act (Kellerman, 2006). Hoffman & Chung (1999) relate to this as an empathetic response in which the organization acknowledges the customer’s complaint. However, according to their point of view, providing an apologetic response fails to take responsibility for resolving the situation.
For handling negative incidents such as service failures and customer complaints, apologies are considered to be one of the most essential response components that
organizations can use (Lee & Cranage, 2012). From the perspective of the complainer, they frequently expect and apology and should therefore be given one (Davidow, 2003). Mack, Mueller, Crotts & Broderick (2000) emphasize that customers expect organizations to admit a mistake has been made, suggesting that a simple apology is sufficient. Davidow (2003) states that apologies may be perceived differently by managers and customers. In his analysis of organizational responses to customer complaints, Davidow (2003) states that expressing regret is non-effective, neither is apologizing. Dens et al. (2015) support this statement and found that apologizing alone is not sufficient to gain the trust of prospective customers. Manika et al. (2017) state that non-customers may perceive the apology as less persuasive than customers. Therefore they recommend managers responsible for service failure apologies to develop clear and consistent messages directed to current and potential
customers. Various authors argue that the positive impact of an apology may be cancelled out by its admission of guilt, which increases the perceived severity of the problem (Kerkhof et al., 2010). To prospective customers, the organization appears to acknowledge the customer’s complaint but fails to resolve the situation (Davidow, 2003). However, Coombs & Schmidt (2000) found that organizations accepting responsibility for a wrongdoing or showing empathy for the victims are evaluated more favorably by the public.
2.1.2 Refutation
Schlenker (1980) states that an apology may fail to ameliorate the negative consequences of an accusation because it involves an acknowledgement of guilt. By acknowledging guilt, it would damage trust more than the benefits that come with an apology, by promising that the violation would be avoided in the future. Therefore, Kim et al. (2004) state that denial may represent a more effective response to a trust violation than apology because individuals might give the accused party the benefit of the doubt.
On the other hand, refutation signals a lack of empathy, which makes the company appear less trustworthy, and prompts customers to rate the problem as more severe than if the company did not respond (Kerkhof et al., 2010; Utz, Matzat & Snijders, 2009). The findings of Utz et al. (2009) contradict with the research of Kim et al. (2004), in which they state that apologies were always more successful in rebuilding trust than denials. Dens et al. (2015) state that refutation scored poorly in their research. This is in line with negative effects appearing in prior studies, indicating that refutation fails to communicate empathy towards dissatisfied customers and enhancing the credibility of the person issuing the negative review (Kerkhof et al., 2010; Utz et al., 2009). Thereby, Kerkhof et al. (2010) state that refutations lead to lower perceived corporate credibility.
2.1.3 Compensation
Besides apologizing and refutation, service providers can provide closure or a solution. Closure results from promises about the future (i.e. that the problem has been solved and will not occur again or offer compensation (Johnston and Mehra, 2002).According to Davidow (2003), compensation is the most researched aspect of complaint handling. Both options are expected to exert a positive influence on customers, due to the perceived effort of companies to resolve the customer dissatisfaction. Therefore, compensation can restore trust better than an apology alone (Mueller et al., 2003; Wen & Chi, 2013). Bitner (1990) states that
dissatisfaction can be resolved by providing customers with an explanation for a service failure and compensating them in some way. Bitner, Booms, & Tetreault (1990) suggest that compensation could turn a dissatisfying event into a memorable, satisfying encounter.
However, these assumptions are mainly based on offline service recovery literature. Although research (Hoffman & Chung 1999; Lewis & McCann, 2004) revealed that
(Levy et al., 2012). According to Dens et al. (2015) the online environment in which reviews are posted changes this perspective, as the compensation is offered to the failed customer and not the bystander. Therefore readers of online reviews, which are considered the bystander, might apply different standards than failed customers in evaluating the justice provided by the recovery (Dens et al., 2015). Davidow (2003) found that full financial or material
compensation is hardly ever offered by companies online. Customers are not only influenced by the compensation that is offered to overcome a service failure (Davidow, 2003), the way the complaint is handled exerts strong influence on the customers’ attitude. Whereas most literature focuses on the perspective of the complainer, it is expected that offering a compensation will have a different effect on the reader.
2.1.4 Effect of type of managerial response on credibility
Coombs & Holladay (2008) evaluated an apology to other accommodative response
strategies such as sympathy and compensation, to determine whether an apology is the best strategy. They found that all three response strategies scored equally on post-crisis reputation evaluations, supporting that an apology is not the best strategy. Coombs & Schmidt (2000) found that an organization accepting responsibility for a wrongdoing or showing empathy is evaluated more favorably by the public. Thereby, organizations expressing concern in their responses are perceived more favorably (Dean, 2004). DiStaso et al. (2014) states that no significant differences were found in the perceived credibility of an apology compared to a compensation. Other studies also showed that if an organization quickly self-disclose negative information about itself, it will result in overall enhanced credibility and message persuasiveness (DiStaso et al., 2014). Scholars have demonstrated that an apology is more effective than the response strategy of denial and better protects an organization’s reputation when held responsible (Kim, Avery & Lariscy, 2009).
Based on the previous discussion and outcomes of the three different type of managerial response, we assume the following hypothesis:
2.2 Level of personalization
Companies can choose to respond in a standardized manner to online customer reviews, by providing the same response to every review being addressed towards them. A standardized response sounds generic does not provide a personalized acknowledgment of the reviewer or the actual complaint being addressed by the consumer. Dens et al. (2015) introduces the standard response strategy, in which service providers might not respond explicitly to
(negative) online reviews. Other companies might not reply because they are not aware of the reviews or lack the necessary resources to do so (Dens et al., 2015). A standardized response does not purely influence the (dis)satisfied customer, posting the online review, but it also exerts an influence on the bystanders reading the response. Although a standardized response indicates some interest by the service provider, it could also be perceived as a lack of effort to respond towards customers or recover a service failure (Dens et al., 2015). Dens et al. (2015) state that more research is needed regarding the complex evaluation processes of consumers. How do current customers and bystanders perceive a standardized response? Less research focused on the effect of a standardized response on the observer of a service failure. Similarly, to the complainant, an observer could perceive responding in a standardized manner as a lack of effort. However, Dens et al. (2015) found that more effort by the service provider is not necessarily provided as more positive. Thereby, generic responses could signal to some extent the level of service quality of firms.
Besides using a standardized response, companies can choose to respond to negative online customer reviews with a personal approach. Personalization covers a variety of
concepts (Vesanen, 2007). However, the basic premise behind the term is to create a message referring to a receiver’s self (Petty, Wheeler & Bizer 2000). The theory of self is a theoretical perspective that can be used to explain the effectiveness of personalization (Van den Putte, Smit & Maslowska, 2011). Personalized communication consists of information recognizable as personal, such as name, referring to an individual’s self. According to Cherry (1953), people are cognitively sensitive to such information. Thereby, individuals are expected to pay more attention to and better memorize personalized communication (Rogers, Kuiper & Kirker 1997).
(2015) state that these responses should be personalized. Management should not only provide an empathetic response, it should also signal that the service provider reads the complaint thoroughly. This could be shown by (partly) paraphrasing the complaint. Min et al. (2015) found that respondents were more satisfied with a response that paraphrased the complaint, compared to a response that did not do so. This finding is supported by research stating that active listening is highly important for handling complaining customers in person (Drollinger, Corner, & Warrington 2006). However, applying active listening in response to online reviews is difficult, since the nonverbal cues of active listening cannot be used. Therefore, it is highly important to use paraphrasing. Paraphrasing the negative complaint in the review, enables organizations to show that they take customer complaints seriously. This is shown to both the complainer and the reviewer. Instead of repeatedly duplicating the generic response, the managerial response should include a personalized acknowledgment of the reviewer and a paraphrase of the actual complaint (Min et al., 2015).
Based on literature, within this research a personalized response is defined as paraphrasing the primary complaint in the negative review. It will be tested whether this managerial response is perceived as more credible by observers.
Based on the previous discussion, we assume the following hypothesis:
H2: The level of personalization will positively enhance the effect of all type of managerial responses on the perceived level of credibility.
2.3 Covariates
As stated in the previous section, how consumers perceive the credibility of types of
2.3.1 Skepticism towards organizations
Personal characteristics of respondents might be an influencing factor. A consumer
characteristic that can possibly influence the tendency to perceive a managerial response as credible is skepticism. Skarmeas & Leonidou (2013) define skepticism as the tendency to disbelieve, doubt, and question information from others. A person’s skepticism might exert an influence on how a managerial response is evaluated. Two forms of skepticism can be defined: skepticism towards others and skepticism towards organizations. Within this research, skepticism towards organizations will be used as a control variable, since the response is provided by the organization and not the customer. Dellarocas (2006) state that participation of the organization leads to skepticism. Consumers being skeptic towards organizations might have the tendency to disbelieve or doubt any information provided by companies. Therefore, credibility of a managerial response might be lower for customers that are skeptic towards organization. Following, less skeptical consumers, might perceive a managerial response as more credible.
We assume the following hypothesis:
H3: Consumers being skeptic towards organizations are less likely to perceive a managerial response as credible.
2.3.2 Need for cognition and credibility
A consumer characteristic that might influence the effect of level of personalization on the perceived credibility of a managerial response is needed for cognition. As mentioned in the previous paragraph, personalized communication consists of information recognizable as personal, such as name, referring to an individual’s self. According to Cherry (1953), people are cognitively sensitive to such information. Xia & Bechwati (2008) identified the role of cognitive personalization in online consumer reviews and state that the cognitive
We assume the following hypothesis:
H4: Individuals with a high need for cognition will perceive a personalized response as more credible compared to individuals with a low need for cognition.
2.3.3 Attribution of responsibility
As a complaint is the expression of dissatisfaction about a perceived misconduct by an
organization, the complainant perceives the organization as partly responsible for the problem they complain about. Responsibility refers to the party that is responsible for the failure which could be the consumer, the company or some other third party. In the case of high perception of responsibility, Coombs (2007) and Coombs & Holladay (2008) argue in favor of more accommodative response strategies. A full apology, compensation or corrective action is recommended when the perception of responsibility is high. If the perceived responsibility is low, more defensive strategies are recommended, which includes denial or refutation. Within this research, attribution of responsibility can be defined as the degree to which the reader of a customer review attributes the responsibility to the service provider.
The following hypothesis will be proposed:
2.4 Conceptual model and hypotheses
Based on the theories discussed, the conceptual model depicted in figure 1 is drawn. The two independent variables are type of managerial response and level of personalization.
The control variables skepticism towards organizations and attribution of responsibility are both expected to influence the effect of the type of managerial response on credibility. Need for cognition is expected to influence the effect of level of personalization on credibility. The following chapter will outline the methodology and measures used in this research.
Hypothesis 1 A managerial response containing an apology or compensation are perceived as more credible compared to a refutation
Hypothesis 2 The level of personalization will positively enhance the effect of all type of managerial responses on the perceived level of credibility
Hypothesis 3 Consumers being skeptic towards organizations are less likely to perceive a managerial response as credible
Hypothesis 4 Individuals with a high need for cognition will perceive a personalized response as more credible compared to individuals with a low need for cognition
Hypothesis 5 If attribution of responsibility is high, managerial responses containing an apology or compensation are perceived as more credible compared to a refutation
METHODOLOGY
This chapter outlines the research design of this study. Firstly, the performed experimental research design is discussed, followed by the data collection procedure. Thereby, the operationalization of all variables used in this research and plan of analysis are explained. An overview of the sample is given, by describing their demographics, and lastly the results of the manipulation checks are provided.
3.1 Research design
To determine the relationship between the variables, causal research in the form of an experiment was performed (Field, 2009). Experimental research design is defined as the determination of a causal relation between a defined number of variables, where one or more are manipulated and other variables are added (Malhotra, 2007). The independent variables used in this research are type of managerial response and level of personalization, the dependent variable is credibility. The independent variables are manipulated by the
researcher. Therefore, each respondent will see a different managerial response, containing either an apology, compensation or refutation. Moreover, these responses vary in their level of personalization in which the response is either personalized or standardized. This results in an experiment consisting of a 3x2 between-subjects design, with six different conditions (see Table 2) (Field, 2009). The between-subjects design is chosen to prevent participants from comparing different type of reviews with each other, which might influence their judgment (Malhotra, 2007). Respondents in all conditions have to evaluate the managerial response on the perceived credibility. Each conditions requires a minimum of 30 respondents to be perceived as reliable (Malhotra, 2009). Therefore, in this research, a total of 210 respondents are needed. The assignment of the respondent to the different conditions is based on
randomization.
Managerial response
Apology Refutation Compensation
3.2 Survey
To obtain the required data, an online questionnaire was created using the professional survey design software Qualtrics. The negative online customer review is equally shown among all respondents, where after a different managerial response is provided, dependent on the condition they are assigned to. All respondent will answer the same questions to measure the perceived credibility of the managerial response on a certain scale. The survey consist of closed-ended questions, measured by a 7-point-Likert-scale, which is commonly used within the research field. Thereby, close-ended questions are suitable for comparison. The survey was administered online, to reach as many respondents as possible and to let the stimulus look as realistic as possible.
3.3 Procedure
Participants were approached via e-mail and different social media channels (Facebook and LinkedIn), with a request to participate in an online study regarding online customer reviews. By opening a link, participants are redirected to the survey. The respondent is first presented with a message explaining the reason for taking part in this study, the estimated time it will take and a reassurance of the anonymity. After that, the respondent will see a short
introductory text in which they have to imagine themselves browsing on the Internet to find a suitable hotel on the review website TripAdvisor. Hereafter, a negative online customer review is shown in which a customer states their dissatisfaction with an overnight stay in Hotel Sea View, published on TripAdvisor. The participant is then asked to what extent the hotel is kept responsible for the negative event described in the review. After this, the respondent is randomly presented with one of the six possible managerial responses. Based on this response, the respondent is asked to answer questions regarding the credibility of that managerial response.
Finally, respondents were asked questions related to the frequency of use of review websites and their earlier experience with the platform TripAdvisor. The survey is concluded with some classification questions to collect information on the respondent, including gender, age and occupation. These questions are presented at the end rather than at the beginning of the survey, as classification questions can be perceived as asking for sensitive information. By starting the survey with these questions, respondents might not complete the rest of the survey (Malhotra & Birks, 2007). Lastly, participants will be kindly thanked for their participation.
3.4 Operationalization
3.4.1 Manipulations
The independent variables in this research are manipulated by the researcher, based on literature. Firstly, the independent variable type of managerial response is manipulated, in which either an apology, refutation or compensation is shown to respondents. Within the apologetic response, an apology will be made towards the customer in which the organization expresses their sincere regret and promise to avoid repeating the same wrongful act
(Kellerman, 2006). The refutable response will refute the arguments provided in the online customer review, by claiming they are untrue and denying responsibility for the failure (Benoit, 1995; Coombs, 1998). Lastly, the compensational response will provide customers with an explanation for a service failure and compensate them in some way (Bitner, 1990).
The negative online customer review is equally shown among all conditions. The consumer expresses a negative experience with the service provider, in which they state that the quality of the beds were not as expected. The failure described in the negative online customer review could be attributed to the service provider, as attribution of responsibility will be considered in this research. An overview is provided in table 3.
Negative online customer review
’Last week, we had an overnight stay at Hotel Sea View. Unfortunately, the beds did not meet our expectations. They are extremely uncomfortable, resulting in a bad night of sleep. These beds need to be renewed!’
Managerial response
Standardized Personalized
Dear guest,
Thank you for your review. We regret to be informed that your stay did not meet your expectations.
Dear guest,
Thank you for your review. We regret to be informed that your stay did not meet your expectations, due to the fact that you experienced our beds are not comfortable. Managerial response - Apology
Standardized Personalized
We sincerely apologize for any
inconvenience you have experienced during your visit. Your feedback will be forwarded to the relevant department.
We sincerely apologize for any
inconvenience you have experienced during your visit. Currently, we are implementing major reforms in the hotel. Therefore, all beds will be replaced within soon. We truly hope to welcome you again in our hotel after the renovation.
Managerial response - Refutation
Standardized Personalized
Of course, people may have different experiences and preferences. We are highly committed to provide you a pleasant stay at
our hotel. However, we rarely receive any complaints.
highly committed to provide you a pleasant stay at our hotel. However, we rarely receive complaints about the quality of our beds.
Managerial response - Compensation
Standardized Personalized
We are highly committed to provide you a pleasant stay at our hotel. Of course, we hope that you want to visit our hotel again. Therefore we would like to offer you a free drink of choice on your next visit at Hotel Sea View. We would like to welcome you anytime soon.
We are highly committed to provide you a pleasant stay at our hotel. During your stay, we noticed that you enjoyed diner in our restaurant. Therefore we would like to offer you a free drink of choice in our restaurant on your next visit at Hotel Sea View. We would like to welcome you anytime soon. Table 3: Negative online customer review and managerial responses
3.4 Operationalized
3.5.2 Variables and scales
Concept Item Scale α Dependent variable Perceived credibility Tate et al. (2006) 1.Source credibility Very trustworthy Very credible Very honest Very questionable Very conclusive 2.Message credibility Very believable Very authentic Very reasonable Very convincing 7-point Semantic differential 1 = Strongly disagree 7 = Strongly agree .884
Independent variable - manipulation check Type of managerial
response
Coombs (2008)
1 Hotel Sea View accepted responsibility for the incident
2 Hotel Sea View provided the guest with a compensation or solution for his/her complaint
3 Hotel Sea View denied the negative experience 7-point Semantic differential 1 = Strongly disagree 7 = Strongly agree Level of personalization
Mittal & Lassar (1996)
1 The management of Hotel Sea View seems polite and courteous
2 The management of Hotel Sea View takes time to respond personally 3 The management of Hotel Sea View seems friendly and pleasant
4 The management of Hotel Sea View takes time to know their
7-point Semantic differential
1 = Strongly disagree 7 = Strongly agree
Covariates
Skepticism towards organizations
Skarmeas & Leonidou (2013)
1 Organizations generally act socially responsible
2 Organizations are generally
concerned to improve the well-being of consumers.
3 Organizations generally follow high ethical standards 7-point Semantic differential 1 = Strongly disagree 7 = Strongly agree .775
Need for cognition
Trilling &
Schoenbach (2012)
1 I prefer complex over simple problems
2 I like to have the responsibility of handling a situation that requires a lot of thinking
3 I would prefer a task that is intellectual, difficult, and more important to one that is somewhat important but does not require much thought 7-point Semantic differential 1 = Strongly disagree 7 = Strongly agree .760 Attribution of responsibility Lee (2004)
1 To what degree should the hotel be blamed for the negative experience? 2 How much responsibility should the hotel bear for the negative experience?
7-point Semantic differential 1 = Not at all 7 = Totally .802 Control variables
Type of argument 1 I perceived the argument contained in the customer review as
Perceived credibility
Credibility is measured by using the scale measuring source credibility and message credibility by Tate, Alexander & Maheshwari (2006) which consist of two different constructs. The scale measuring source credibility consists of five items using a 7-point semantic differential. Factor analysis showed that four items measure the same construct. One item measuring ‘unquestionable – questionable’load onto a second construct. The reliability analysis on the first construct shows that the scale measuring source credibility is reliable (α = .805) with four items. Secondly, message credibility is measured by four items, using a 7 point semantic differential. Factor analysis showed that two items, measuring ‘unbelievable - believable’ and ‘unauthentic - authentic’ have communalities below 0.5 (.321 and .491), which is a prerequisite. Therefore, factor analysis is performed on all nine items measuring source and content credibility. After removing the item measuring ‘unquestionable - questionable’, which measures a different construct, and the item ‘authentic - unauthentic’, which has not the required communalities, all items clearly measure one construct. Reliability analysis was performed and the scale is proven to be reliable (α = .764). It is stated by
literature (Metzger, Flanagin, Eyal, Lemus & McCann, 2003) that source and message credibility are perceived as overlapping concepts in literature. Therefore, source and message credibility will be combined into one construct, named ‘perceived credibility’.
Level of personalization
The independent variable, level of personalization, can be divided into two levels: a personalized or a standardized response. The scale measuring level of personalization by Mittal & Lassar (1996) is used, which consists of four items. A high score indicates that the managerial response is perceived as personalized. Factor analysis showed that all items clearly load to one construct. Reliability analysis was performed and the scale is reliable (α = .764).
Skepticism towards organizations
organizations while a low score indicates a high level of skepticism. Factor analysis showed that the items clearly load onto one construct. Additionally, reliability analysis was performed and this shows that the scale has a high reliability (α = .774).
Need for cognition
Need for cognition will be measured by using a scale from Trilling & Schoenbach (2012), based on the original 34-item NFC scale of Cacioppo & Petty (1982). The scale of Trilling & Schoenbach (2012) consist of three items.A high score on the scale indicates a high need for cognition while a low score indicates a low need for cognition. Factor analysis showed that the items clearly load onto one construct. Additionally, reliability analysis was performed and this shows that the scale has a high reliability (α = .760).
Attribution of responsibility
Attribution of responsibility will be measured by using a scale from Lee (2004), which is originally used to measure perceived organizational responsibility for a crisis. It is measured by two items. Factor analysis showed that the two items measure one construct and that the reliability of the scale is high (α = .802).
Control variables
The results were controlled for type of argument, to measure to which extent the negative online customer review was perceived as negative, neutral or positive. Thereby, frequency of use of online review websites is measured with three separate questions. Lastly, the
demographic variables gender, age and occupation are one-item objective measures and thus do not require an additional factor analysis.
3.5 Sample
A total of 298 respondents filled out the survey. Due to incomplete data of some respondents, 67 surveys were excluded. Each condition received the minimum requirement of 30
a full-time job and 40 part-time. A total of 10 respondents filled out that they currently have no occupation. On average, respondents booked between two to five times a holiday via the Internet over the past two years (M = 3.11). When booking a hotel on the internet, they made frequent use of online review websites to read OCRs (M = 4.80). Thereby, they were on average familiar with the online review website TripAdvisor (M = 5.11). The type of argument ( 1 = ‘very negative’ - 7 = ‘very positive’) was perceived as moderately negative (M = 3.42).
Sample descriptive per condition Condition Mean
age
Gender Occupation N
Male Female Student Fulltime job Parttime job No occupation 1 33 46,15% 53,85% 41,0% 33,3% 23,1% 2,6% 39 2 32 50% 50% 32,5% 45,0% 20,0% 2,5% 40 3 33 24,39% 75,61% 41,5% 29,3% 24,4% 4,9% 41 4 30 34,88% 65,12% 55,8% 32,6% 9,3% 2,3% 43 5 28 40,63% 59,38% 43,8% 34,4% 15,6% 6,3% 32 6 32 38,89% 61,11% 36,1% 33,3% 22,2% 8,3% 36 Total 31 61.40% 38.60% 42,0% 34,6% 19,0% 4,3% 231
Table 5: Overview demographics per condition
Overall sample descriptive
Mean Frequency online hotel
booking
3.11
Frequency of use online review website
4.80
Familiarity TripAdvisor 5.11
Type of argument 3.42
3.7 Randomization check
A randomization check has been performed to get an overview whether there is systematic difference between respondents in the separate conditions. Some analyses have been conducted, where the independent variable was the condition and the dependent variable gender, age or occupation. The results of the Chi-square tests (p=.228) (p=.757) and the one-way ANOVA for age (F (5)=.707, p>.05) show that all outcomes are insignificant (see Table 7). This means that there are no significant differences between the groups. Therefore, the assignment of the participants to a condition in a random way was successful (see Appendix D).
Variable Significance Df
Gender .228 5
Age .555 36
Occupation .757 15
Table 7: Randomization check
3.8 Plan of analysis
Data will be analyzed by making use of the statistical analysis program Statistical Software Package for Social Science (SPSS). The collected data will be adjusted and prepared for further analysis. To test the hypotheses, different analysis methods will be used.
3.8.1 Data preparation
3.8.2 Analyses
First, an analysis of variance is performed to obtain an overview of the first results of the research. The proportionate variance caused by the different levels of independent variables in the measured variables will be examined, and to see if there are any significant interaction effects. Hereafter, multiple regression analysis will be used to test the effect of the two independent variables on the dependent variable. The independent variables consists of multiple categories, therefore these variables have been coded into dummy variables to carry out subsequent analyses. Due to the fact that type of managerial response consist of three categories, refutation will be chosen as the base in this model.
The first model consist of the dependent variable credibility, the two independent variables and the interaction effects. Two interactions terms are included: type of managerial response (compensation)*level of personalization and type of managerial response
(apology)*level of personalization. The regression equation for the model including the interaction effects is Y=b0 + b1X1 + b2X2 + b3X1X2. In the second model, the covariate scepticism towards organizations is added, as well as the interaction effects of scepticism towards others with type of managerial response. In the third model, the covariate need for cognition is added, together with the interaction effect of need for cognition and level of personalization. In the fourth model, the covariate attribution of responsibility is added, as well as the interaction effect of attribution of responsibility and type of managerial response. Eventually, the full model is created with all independent variables, the dependent variable, the interactions effects and all covariates.
3.9 Results manipulation checks
The two independent variables are both manipulated by the researcher. A manipulation check was executed to examine whether the independent variables were perceived by the
respondents as intended. The mean scores are computed and several tests were conducted to analyze whether there were significant differences between the conditions. The results show that the manipulation of the two independent variables worked as intended. Next, the results per manipulation will be discussed. The output of the statistical tests can be found in
3.9.1 Type of managerial response
To check whether the manipulation of the first independent variable worked as intended, participants were asked to evaluate the type of managerial response on three 7-point Likert scale items (1=’strongly disagree’, 7= ‘strongly agree’). For the first item measuring apology, a high score means that the response is perceived by the respondent as an apology. As the mean score for apology is above the mean scores for refutation and compensation, this shows that respondents in the apologetic condition perceived the response as intended. This
difference is significant (p < .05), although compensation received a slightly smaller mean. This result could be explained by the fact that a compensation is perceived as type of apology. For the items measuring refutation and compensation, the means of the conditions are significantly different compared to the other two conditions. Overall, this implies that the manipulation has worked. To test the manipulation, a one-way ANOVA has been conducted (Appendix E).
Apology Refutation Compensation P-value
Apology 4.65 3.43 4.31 .000
Refutation 2.24 4.21 2.71 .000
Compensation 2.62 1.99 4.37 .000
Table 8: Manipulation type of managerial response
3.9.2 Level of personalization
To check whether the manipulation of the second independent variable worked as intended, participants were asked to evaluate level of personalization on four 7-point Likert scale items (1=’strongly disagree’, 7=’strongly agree’). A high score on the scale means that the
managerial response was perceived as personalized compared to standardized. An
independent t-test was performed and this showed that the groups are significantly different from each other (p < .05). It can be concluded that the manipulation worked (Appendix E).
Condition Mean SD P-value
Standardized 4.6161 1.13201 .006
Personalized 5.0169 1.05750
RESULTS
The results of the research will be discussed in this chapter. First, the outcome of the analysis of variance and test of simple effects are discussed to examine the relationship between the variables. Hereafter, the results of the regression analysis are discussed in which, to conclude, findings are used to discuss the separate hypotheses provided in chapter two.
4.1 Analysis of variance
Firstly, to gain an understanding of the relationship between the variables, a two-way ANOVA test was conducted with the two independent variables and credibility as the dependent variable. An overview of the results of the ANOVA can be seen in table 10. The full output is provided in Appendix F.
4.1.1 Main effects
The results of the ANOVA-tests show that the independent variable type of managerial response does not significantly influence credibility (p=.728). Similar results are found for the second independent variable, level of personalization (p=.147). This means that having a different managerial response, which could either be an apology, compensation or refutation, does not significantly influence the perceived level of credibility of that particular managerial response. The means of each type of managerial response show that offering a compensation resulted in the highest overall credibility (M = 4.45), with a refutable response receiving the lowest levels of credibility (M = 4.32). A managerial response with an apology received an overall credibility with a mean of 4.38. Thereby, offering either a personalized or
DV = Credibility
Type of managerial response
Condition Apology Refutation Compensation Total
Standardized 4.03 (.913) 4.32 (.986) 4.51 (1.016) 4.28 (.981) Personalized 4.72 (1.095) 4.33 (.1.008) 4.40 (1.073) 4.48 (1.063) Total 4.38 (1.061) 4.32 (.991) 4.45 (1.040) p-value type of management response .728 p-value level of personalization .147 p-value MR * pers .034
Table 10: Main results ANOVA credibility
4.1.2 Interaction effect
Figure 2: Plots interaction effect
4.1.3 Test of Simple Effects
Managerial response
Apology Refutation Compensation
Standardized 4.033 (.163) 4.342 (.159) 4.509 (.180) Personalized 4.721 (.161) 4.509 (.180) 4.405 (.169)
P-value .003 .994 .673
Based on estimated marginal means
* The mean difference is significant at the 0.05 level Table 11: Results of Test of Simple Effects
4.2 Regression analysis
Multiple regression analysis has been performed to see how the independent variables influence the dependent variable. The direct effects of the independent variables on credibility were tested first, in combination with the interaction effects between the
independent variables. Then, the control variables skepticism towards organizations, need for cognition and attribution of responsibility were added to test their main effects and
interaction effects. Finally, the full model was tested with all independent variables, control variables and the dependent variable. The output of the regression analysis can be found in Appendix G. The results in table 12 show that adding the interaction terms and the control variables result in an increase in the adjusted R square.
4.2.1 Regression model with the two independent variables
4.2.2 Regression model with skepticism towards organizations
The control variable skepticism towards organizations is added in the second model. Results in table 12 show a positive significant main effect of skepticism towards organizations on credibility. This means that the less skeptic someone is towards organizations, the more a managerial response is perceived as credible. The other way around, this shows that if someone is skeptic towards organizations, managerial responses are perceived as less credible. The interaction effect of skepticism towards organizations and compensation, as well as the interaction effect of skepticism towards organizations and apology, have no significant effect on the credibility of a managerial response compared to the interaction effect of the base (refutation). This means that skepticism towards organizations is not significantly different per type of managerial response, in turn affecting the credibility of a response.
4.2.3 Regression model with need for cognition
The control variable need for cognition is added in the third model. Results in table 12 show that there is a negative significant main effect of level of personalization on credibility, meaning that having a less personalized response results in lower levels of credibility. However, there is also a positive significant interaction effect of need for cognition and level of personalization on credibility. This means that the higher one’s need for cognition, the stronger the effect of level of personalization on credibility. Thus, the higher one’s need for cognition, the higher the credibility of a personalized response. Thereby, the results in table 12 show a positive significant interaction effect of managerial response with an apology and level of personalization (p=.073). In comparison with the base, this implies that offering a personalized apology instead of a personalized refutation result in a significant increase in perceived credibility.
4.2.4 Regression model with attribution of responsibility
credibility. Within this research, the attribution of responsibility reflects the degree to which the hotel should be kept responsible for the customer’s negative experience in the online customer review. This significant interaction effect shows that higher levels of attribution of responsibility, results in lower credibility if management responds with a compensation. The positive effect of a compensation decreases if the attribution of responsibility increases.
4.2.5 Full regression model
In the fifth model, all independent variables, the dependent variable, the interaction effects and all covariates are added, of which the results are shown in table 12. This shows that there is a significant effect of a personalized apologetic managerial response on credibility,
compared to a personalized refutable response, at a confidence level of p<.05. This implies that offering a personalized apology results in higher perceived credibility. Thereby,
skepticism towards organizations has a significant direct effect on credibility, at a confidence level of p<.05. Laslty, the interaction effect of need for cognition and level of personalization on credibility is found to be significant. This means that need for cognition as a consumer characteristic influences the effect of level of personalization on credibility.
DV = Credibility
Model 1 2 3 4 5
Constant 4.293** 3.022** 4.135** 4.142** 2.757**
IVIa = type of managerial response (Compensation)
.096 -.092 .088 .640** .530
IV1b = type of managerial response (Apology)
-.101 -.118 -.105 .010 .050
IV2 = Level of personalization .016 -.048 -.681** .011 -.711* IV1_a * IV2
Managerial Response (Compensation)* Level of personalization
-.048 .003 -.036 -.036 .021
IV1_b * IV2
Managerial Response (Apology)* Level of personalization
Control variable 1
Skepticism towards organizations
.312** .324**
Skepticism towards organizations* Managerial Response (Compensation)
.163 -.020
Skepticism towards organizations* Managerial Response (Apology)
-.028 -.122
Control variable 2 Need for cognition
.381 .005
Need for cognition * Level of personalization .730** .690** Control variable 3 Attribution of responsibility .038 .049 Attribution of responsibility *
Managerial Response (Compensation)
-.575* -.470
Attribution of responsibility *
Managerial Response (Compensation)
-.114 -.086
R2 .039 .143 .096 .061 .199
R2 adjusted .017 .112 .067 .032 .151
F-statistic 1.817 4.613 3.367 1.819
Multicollinearity No Yes Yes Yes Yes
Table 12: Regression model with DV = Credibility ** = p < .05
* = p < .10
4.3 Multicollinearity
Within the first model, which includes the main and interaction effects of the two independent variables and the dependent variable, no multicollinearity occurred as all VIF scores were below 10. (Appendix G). When adding the control variable skepticism towards organizations, multicollinearity occurs with the predictor type of managerial response (VIF = 17.503). The same accounts for adding the control variable need for cognition, in which multicollinearity occurs with the predictor level of personalization. (VIF = 24.511). Lastly, the control variable attribution of responsibility has multicollinearity with the predictor type of managerial response (VIF = 13.239). All multicollinearity occurs due to overlapping constructs of the control variables with one of the independent variables. Therefore, results should be interpreted with caution.
4.4 Discussion of hypotheses
A total of five hypotheses were introduced at the end of chapter two. The hypotheses were tested by making use of statistical tests in SPSS, as discussed in chapter three. In the previous section, the results of the ANOVA and multiple regression analysis were presented. Based on the outcome of this, the proposed hypotheses will either be accepted or rejected. The
influence of the coefficients is assumed to be significant when the coefficients have
significance levels of ten percent or lower. The most common used significance level is five percent, but significant levels of ten percent are also appropriate (Hair et al., 2010).
4.4.1 Main effect of type of managerial response on credibility
A managerial response with an apology or compensation will be perceived as more credible compared to a managerial response with a refutation.
4.4.2 Effect of level of personalisation on credibility
The level of personalization will positively enhance the effect of all type of managerial responses on the perceived level of credibility.
To test whether level of personalization positively enhance the effect of all type of managerial responses on the perceived level of credibility, a regression analysis was
performed (Appendix G). This shows that the effect of a personalized apology is significant (t = 1.919, p = .056). The Beta-value (.226) shows a positive relationship. This means that if management responds towards a negative online customer review with a personalized apology, compared to a personalized refutation which is chosen to be the base in this model, the apologetic response is perceived as significantly more credible. The effect of personalized compensation was not significant (t = -.409, p = .683). This means that if management
responds towards a negative online customer with either a personalized compensation or personalized refutation, does not result in a significant difference in perceived credibility of that managerial response. This means that the hypothesis cannot be accepted.
4.4.3 Effect of skepticism towards organizations on credibility
Consumers being skeptic towards organizations are less likely to perceive a managerial response as credible.
To test the effect of the covariate skepticism towards organizations on credibility, a regression analysis was performed (Appendix G). This shows that skepticism towards organizations has a significant effect on credibility (t = 3.141, p = .002). The Beta-value (.312) shows a positive relationship. Therefore, skepticism towards organizations as a
4.4.4 Effect of need for cognition on level of personalization
Individuals with a high need for cognition will perceive the personalized response as more credible compared to individuals with a low need for cognition.
A regression analysis was performed to test the effect of need for cognition on level of personalization (Appendix D). Need for cognition has no significant influence on credibility (t = .381, p = .704). However, need for cognition significantly influences the effect of level of personalization on credibility (t = 2.314, p = .022). The Beta-value (.730) shows a positive relationship. This means that having a high need for cognition enhances the effect of level of personalization on credibility, in which a personalized managerial response will be perceived as more credible. Therefore, the hypothesis can be accepted.
4.4.5 Main effect of attribution of responsibility on credibility / interaction effect of attribution of responsibility on type of managerial response and credibility
If attribution of responsibility is high, managerial responses containing an apology or compensation are perceived as more credible compared to a refutation.
A regression analysis was performed to test the effect of attribution of responsibility on credibility (Appendix D). Attribution of responsibility has no significant influence on
4.5 Hypotheses
Hypothesis Result
H1: A managerial response with an apology or compensation will be perceived as more credible compared to a managerial response with a refutation.
Rejected
H2: The level of personalization will positively enhance the effect of all type of managerial responses on the perceived level of credibility.
Rejected
H3: Consumers being skeptic towards organizations are less likely to perceive a managerial response as credible.
Accepted
H4: Individuals with a high need for cognition will perceive the
personalized response as more credible compared to individuals with a low need for cognition.
Accepted
H5: If attribution of responsibility is high, managerial responses containing an apology or compensation are perceived as more credible compared to a refutation.
