How do you motivate consumers to write online reviews?
A study examining the effects of different types of incentives and the certainty of
receiving these incentives on consumers’ eWOM behavior
By
M.C. (MARNIX) TEENSMA
“A Master Thesis submitted to the faculty of Economics and Business at the University of Groningen in partial fulfilment of the requirements for the degree of MSc. Marketing
Management.’’
How do you motivate consumers to write online reviews?
A study examining the effects of different types of incentives and the certainty of
receiving these incentives on consumers’ eWOM behavior
By M.C. (MARNIX) TEENSMA Peizerweg 53B 9726 JD Groningen (06) 40214034 m.c.teensma@student.rug.nl student number S3271498 June 2018 University of Groningen Faculty of Economics and Business
MSc. Marketing Management
ABSTRACT
Electronic word of mouth (eWOM) is the most prevalent and prolific source of online product information to date. Yet, while nearly all consumers use eWOM in their decision-making process, far fewer actually generate eWOM. Marketers are still unsure how to effectively motivate consumers to generate eWOM content. This study takes a consumer-oriented focus on rewarding consumers, with both monetary and non-monetary incentives, to write online reviews. This study also examines the effect of the certainty by which these incentives are received by consumers. A digital survey was conducted among 331 consumers using an experiment of statistical design. The results show that offering consumers certain money rewards, leads to a significantly higher intention to write online reviews compared to other incentives. The findings in this study challenges the status quo and can help marketers to develop a more proactive marketing program in order to increase consumers’ eWOM participation and, thus, increasing sales.
Keywords: Electronic Word-of-Mouth (eWOM), Online Consumer Reviews (OCRs), Online
Consumer Behavior, Incentives, Self-sufficiency, Uncertainty, Reward Sensitivity.
Research Topic: The When and How of eWOM, and More Specifically: OCRs
Seminar supervisor: Dr. J. A. (Liane) Voerman, Department of Marketing, University of
TABLE OF CONTENTS
1.
INTRODUCTION ... 1
1.1 INTENTIONTOWRITEANONLINEREVIEW ... 2
1.2 USEOFINCENTIVES ... 2
1.2.1 Monetary incentives ... 3
1.2.2 Non-monetary incentives ... 3
1.3 CERTAINANDUNCERTAININCENTIVES ... 4
1.4 PROBLEMSTATEMENT ... 4
1.5 RESEARCHQUESTIONS ... 5
1.6 ACADEMICANDMANAGERIALRELEVANCE ... 5
1.7 STRUCTUREOFTHETHESIS ... 5
2.
THEORETICAL FRAMEWORK ... 6
2.1 TYPEOFINCENTIVE ... 6
2.2 CERTAINTYOFTHEINCENTIVE ... 7
2.3 REWARDSENSITIVITY... 8
2.4 CONTROLVARIABLES ... 8
2.5 CONCEPTUALMODEL ... 9
3.
RESEARCH DESIGN ... 10
3.1 TYPEOFRESEARCH ... 10
3.2 PARTICIPANTANDDESIGN ... 10
3.3 PRELIMINARYTEST ... 11
3.4 SURVEYDESIGN ... 11
3.4.1 Scenario ... 11
3.4.2 Choice for incentives ... 12
3.4.3 Magnitude of the incentives ... 13
3.5 OPERATIONALIZATION... 14
3.5.1 Dependent variable measurement ... 16
3.5.2 Manipulation check: perceived uncertainty ... 16
3.5.3 Extraneous variables ... 17
3.6 POPULATIONANDSAMPLINGMETHOD ... 18
3.7 PLANOFANALYSIS ... 18 3.7.1 Factor analysis ... 18 3.7.2 Reliability analysis ... 19 3.7.3 Analysis of variance ... 20 3.7.4 Regression analyses... 20 3.7.5 Mean centering ... 20 3.7.6 Dummy coding ... 21 3.7.7 Multicollinearity ... 21
3.8 DESCRIPTIVESTATISTICSOFTHESAMPLE ... 21
3.8.1 Data preparation ... 21
4.
RESULTS ... 23
4.1 ANALYSISOFVARIANCE ... 23
4.2 REGRESSIONANALYSIS ... 25
4.3 HYPOTHESESTESTING... 28
5.
DISCUSSION ... 30
5.1 DISCUSSIONOFTHERESULTS ... 30
5.2 MANAGERIALANDACADEMICIMPLICATIONS ... 32
5.3 LIMITATIONS ... 34
5.4 RECOMMENDATIONSFORFUTURERESEARCH... 35
REFERENCES ... 36
APPENDICES ... 40
APPENDIXA:SURVEY ... 40
APPENDIXB:PRELIMINARYTESTRESULTS ... 45
APPENDIXC:FACTOR&RELIABILITYANALYSIS ... 46
APPENDIXD:ANOVARESULTS ... 50
APPENDIXE:SPSSOUTPUTMODEL1-4 ... 52
- SPSS output - model 1 ... 52
- SPSS output - model 2 ... 53
- SPSS output - model 3 ... 54
1
1. INTRODUCTION
Consumers provide information about goods, services, brands, or companies to inform and advice their peers (Babić, Sotgiu, De Valck, & Bijmolt, 2016). This phenomena is known as word of mouth (WOM) and has been called “the world's most effective, yet least understood marketing strategy” (Misner, 1999). Word of mouth plays a major role in consumers’ buying decisions (Richins & Root-Shaffer, 1988), as it helps consumers to minimize uncertainty about a product or service and, thus, helps them to choose the best possible offering available (Dichter, 1966; Roselius, 1971). The underlying reason for this is that customer-to-customer communication is perceived as much more credible and persuasive, compared to traditional marketing (Trusov et al., 2009). With the rise of the internet, many new options have emerged for consumers to gather product information from other consumers and also to inform co-consumers on products and services, also known as Electronic Word of Mouth (Babic et al., 2016). Electronic word of mouth (eWOM) is defined as the most prevalent and prolific source of online product information to date (Fennis & Stroebe, 2016). Electronic word of mouth behavior is a broad concept and can be defined as any word of mouth which is communicated through the use of the internet (Babic et al., 2016), e.g. through the use of tweets, blog posts, reviews, “likes,” “pins,” images, video testimonials and more. Yet, while nearly all shoppers use eWOM in their decision-process, far fewer actually engage in eWOM behavior (Hennig-Thurau, 2004). This fact makes it a key task for marketers to gain knowledge into the main drivers and barriers for consumers to engage in eWOM behavior.
2
application of experimental psychology and behavioral psychology to the discipline of human decision-making, including economic decision-making (Schweyer, 2017). Behavioral economics (BE) is described in the academic term behind what many practitioners describe as “nudges” (Schweyer, 2017). A ‘nudge’ is essentially a means of encouraging or guiding behavior, but without mandating or instructing. And while these nudges can be used to encourage people to make choices that are not in their best interests, this paper is about good-intentioned nudges, i.e. for consumers to express themselves on the internet to help inform their peers on goods and services, thereby reducing risk and uncertainty in co-consumers’ decision-making process.
1.1 INTENTION TO WRITE AN ONLINE REVIEW
This study focuses on a special type of eWOM, i.e. online consumer reviews (OCRs). Online consumer reviews are defined as any negative or positive statement about products, services and/or a company, created by consumers (potential, actual, or former) who want to inform and advice their peers (Hennig-Thurau et al., 2004; Park & Park, 2008). Research shows that customers’ satisfaction with a product or service alone does make consumer engages in eWOM behavior (Wirtz and Chew, 2002). Customers need to feel motivated to write online reviews. Motivation is a key determinant of any behavior and is defined as an internal state or condition that serves to arouse or energize behavior and give it a goal-direction (Kleinginna and Kleinginna, 1981). The foremost theoretical approach to motivation is the self-determination theory by Deci & Ryan (2000), which distinguishes two types of motivation based on different sources of initiating a certain action: extrinsic and intrinsic motivation. Extrinsic motivation means that engagement in an activity is goal-driven, i.e. done in order to attain a separable outcome or reward. Intrinsically motivated individuals do an activity for its own sake rather than for external rewards, referring to the fact that the engagement itself provides hedonic or altruistic satisfaction (Amabile et al., 1994; Huang, 2003). Intrinsic motivations, like the altruistic one, are very difficult to influence for companies, as they are embedded in the thoughts and beliefs of the consumer (Dichter, 1966). Thus, this study focuses on influencing consumers’ extrinsic motivation, with the use of incentives, to write online reviews.
1.2 USE OF INCENTIVES
3
provide a positive assurance for fulfilling the needs and wants of individuals. This means that if you expect that some behavior will lead to a valued outcome, you will be more motivated to engage in that behavior. All of these actions were influenced by an incentive to gain something in return for your efforts. This phenomenon is called the reciprocity principle (Cialdini, 2007). Incentives are defined as inducements or rewards that serve as motivational devices for a desired action or behavior (Hansen, 1980). Incentives have been demonstrated to be an important driver of human behavior in general and are considered by the recipient as a sign of appreciation of his or her own behavior by the incentive giver (Lawler, 1984). In this study, we distinguish two major forms of incentives; monetary and non-monetary incentives.
1.2.1 Monetary incentives
A monetary incentive is a transaction which involves a reward based on money (Hansen, 1980). Monetary incentives are often used as a reinforcer to shape behavior (Buhler, 1992; Gupta and Shaw, 1998), as people who are rewarded for a behavior are more likely to engage in that behavior again. Monetary incentives can take many different forms (e.g. cash money, discounts, coupons, etc.). This study focusses on the effects of money and gift card incentives as drivers of eWOM behavior. Cash money as a monetary incentive has been proven to influence people’s extrinsic motivation (Vohs, 2006). Research has proven the effectiveness of money as an incentive, as people want money to attain a state of self-sufficiency in order to attain personal goals and to trade money for goods or services (Vohs, 2006). In addition, using gift card incentives can significantly increase the participation in eWOM programs (Incentive Marketing Association, 2018). Gift card incentives hold several benefits compared to cash money rewards for companies, as they are perceived as more memorable by consumers – they create a lasting reminder of achievement (i.e. trophy value). In addition, awarding gift cards to consumers give companies the certainty that the consumers will revisit their (web)shop in order to spend them.
1.2.2 Non-monetary incentives
4
1.3 CERTAIN AND UNCERTAIN INCENTIVES
Not all incentives are equal and an incentive that one person may find motivating might not be enough to inspire another to take action. When developing incentives, it is key for marketers to understand the company’s audience. Both monetary and non-monetary incentives can take very different characteristics depending on, for instance, the target audience or product category. In this study, we focus on one characteristic on which incentives are different, i.e. the certainty of which the incentive is received by the consumer.
Certain incentives can be defined as rewards which are surely granted by the sender to the recipient after performing a desired action or behavior. In other words, there is a 100% chance of receiving the incentive. Unlike uncertain rewards, which are not surely granted to the consumer. For example, when there is a chance of winning something after you have performed the desired action. In the field of eWOM communication, the use of so called ‘sweepstakes’ is the most commonly applied tool to induce consumers to write online reviews (Incentive Marketing Association, 2018). A sweepstake is a type of contest where a prize may be awarded to an entrant of the contest. In this case, the outcome of the contest is uncertain for the entrant. On the other hand, offering gift cards, as a certain incentive, in return for an online review can increase the participation in the eWOM program (Incentive Marketing Association, 2018).
1.4 PROBLEM STATEMENT
5
1.5 RESEARCH QUESTIONS
Specific key questions are formulated directed towards addressing the major issues derived from the problem statement. These research questions are the fundamental core of this thesis. It focuses the study, determines a large part of the methodology, and guides all stages of inquiry, analysis of the data, and reporting of the results which are to follow in the remaining chapters. The research questions are proposed in order of importance to the topic;
i. What are the effects of monetary versus non-monetary incentives on consumers’ intention to write an online review?
ii. What is the effect of the certainty of receiving (non-)monetary incentives on consumers’ intention to write an online review?
iii. How does the certainty of the incentives affect the strength of the relation of (non-) monetary incentives on consumers’ intention to write an online review?
iv. How do extraneous factors control for the proposed effects?
1.6 ACADEMIC AND MANAGERIAL RELEVANCE
The outcome of this research aims to help marketers to increase the effectiveness of their eWOM management strategy. The findings of this research contributes to the research field of consumers’ eWOM behavior, by attempting to answer the specific research questions proposed in the previous paragraph. Furthermore, the study helps marketers to better understand how to extrinsically motivate consumers to increase their intention to write online reviews on products or services. The results can help marketers to develop a more proactive marketing program in order to stimulate eWOM behavior and, thus, increase their sales.
1.7 STRUCTURE OF THE THESIS
6
2. THEORETICAL FRAMEWORK
In this chapter, existing literature and theories on relevant subjects are reviewed and used to formulate the hypotheses of this research. The goal of this chapter is to bring together opinions, concepts and theories that help to build a conceptual model for the study.
2.1 TYPE OF INCENTIVE
This paper is attempting to measure the effects of different types of incentives, both monetary and non-monetary, on eWOM behavior. Prior research on incentivizing consumers to provide (e)WOM has shown that monetary incentives are found to be an effective tool (e.g. Ahrens et al., 2013; Hennig-Thurau, 2004; Wirtz and Chew, 2002). However, these studies fail to infer the effectiveness of different forms of monetary incentives on eWOM behavior. In addition, Hansen (1980) studies the effects of both monetary and non-monetary on mail survey response. He finds that monetary incentives proved to have a higher effect on mail response than non-monetary incentives. The paper from Hennig-Thurau (2014) defines economic incentives as one of the primary reasons for consumers to express themselves on goods and services. From these studies, it can be assumed that monetary incentives will have a higher positive effect on consumers’ intention to write an online review compared to non-monetary incentives. The psychological explanation behind the outcomes from these studies can be derived from the effect of money itself on consumers. Leading research on this topic is the meta-analytical study from Katleen Vohs (2006) on the psychological consequence of money on people. She suggests that money brings about a self-sufficient orientation in which people prefer to be free of dependency. Self-sufficiency is defined as an insulated state wherein people put forth effort to attain personal goals and prefer to be separate from others (Vohs, 2006). Monetary incentives have the preference over non-monetary incentives since they enable people to achieve personal goals; they are more self-sufficient. Non-monetary incentives cannot provide consumers with this ability, or at least not to the same extent as money. Thus, the first hypothesis of this study states that monetary incentives (i.e. high level of self-sufficiency) will have a higher positive effect on consumers’ intention to write an online review compared to non-monetary incentives (i.e. low level of self-sufficiency);
7
2.2 CERTAINTY OF THE INCENTIVE
Other than the effect of the type of incentive, this study measures the effect of the certainty of receiving these incentives on consumers’ intention to write online reviews. As certainly or uncertainly rewarding consumers after writing an online consumer review might have a different effect on consumers’ intention to write online reviews. This effect has not been accounted for in prior studies regarding incentivizing consumers to stimulate eWOM behavior. The major theory on financial decision making involving uncertainty is the prospect theory from Kahneman and Tversky (1979). This behavioral economic theory explains the way how people choose between alternatives that involve risk, when the probabilities of the outcomes are unknown. The theory implies that people make decisions based on the potential value of losses and gains rather than the final outcome, and that people evaluate these losses and gains using certain heuristics. An important implication of the prospect theory is that the way in which economic agents subjectively frame an outcome in their mind, affects the utility they expect. Deriving from this theory, people have the general tendency to be risk-averse in financial decision-making situations. Risk aversion is defined as the behavior of humans, when exposed to uncertainty, in attempting to lower that uncertainty (Kahneman and Tversky, 1979). It is the consumers’ restraint to agree to a situation with an unknown outcome (i.e. uncertain incentive), rather than another situation with a more predictable outcome (i.e. certain incentive). Thus, the second hypothesis of this study is that certain incentives have a higher positive effect on consumers’ intention to write online reviews compared to uncertain incentives;
H2: Certain incentives have a higher positive effect on consumers’ intention to write online reviews compared to uncertain incentives
In addition to the certainty of the incentive having a direct effect on the intention to write a review, it might also play a moderating role in the effects of the type of incentives used in this study. Given the fact that people are risk-averse and that there is a general preference for incentives that allow people to pursue personal goals (i.e. incentives that are more self-sufficient), it is assumed that the effect of high self-sufficient incentives (i.e. monetary incentives) on consumers’ intention to write reviews is enforced when combined with a higher certainty by which the incentive is received by the consumer;
8
2.3 REWARD SENSITIVITY
To analyze the results from the study accurately, a third, moderating variable, is added to the research to control for the plausible alternative explanation of the respondents sensitivity to receiving rewards in general. As the effects of the types of incentives might significantly differ across individual respondents. For instance, there are people who might value receiving rewards highly, while there are others who will not be affected by them at all. Or there could be instances where it might even have a negative effect on their intention to write an online review when you are getting rewarded for it. A major theory derived from the field of psychology behind this aspect is the reinforcement sensitivity theory (Gray, 1970). In his study, Gray proposes there are three brain behavioral systems (neural structures) which underlie the individual differences in people’s sensitivity to rewards. The relevant system in the case of the responsiveness to rewards is the behavioral activation system (BAS). This system includes the activation of the brain regions which are involved in regulating human arousal when receiving rewards. In general, individuals whose behavioral activation system is more active, tend to be more impulsive and may experience difficulty inhibiting their behavior (Gray, 1970). The theory could imply that respondents will be more likely to comply with the request of writing a review after being exposed to a reward. Thus, it can be assumed that the effects of the types of incentives is enforced when someone has a more active BAS system (i.e. highly sensitive to rewards) compared to someone who has an inactive BAS system (i.e. not sensitive to rewards);
H4: (Non-)monetary incentives will have a higher positive effect on consumers’ intention to write an online review when consumers are highly sensitive to rewards
2.4 CONTROL VARIABLES
9 H1 (+) H2 (+) H4 (+) H3 (+/-)
To complete the set of control variables, the extent to which the consumer likes the specific brand used in the experiment is added. Adding this variable to the model might explain some of the variation in intention to write a review caused by the brand likeability effect (Nguyen, Melewar, & Chen, 2013).
2.5 CONCEPTUAL MODEL
In this paragraph, a conceptual framework is constructed to graphically display the variables used in this study and their assumed relations (i.e. hypotheses) to each other. In the conceptual model (figure 2.1), the hypothesis numbers and their assumed (positive and/or negative) effects on the dependent variable are included (indicated with a +/- sign). The conceptual model displayed below will serve as the main plotline in this study.
FIGURE 2.1 Conceptual model (Experimental variable 1) Type of incentive Monetary vs. Non-monetary (Experimental variable 2)
Certainty of the incentive
Certain vs. Uncertain
(Dependent variable)
Intention to write an OCR
(Control variables) - Tendency to write an OCR
- Brand likeability - Gender
- Age (Moderating variable)
10
3. RESEARCH DESIGN
The third chapter translates the conceptual model into a design for a research. The various aspects which are needed to do so are discussed in the following paragraphs.
3.1 TYPE OF RESEARCH
To test the causal relations stated by the hypotheses, a conclusive study of statistical design is conducted. A statistical design is a series of basic experiments that allows for statistical control and analysis of external variables (Malhotra, 2010). Only causal designs encompassed by experimentation are appropriate for inferring cause-and-effect relationships (Malhotra, 2010), like the ones in this study. In addition, interactions to the relation add specificity to the phenomena; they make the predictions of this research more specific. An interaction is said to take place when the simultaneous effect of two or more variables is different from the sum of their separate effects (Malhotra, 2010).
3.2 PARTICIPANT AND DESIGN
In order to test the hypotheses, a factorial between-subject design is applied in this study, which means that participants of the survey will serve in only one condition (Malhotra, 2010). This type of experiment is used to eliminate any carry-over effects between subjects. The factorial design measures the effects of every level of every independent variable crossed with every level of every other independent variable. The factorial design of this study is conceptualized in figure 3.1.
FIGURE 3.1 2x3 Factorial design
Type of incentive
Monetary Non-monetary
Money Gift card Gift
Cer ta in ty o f t he in ce nt iv e Cer ta in
Condition 1 Condition 3 Condition 5
U
nce
rt
ai
n
11
As shown in figure 3.1, this will require 2 x 3 = 6 conditions. In the factorial design, each level of the first experimental variable (i.e. type of incentive) represents a column and each level of the second experimental variable (i.e. certainty of the incentive) represents a row. The respondents are randomly assigned to one of these six conditions.
3.3 PRELIMINARY TEST
A preliminary test is conducted with a sample of 26 subjects to determine the manipulations of the uncertain incentive conditions in the experiment. In this pre-test, the participants are asked to indicate which amount (in €’s) in an uncertain incentive scenario, could be of an equal value alternative to that of a €5,- certain incentive. The preliminary test results can be found in the appendices of this report (Appendix B). The results show that 65,38% of the participants indicated that €500,- is an appropriate equal value alternative for the uncertain incentive scenarios. From these results, it is adopted in the experiment that every €5,- certain incentive scenario will have an opposing €500,- uncertain incentive scenario.
3.4 SURVEY DESIGN
In this paragraph, the considerations regarding the choices in the questionnaire are explained. For each of the conditions (figure 3.1), a specific stimulus is designed. In experimental psychology, a stimulus is an event or an object to which a response is measured (Malhotra, 2010). In this study, the response measured is the intention to write an online review. The process of the construction for the scenario and stimuli is explained in the following paragraphs. The complete survey used in the experiment is included in Appendix A.
3.4.1 Scenario
12
(one out of six possible reward scenarios). The introduction of the Coolblue brand in the stimuli, resulted in having to control for the brand likeability effect (Nguyen, Melewar, & Chen, 2013). The e-mail invitations are the different incentive scenarios, explained in the following paragraphs.
3.4.2 Choice for incentives
The construction of the incentive scenarios is a critical part of this experiment. Literature suggests that there is a preference for incentives that allow people to pursue personal goals, as they are more self-sufficient (Vohs, 2006). Self-sufficiency is the underlying mechanism for the choice and construction of the stimuli. The monetary and non-monetary incentives in this research are designed into 3 incentive categories, each representing a different level of self-sufficiency (figure 3.2). Monetary reward scenarios are divided into a money and a gift card incentive. The non-monetary reward scenarios will offer participants a pre-determined gift.
FIGURE 3.2
Level of self-sufficiency of the incentive categories
13
within the gift category, a chance of winning a digital SLR-camera by Canon (valued at €500,-) is used. This gift is related to the store where the laptop is bought from and was actually used by Coolblue as a prize for a sweepstake contest in 2018 to increase eWOM participation.
3.4.3 Magnitude of the incentives
A critical issue regarding the manipulation of the experimental variables remains the determination of the magnitude of the incentives in this survey. For the certain reward scenarios, rewards worth €5,- were used. In the opinion of the researcher, the certain reward scenarios have to be of a sufficient magnitude in order to elicit a certain behavior from consumers. Reward scenarios of either €1,- or €2,- euro would be too low of an impact on respondents. The pre-test results (3.3) inferred that, in the main survey, every €5,- certain incentive will have an opposing €500,- uncertain incentive scenario. To support this result, e-commerce company’s (i.e. Coolblue, Mediamarkt, BCC) current product review invitations use sweepstake contests with similar uncertain incentives worth (≈) €500,-. In the below figure, two examples of e-mail invitations from the main survey are shown. These are the certain and uncertain incentive scenarios of the incentive category: money.
FIGURE 3.3
A certain versus an uncertain incentive scenario
14
To conclude, the magnitude of the incentives is a crucial aspect in this study. Although the risks for the choices made for this aspect have been reduced through pre-testing and the analysis of secondary data, it can still be argued for that they are assumption based. Thus, magnitude of incentives will return as an issue in the limitations of this research.
3.5 OPERATIONALIZATION
The research method of obtaining information is based on the questioning of respondents through a digital survey. In this survey, respondents are asked a variety of questions regarding their behavior, attitudes, motivations and demographic characteristics towards writing online reviews after being exposed to one of the six incentive scenarios. The digital survey is developed with the use of Qualtrics. Qualtrics is an online survey software solution creating online surveys easy and which has excellent analytical capabilities. This helped greatly in the process of interpreting the results from the survey.
15 OPERATIONALIZATION TABLE (3.1)
Underlying construct
Source Items Scale
Eigen-value Cronbach’s Alpha Intention to write an OCR Based on; Reimer & Benkenstein (2016); Cheung and Lee (2012)
1. I would consider to write a review on the Coolblue website 2. It is very likely that I will write a review on the Coolblue website 3. I intend to share my experiences
by writing a review on the Coolblue website
4. I am determined to write a review on the Coolblue website
7-point likert scale: ranging from strongly disagree –
strongly agree 3,445 ,946
Probability to write an OCR
N/A 1. Please indicate the probability that you will write a review on the Coolblue website
100-point slider scale: ranging from 0% -100% N/A Uncertainty (Manipulation check) Based on; Priester et al. (2004)
1. Please indicate how certain you think it will be that you will actually receive this reward
7-point bipolar scale very uncertain – very certain N/A Reward Sensitivity Based on; Lichtenstein et al. (1990); Wirtz & Chew (2002)
1. Receiving rewards makes me feel good
2. I enjoy receiving rewards, regardless of the value of the reward itself
3. Receiving rewards gives me a sense of joy
4. Rewards, regardless of the value they hold, have caused me to do things I normally would not do
7-point likert scale: ranging from strongly disagree – strongly agree. 2,562 ,796 (,852 if item4 deleted)
Attention check Based on;
Hauser & Schwarz (2015)
1. To check if you are still paying attention to the survey, please select ‘Strongly disagree’ here
7-point likert scale: strongly disagree – strongly agree N/A Co nt ro l v ari ab le s Brand Likability Based on; Martin & Stewart (2001)
1. Please indicate to what extent you like the Dutch e-commerce company Coolblue
7-point bipolar scale: ranging from dislike very much – like very much
N/A Tendency to write an OCR Based on; Hennig-Thurau et al, (2004); Reimer & Benkenstein (2016)
1. I usually write online reviews when I am satisfied with a product or service
2. I regularly write reviews, because I enjoy writing reviews itself. 3. I often write reviews because I
want to help others in their decision-making process 4. I always write reviews because I
want to share my experiences with products and services
7-point likert scale: ranging from strongly disagree –
strongly agree. 2,879 ,865
Gender N/A 1. Please indicate your gender Male – Female N/A
16
3.5.1 Dependent variable measurement
The operationalization table displays the steps through which participants answered the different items in the online survey. The survey begins with the description of the scenario, after which, participants were randomly exposed to one of the six possible incentive scenarios. Directly after being exposed to a stimulus, participants’ intention to write an online review was measured using a 4-item measurement scale. Participants had to indicate the degree of (dis)agreement on a 7-point likert scale on 4 statements about their intention to write an online review. This scale is based on the studies of Reimer & Benkenstein (2016) and Cheung & Lee (2012).
An additional item was added in the survey to measure participants’ intention to write a review. After answering the 4 items on the intention to write a review, participants had to indicate the probability they will write a review by using a slider scale. This question was added as a back-up, in case the main items failed to measure the same underlying dimension accurately (i.e. intention to write a review). In addition, it is interesting to see if there were any significant differences in the results on both measurement scales. Fortunately, the main dependent variable questions passed factor and reliability analysis and since these questions were based on existing literature, they were used in further analysis. Even though all the tests were run on both dependent variables, no significant differences between the two outcome variables were observed.
3.5.2 Manipulation check: perceived uncertainty
A manipulation check question was asked to the respondents who were exposed to an uncertain reward scenario. Manipulation check is a secondary evaluation of the manipulation in the experiment (Malhotra, 2010). This question was added to check whether the participants really perceived the reward a(n) (un)certain scenario. The question added, which can be seen in Appendix A, is a 7 point bipolar scale ranging from ‘very uncertain’ to ‘very certain’. This scale is based on the study from Priester et al. (2004), who used the scale as a measure for attitudes and considerations of choice for a product.
17 TABLE 3.2
Manipulation check results: perceived uncertainty
Scenario Mean level of
uncertainty T-statistic df Sig. (2-tailed)
Man. Check 2 1,91 -9,609 45 ,000
Man. Check 4 1,78 -10,306 45 ,000
Man. Check 6 2,29 -6,750 44 ,000
Test value = 4 (neutral perception)
The results show a significant difference between the means from the manipulation checks and the known population mean of the certain reward scenarios (µ =7) and from a neutral perception (µ =4). This is indicated by the high t-statistics (Malhotra, 2010). This means that participants actually perceived the uncertain reward scenarios as uncertain, so the manipulation applied in this research was proven to be successful.
3.5.3 Extraneous variables
After the manipulation check, all extraneous variables included in this research are measured. Participants’ reward sensitivity was measured using a 4-item measurement scale. Participants had to indicate the degree of (dis)agreement on a 7-point likert scale with a series of 4 statements about their sensitivity of receiving rewards in general. This scale is based on the studies Lichtenstein et al. (1990) and Wirtz & Chew (2002).
A fifth statement was added to the measurement of the reward sensitivity construct. This statement is added to check if participants are still paying attention to the statements in the survey and, thus, ensuring they are contributing useful data in this research. The statement instructed participants to select the ‘Strongly disagree’ answer option on a 7-point likert scale. After the measurement of reward sensitivity, the control variable constructs are measured. First brand likeability was measured using a single item question with answer options on a 7-point bipolar scale ranging from ‘dislike very much’ to ‘like very much’. This scale is based on the study of Martin & Stewart (2001). Next, participants’ general tendency to write online reviews was measured using a 4-item measurement scale. Participants had to indicate the degree of (dis)agreement on a 7-point likert scale with a series of 4 statements about their general tendency to write online reviews. This scale is based on the studies of Hennig-Thurau et al. (2004) and Reimer & Benkenstein (2016).
18
3.6 POPULATION AND SAMPLING METHOD
The population of interest in this study consists of people who either use and/or create electronic word of mouth. The sampling method used in this research is convenience sampling. Convenience sampling is a non-probability sampling technique where respondents are selected because of their convenient accessibility and proximity to the researcher (Malhorta, 2009). The simple reason for the use of this sampling method is that a lot of data needs to be generated in a relatively short period of time. As an effort to increase response rate, a respondent is randomly selected from the total pool of total subjects to win a prize (i.e. gift card).
The time period in which the data was collected, runs from the 15th of May to the 22nd of May 2018. Participants were asked to fill in the survey through e-mail, social media platforms like Facebook and LinkedIn, and survey sharing services like surveyswap.io and surveystudent.nl. At least 25 respondents per condition are needed in order to get sufficient data in order to use for analytical testing (Malhotra, 2010), resulting in a minimum set of subjects of 150 respondents. With 300 recorded responses this requirements was sufficiently satisfied.
3.7 PLAN OF ANALYSIS
When sufficient data was collected using Qualtrics, the data analysis process started using SPSS software. SPSS (Statistical Package for the Social Sciences) is a predictive analytics software package that provides statistical analysis/reporting, predictive modelling, etc. in this research. The tests that will be used in this research are explained in the following paragraphs.
3.7.1 Factor analysis
Factor analysis is a data reduction technique that tests survey items that are attempting to measure the same underlying construct. The purpose of the analysis is to reduce a large quantity of data that has been collected by finding common variance to retrieve an underlying dimension/construct in the dataset. The main results from the factor analysis per construct are included in the operationalization table. The complete factor analyses per construct can be found in Appendix C.
19
Principal component analysis method is used (Malhorta, 2009). In this method of factor analysis, the total variance in the data is considered. The factors are then rotated after extraction. Rotation does not improve or deteriorate the solution, but it makes the results more interpretable (Malhotra, 2010). Rotating the factors also ensures that they are orthogonal, which eliminates problems of multicollinearity in later regression analyses (Malhotra, 2010).
The second step in the analysis is determining how many components to extract from the data. To do this, we use Kaiser criterion (Malhorta, 2009). In this criteria, we are interested only in components that have an initial Eigenvalue greater than 1.0. As an additional check, by looking at the percentage of variance explained, the criteria for the number of factors to retain are the components that explain at least 60% of the variance (Malhotra, 2010). As an additional check, we look at the scree plots generated by SPSS, included in Appendix C. We see that the ‘break’ (i.e. elbow) in the graph occurs at item number 2. Only components above this ‘break’ are retained for further testing (Malhorta, 2010). As a final check, we look at the communalities. The communalities measure the percent of variance in a given variable explained by all the extracted factors (Malhotra, 2010). The criteria for the communalities is that they have to be above .4. This final condition is also satisfied in all cases. Thus, we can assume there is unidimensionality in the variables, meaning they are explaining the same underlying dimension (Malhotra, 2010).
3.7.2 Reliability analysis
As an additional check, Cronbach's alpha score is calculated. Cronbach’s alpha is an estimate of the internal consistency reliability associated with the scores that can be derived from a scale or composite score. Determining the reliability is important, because in the absence of reliability it's impossible to have any validity associated with the scores of a scale. A Cronbach's alpha score above .60 is commonly regarded acceptable (Malhorta, 2010).
20
3.7.3 Analysis of variance
To test the main effects of the experimental variables and their interaction with each other, an analysis of variance (ANOVA) is conducted. Analysis of variance is used as a test of means for two or more populations (Malhotra, 2010). Since the analysis of variance is a test for comparing groups, there is no need to use dummy variables yet. The following ANOVA model is constructed: Y = βo + Ev1 + Ev2 + Ev1*Ev2
3.7.4 Regression analyses
In addition to testing the main effects and their interaction with each other, we are also interested in the effects of possible other extraneous variables influencing the dependent variable. In table 3.3, you find an overview of the models of interest for this research.
TABLE 3.3
Overview of regression models
Model # Description of the model Regression formula
Model 1 Base model with only the main
effects on the DV Y = βo + β1 * Ev1 + β2 * Ev2_Dummy1 + β3 * Ev2_Dummy2 Model 2 Base model with the main effects
on the DV and also the interaction of the main effects
Y = βo + β1 * Ev1 + β2 * Ev2_Dummy1 + β3 * Ev2_Dummy2 + β4 * Interaction certainty and Dummy1 + β5 * Interaction certainty and Dummy2
Model 3 Base model with the main effects and the interaction of these effects, complemented by the main effect of reward sensitivity and their interaction with the first experimental variable
Y = βo + β1 * Ev1 + β2 * Ev2_Dummy1 + β3 * Ev2_Dummy2 + β4 * Interaction certainty and Dummy1 + β5 * Interaction certainty and Dummy2 + β6 * Reward Sensitivity + β7 * Reward sensitivity and Dummy1 + β8 * Reward sensitivity * Dummy2
Model 4 Base model with the main effects and their interactions, the main effects of the moderator and their interactions, complemented by all 4 control variables on the intention to write a review.
Y = βo + β1 * Ev1 + β2 * Ev2_Dummy1 + β3 * Ev2_Dummy2 + β4 * Interaction certainty and Dummy1 + β5 * Interaction certainty and Dummy2 + β6 * Reward Sensitivity + β7 * Reward sensitivity and Dummy1 + β8 * Reward sensitivity * Dummy2 + β9 * Tendency to write a review + β10 * Brand likability + β11 * Gender + β12 *Age
3.7.5 Mean centering
21
3.7.6 Dummy coding
Since the experimental variable ‘type of incentive’ contains three levels (i.e. money, gift card and gift), two dummy variables were computed in SPSS to be used in further regression analysis (Malhotra, 2010). A dummy was created for people who were exposed to a money incentive (value =1) versus people who did not get a money incentive scenario (value =0). The second dummy variable gives information on people who were shown a gift incentive (value =1) versus people who were not shown a gift (value =0). This means that the gift card incentive was used as a base. The additional reason for this choice is that gift cards are the most commonly used tool to induce consumers to write online reviews (Incentive Marketing Association, 2018). And so, in the regression analyses, we will see the differential effect of money and gifts, relative to gift cards (which is the base effect).
3.7.7 Multicollinearity
Multicollinearity is a common problem when estimating linear or generalized linear models (Farrar & Glauber, 1967). Multicollinearity occurs when there are high correlations among predictor variables, and as such predicting the variance of another variable, leading to unreliable and unstable estimates of regression coefficients (Malhotra, 2010). The most widely-used diagnostic for multicollinearity is the variance inflation factor (VIF). Looking at the VIF scores in the study’s full model (model 4) without mean centering (see appendix C), we see that the introduction of interaction variable of the moderator (i.e. reward sensitivity) with the experimental variable dummies has resulted in high VIF scores. Thus, this specific variable introduces multicollinearity in model 4. A solution to this problem is to split up this continuous variable into a dummy variable based on the median (Malhotra, 2010). This means that a 0 represents a value below the median and 1 a value above the median. Rerunning the regression analysis of model 4, with the use of this new dichotomous variable, resulted into a decrease of multicollinearity in model 4.
3.8 DESCRIPTIVE STATISTICS OF THE SAMPLE
The introduction of this paragraph covers the process of preparing the dataset for further analyses and testing in SPSS. The second paragraph describes the actual sample that is used in further testing.
3.8.1 Data preparation
22
process, 331 people participated in the survey. Of this sample, 31 respondents did not fully complete the survey and so their responses were still in progress. After one week the survey was online, their responses were automatically deleted from Qualtrics. In addition, an attention check was integrated in the survey. An attention check is designed to ensure that participants are still paying attention to the survey and thus contributing useful data to this research. Unfortunately, 21 participants failed to pass the attention check and therefore their responses were deleted. This leaves 279 respondents to have fully completed the survey. In the end, the analyzed net sample size (N) was determined at 279 responses.
3.8.2 Descriptive statistics of the sample
The first step in the data analysis is to get a better insight in the data that is collected through descriptive statistics. The total valid sample size consists of 279 respondents. These respondents were randomly and evenly divided over the six scenarios. Respondents’ gender and age per scenario are displayed in the table below. 170 respondents (60,73%) were male and 109 respondents (39,27) were female. The average age of respondents is 26,7 years old. The age of the respondents do not significantly differ across scenarios (F = ,683 , p = ,637 > .05). In table 3.4 you find an overview of the descriptive statistics of the sample.
TABLE 3.4
Descriptive statistics of the sample
Type of incentive Scenario N Gender Average
Age Male (%) Female (%) M on et ar y Money 1 2 43 46 60,5 54,3 39,5 45,7 25,5 26 Gift card 3 49 73,5 26,5 26,5 4 46 54,3 45,7 26,5 N on -mo ne ta ry Gift 5 50 64,0 36,0 27,7 6 45 57,8 42,2 28 TOTAL 279 60,73 39,27 26,7
a. Chi-square test not significant at the ,05 level b. No association found between scenarios and age
23
4. RESULTS
This chapter contains the results from the analytical tests set in the plan of analysis (paragraph 3.7). The extended outputs of the tests are included in the appendices (Appendix D and E).
4.1 ANALYSIS OF VARIANCE
In order to analyze the influence of the type and certainty of receiving the incentive on consumers’ intention write a review, a two (certainty of incentive: certain vs. uncertain) by three (type of incentive: money vs. gift card vs. gift) analysis of variance (ANOVA) was conducted on consumers’ intention to write a review. In table 4.1 you find an overview of the distribution of the average intention to write a review per condition. In table 4.2, the test of between subject effects is displayed.
TABLE 4.1
Average intention to write a review per condition
Ev2 Ev1
Monetary Non-monetary Total
Money Gift card Gift
Certain 5,3256 4,0102 4,0100 4,4085
Uncertain 3,8804 3,7283 3,4500 3,6880
Total 4,5787 3,8737 3,7447 4,0547
Dependent variable = Average intention to write a review (range 1 – 7)
TABLE 4.2
Test of between subjects effects
Source Type III Sum
of Squares
Df Mean Square F Sig.
24
In table 4.1, we see that the highest score on the intention to write a review are the participants who were exposed to a certain money incentive (M = 5,3256). The lowest score on the intention to write a review are the participants who were exposed to the uncertain gift scenario (M = 3,4500). Participants who were exposed to a certain incentive scored higher (M = 4,4085) than people who were exposed to an uncertain scenario (M =3,6880). From the ANOVA results (table 4.2), the main effect of certainty is proven to be significant (F= 15,117 , p < .01). In addition, participants who were exposed to a money scenario, scored higher on their intention to write a review (M = 4,587) than people who were shown a gift card (M = 3,8737) or a gift (M =3,7447). From the ANOVA results, the main effect of the type of incentive is proven to be significant (F = 7,473 , p < .01). However, the difference between a gift card scenario and a gift scenario was not proven to be significant (p > ,05). The interaction effect between the two main effects is proven to be significant (F = 3,139, p < 0,01).
In addition, the mean differences of the average intention to write a review are compared pairwise to check for any significant differences within the experimental variable. The pairwise comparisons table is included in Appendix D. By pairwise comparing the types of incentives, we see that there are only significant differences in the means of the dependent variable when a money incentive is present (p < ,05). From this table, we see that for participants who were offered a gift card or a gift, it does not matter if the gift card was certain or uncertain. In both cases the mean difference on their average intention to write a review do not significantly differ. However, participants who got a money incentive, the certainty of receiving money does make a significant difference in their intention to write a review (F= 15,117 , p < .01).
25 FIGURE 4.1
Estimated marginal means of intention to write a review
In figure 4.1, we see the main effect of the certainty of receiving an incentive is clearly visible. The certainty of the incentive has a higher average score on the intention to write a review across all the types of incentives. The intention to write a review shows a remarkable jump from a certain gift card to a certain money incentive. This means that the interaction effect of certainty on the type of incentive really comes into play when the type of incentive is a money incentive. The most important conclusion from the analysis of variance on the experimental variables is that the certainty of receiving the incentive has a higher positive effect on the intention to write a review on incentives that have a high level of self-sufficiency (i.e. money), compared to incentives that have a low level of self-sufficiency (i.e. gift).
4.2 REGRESSION ANALYSIS
26 TABLE 4.3
Regression results per model
Model 1 Model 2 Model 3 Model 4
Intercept 3,488*** 3,728*** 3,734*** 3,399*** Main effects Certainty ,748*** ,282 ,550 ,346 Dummy 1 Money ,729** ,152 ,021 ,150 Dummy 2 Gift -,137 -,278 -,381 -,536 Interaction effects Certainty * Dummy1 1,163** ,992** ,897** Certainty * Dummy2 ,278 ,039 ,025 Moderating effects Reward sensitivity ,562*** ,475 Reward sensitivity * Dummy1 ,090 ,239 Reward sensitivity * Dummy 2 ,007 ,478
Control variable effects
Tendency to write a review ,538*** Brand likability ,226*** Gender ,132 Age -,022** Explanatory power R2 ,092 ,112 ,283 ,352 Adjusted R2 ,082 ,096 ,262 ,323 F-value 9,278*** 6,909*** 13,312*** 12,063***
*** p-value < .01, **p-value < .05, * p-value < .10
b. multicollinearity has been reduced through median splitting
Model 1
27
Model 2
Model 2 shows the main effects and their interaction with each other on the average intention to write a review. The results indicate that the overall model is significant (F = 6,909 , p < .01). There is a small increase in the model’ explanatory power (R2 = ,112), compared to model 1 (R2 = ,092). Surprisingly, due to the introduction of the interaction variables to the model, the
main effects become insignificant (p > .05). However, the interaction effect of certainty on the change from a gift card to a money incentive is significant (β = 1,163, p < .05). This means that the interaction effect of certainty is only significantly present when respondents were exposed to a money incentive scenario.
Model 3
Model 3 shows the main effects, their interaction with each other and the effects of the moderating variable (i.e. reward sensitivity). The results indicate that the overall model is significant (F = 13,312 , p < .01). This model explains a lot more of the variance in the average intention to write a review than the previous models. The adjusted R2 increases with 16,6%, compared to model 2. This means that adding the reward sensitivity variable to the model increases its explanatory power significantly. Furthermore, the main effects of the type of incentive and the certainty of receiving incentives are insignificant (p >.05). The interaction of certainty and money is still significant (β = ,992 , p < .05). The main effect of reward sensitivity is significant (β = ,562 , p < .01). The interactions of rewards sensitivity on the types of incentives are not significant (p > .05).
Model 4
28
< .01). Age seems to have a significant effect as well (β = -, 022 , p < .05). Lastly, gender does not have a significant effect on the intention to write a review (p > .05).
4.3 HYPOTHESES TESTING
In this paragraph, statistical decisions regarding the rejection or (partial) support of the hypotheses are made. In table 4.4 you find an overview of the hypotheses followed by an indication if the hypothesis is rejected or (partially) supported. After the table, the inferences made will be explained and supported by data from the study.
TABLE 4.4 Hypotheses overview
Hypothesis # Hypothesis description Rejected/Supported
H1 Monetary incentives have a higher positive effect on consumers’
intention to write online reviews compared to non-monetary incentives
Partially supported
H2 Certain incentives have a higher positive effect on consumers’
intention to write online reviews compared to uncertain incentives Supported
H3 Monetary incentives will have a higher positive effect on consumers’
intention to write an online review when the incentive is certain compared to when they are uncertain
Partially supported
H4 (Non-)monetary incentives will have a higher positive effect on
consumers’ intention to write an online review when consumers are highly sensitive to rewards
Rejected
29
2nd hypothesis. Meaning that the certainty of receiving incentives has a significant effect on the
intention to write online reviews.
To test the 3rd hypothesis (H3), we rely on the combination of data from the ANOVA analysis and the results from model 2 of the regression analyses. With this data we are only able to partially support this hypothesis. As certainty only has a significant interaction effect when it is combined with a money incentive (β = ,992 , p < .05). However, the effect of certainty on the gift card incentive was not proven to be significant. As gift cards are defined as a monetary incentive in chapter 2, we are not able to fully support this hypothesis.
30
5. DISCUSSION
The final chapter discusses the results of this study. It describes the meaning of the results and how they relate to the research questions proposed in paragraph 1.5. The managerial and academic implications of the findings are discussed and the limitations of this study are acknowledged. The last paragraph recommends directions for future research in this area.
5.1 DISCUSSION OF THE RESULTS
The purpose of this study was to examine the effects of different types of incentives and the (un)certainty of receiving these incentives on consumers’ intention to write an online review. The first research question addressed the contrasting effect of monetary versus non-monetary incentives on consumers’ intention to write online reviews;
RQ1 = What are the effects of monetary versus non-monetary incentives on consumers’ intention to write an online review?
The findings of this report indicate that offering money to consumers has the highest effect on their intention to write online reviews. In addition, no significant difference was found between offering consumers gift cards or gifts. This means that the type of monetary incentives does matter when it comes to their effects. From this study, it is not clear that we can conclude that all types of monetary incentives have a significant higher effect versus non-monetary incentives. However, we can infer that money, as a monetary incentive, has a significantly higher effect on consumers’ intention to write reviews than gifts, as a non-monetary incentive. The second research question addressed the effect of the (un)certainty characteristic of incentives on consumers’ intention to write online reviews;
RQ2 = What is the effect of the certainty of receiving (non-)monetary incentives on consumers’ intention to write an online review?
31
The third research question addressed the possible interacting effect of the (un)certainty characteristic with the different types of incentives on consumers’ intention to write online reviews;
RQ3 = How does the certainty of the incentives affect the strength of the relation of (non-) monetary incentives on consumers’ intention to write an online review?
The findings of this report indicate that the interaction effect of the certainty of receiving (non-)monetary incentives only plays a significant role when you offer consumers money as an incentive. No significant interaction takes place when comparing certain and uncertain gift (card) incentives. Although the certain incentives have a higher effect on the intention to write online reviews is higher across all incentive types, the effects are not enforced by each other when you offer gift (cards). However, the intention to write online reviews increases drastically when the money incentive is certain to be received by the consumer (see figure 4.1). This means that consumers are somehow more triggered to certainty when it comes to receiving money, compared to gift (cards).
The fourth research question addressed the issue if there are any plausible alternative explanations why consumers intend to write online reviews, other than the proposed effects of the experimental variables;
RQ4 = How do extraneous factors control for the proposed effects?
32
To conclude, the title of this thesis addresses how to motivate consumers to write online reviews. The answer to this question: It depends! When developing an incentive program to increase consumers’ eWOM participation, it is key for marketers to understand the context of the company and its target audience. Based on this research, monetary incentives in the form of certainly received money has the highest effect on consumers’ intention to write online reviews. However, this study describes a scenario in which people are (already) satisfied with a new laptop, which is a high involvement product. The findings of this research cannot be generalized for products in which people are lower involved or when they are not satisfied with the product itself. In addition, both monetary and non-monetary incentives can take many different forms, other than the forms used in this research. This means that other possible forms of (non-)monetary incentives may have a different effect on consumers’ intention to write online reviews. These limitations are further discussed in paragraph 5.3.
5.2 MANAGERIAL AND ACADEMIC IMPLICATIONS
In this section, the possible managerial implications of the results in this study are discussed. In addition, academic implications are discussed by reflecting the findings of this report with the findings of literature research in the theoretical framework (chapter 2).
As a managerial implication, the outcome of this research aims to help marketers to increase the effectiveness of their eWOM management strategy. This research tries to broaden marketers’ perspective on the ways they can induce consumers to write online reviews. It shows that different types of incentives have different effects on eWOM behavior, some more effective than others. Also, from this study it is shown that the aspect of (un)certainty of receiving incentives plays a significant role in this effect. The findings of this research show that the present-day tool to induce consumers to write online reviews (i.e. sweepstake contests using gift card incentives) is not the most effective way to do so. It shows that consumers have significantly lower intentions to write online reviews, when the incentive is uncertain to receive by consumers. Even though the value (in €’s) of the uncertain incentives are much higher. The results show that incentives that hold a high level of self-sufficiency (i.e. money) in combination with the interaction effect of certainty, has the highest effect on consumers’ intention to write online reviews.
33
desired to boost sales (e.g. Babic et al., 2015; Floyd et al., 2014; Pauwels et al., 2016), they could consider to use a certain money incentive over other incentives. In addition, when companies want to increase eWOM for their products or services for only a short period of time, they could consider to use certain money incentives. These results can help marketers to develop a more proactive marketing program in order to stimulate eWOM behavior and, thus, increase their sales.
As academic implications, some findings are not in line with the results of the literature research. Literature suggests that there is a general preference for incentives that allow people to pursue personal goals, i.e. which have a higher level of self-sufficiency (Vohs, 2006). And although this statement holds for money as an incentive versus a gift (card) incentive, the difference in effects between gift cards and gifts is not significant. This means that participants in this study did not evaluate the increase in self-sufficiency of the incentive from a gift to a gift card. The study hypothesized that gift cards, as a monetary incentive, has a higher effect on consumers’ intention to write a review, compared to non-monetary incentives (i.e. gifts). However, this hypothesis is not fully supported by the data from this study. The difference of effects between offering consumers gift cards or gifts is not significant. This could imply that people do not perceive gift cards to be more self-sufficient than gifts. i.e. gift cards do not allow them to pursue personal goals. It could also mean that consumers perceive gift cards as a trick to lure them in to spending more at the specific company’s (web)shop. Further research addressing this issue has to be made to clarify this result.
34
5.3 LIMITATIONS
Encompassing this study, several limitations of the research related to the data and the methodology are acknowledged.
In this study, the construction of the stimuli used in the experiment remains an issue. Even though the risks concerning this aspect have been tried to be reduced through pre-testing, it is still not safe to assume that the stimuli are safe from biasing effects. For instance, the gift incentives (i.e. the USB flash drive and an SLR camera) might not be favorable/desirable to some respondents, which might have affected their responses. The magnitude of the certain incentive scenarios (€5,-) versus the uncertain incentive scenarios (€500,-) is still an issue in this research. Although pre-test results show that people valued a €500,- uncertain incentive as an equal valued alternative for a €5,- certain incentive, this inference cannot be generalized. Some people might think this amount is either too low or too high. Further testing using a larger sample size and different magnitudes of incentives is needed.
In addition, the sample size was required through a convenience sampling method; respondents were obtained via social networking of the researcher. Although this sampling method resulted in a sufficient sample size in relatively a short period of time, it also has some negative consequences regarding the external validity of the results of this study. Thus, the heterogeneity of the sample cannot be argued for. The sample size was determined at 331 respondents, which can be argued as a small sample size when making inferences about general consumer behavior. Furthermore, this study was conducted with (young) Dutch consumers and may not reflect the consumer behavior in other countries. In addition, the average age of respondents in this study is 27 years old and there is not a lot of variance in the age of respondents. Thus, the results may not accurately reflect how other age category samples might behave.
35
5.4 RECOMMENDATIONS FOR FUTURE RESEARCH
Based on the limitations of this research, recommendations for future research are proposed. This research broadens existing knowledge on how to extrinsically motivate consumers to write online reviews. Furthermore, the findings also highlights the necessity to continue research in this specific area.
This study encompassed the effects of three types of incentive categories. However, there are many other possible forms of (non-)monetary incentives to use as a tool to influence eWOM behavior. For instance, using loyalty programs where consumers can receive credits for writing reviews can prove to be an effective tool. Also, offering shoppers free shipping or extended warranty licenses to compensate them for writing reviews. Thus, future research should focus on bringing other types of incentives into scope, testing their effects on consumers’ intention to write online reviews. In addition, an interesting question is whether the intrinsic motivations to engage in eWOM can be influenced by companies through marketing actions. Simultaneously overcoming the main drawback of using monetary incentives, as these reviews suffer from a loss of credibility of the review (Martin, 2014).
As mentioned in the limitations, future research should test the effects of different magnitudes of the certain/uncertain incentives. Also, it might be interesting to observe the effects of incentives in a scenario where people are unsatisfied with the product. In addition, future research should assess the effects of incentives in different product categories.
In addition, the main and interaction effects of reward sensitivity was not proven to be significant in model 4. However, model 3 (without the control variables) does show a significant main effect of rewards sensitivity. These contrasting results make it difficult to interpret the effect of this variable. Thus, future research should incorporate a condition where no incentive is offered to participants. In doing so, the effect of reward sensitivity (among others) can better be interpreted by the researcher.
36
REFERENCES
Ahrens, J., Coyle, J. R. & Strahilevitz, M. Ann. (2013). Electronic word of mouth: The effects of incentives on e-referrals by senders and receivers. European Journal of Marketing, 47 (7), 1034-1051.
Amabile, T. M., Hill, K. G., Hennessey, B. A., & Tighe, E. M. (1994). The Work Preference Inventory: assessing intrinsic and extrinsic motivational orientations. Journal of personality and social psychology, 66(5), 950.
Babić Rosario, A., Sotgiu, F., De Valck, K., & Bijmolt, T. H. (2016). The effect of electronic word of mouth on sales: A meta-analytic review of platform, product, and metric factors. Journal of Marketing Research, 53(3), 297-318.
Bitner, M.J., B.H. Booms, and M.S. Tetreault (1990), The Service encounter: Diagnosing Favorable and Unfavorable Incidents, Journal of Marketing, 54, 71-84
Buhler, P. (1992). The keys to shaping behavior. Supervision, 53(1), 18-20.
Chen, Y., Q. Wang, and J. Xie (2011), Online social Interactions: A Natural Experiment on word of Mouth versus Observational Learning, Journal of Marketing Research, 48, 238-254.
Cialdini, R. B., & Cialdini, R. B. (2007). Influence: The psychology of persuasion (pp. 173-174). New York: Collins.
Corr, P. J. (2004). Reinforcement sensitivity theory and personality. Neuroscience & Biobehavioral Reviews, 28(3), 317-332.
Deci, E. L., & Ryan, R. M. (1987). The support of autonomy and the control of behavior. Journal of personality and social psychology, 53(6), 1024.
Dellarocas C. (2003). The digitization of word of mouth: promise and challenges of online feedback mechanisms. Journal of Management Science
Dichter, E. (1966). How word-of-mouth advertising works. Harvard business review, 44(6), 147-160.
Farrar, D. E., & Glauber, R. R. (1967). Multicollinearity in regression analysis: the problem revisited. The Review of Economic and Statistics, 92-107.
