Message Oriented Approach to WOM Effects in Service Industries

Journal of Economics and Financial Analysis, Vol:1, No:2 (2017) 17-47


Message Oriented Approach to WOM Effects in Service Industries


Mustafa DILBERa, Lokman INCIRKUSb,*

a Department of Public Relations and Publicity, Yeditepe University, Turkey

b Department of Management, Istanbul University, Turkey


Abstract

In this study a search-based and an experience-based service were examined in Word-of-Mouth (WOM) context. The study aimed at revealing the extent to which consumer choices in movie theatre and repair and maintenance shop services are influenced by the experience communicated by personal sources. Four key contributions are planned. Firstly, although it has attracted attention and criticism there is no empirical examination regarding message characteristics. In this study, the effects of messages delivered by senders concerning purchase decisions are investigated. Secondly, a more powerful scale regarding active information search was developed. Thirdly, perceptual homophily and sender characteristics were added to the model in a unique construct. Fourthly, to measure the effects of personal sources a classification of services is applied for the first time in WOM researches in this context. Data was subjected to an exploratory factor analysis (EFA) and reliability analysis in the first stage of the analysis. At the second stage, confirmatory factor analysis (CFA) was conducted and the model's hypotheses were tested by using structural equation modeling (SEM).

Keywords: Word-of-Mouth, Message Characteristics, Interpersonal Communications, Credence-based Service

JEL Classification: D12




* Correspondance author. Tel: +90 533 694 9409

E-mail address: [email protected] (L.Incirkus)


  1. Introduction

    Interpersonal communications are at least as important in purchase decisions for services as for products. A multitude of researchers have focused on service- oriented issues (Bansal and Voyer, 2000, p.172; Sweeney et al., 2008, p.358; Wangenheim and Bayon, 2004, p.1179; Gatignon and Robertson, 1986, p.535). However, some areas were neglected or less vigorously examined in WOM. From a relevant literature review, some fundamental issues identified and focused on in this study. Four key contributions are planned. Firstly, although it has attracted attention and criticism, there is no empirical examination regarding message characteristics. In this study, the effects of messages delivered by senders on purchase decisions are investigated. Secondly, a more powerful scale regarding active information search was developed. Thirdly, perceptual homophily and sender characteristics were added to the model in a unique construct. Fourthly, to measure the effects of personal sources a classification of services is applied for the first time in WOM researches in this context.


  2. Literature

    WOM is a concept which provides consumers information they need to make rational purchase decisions. WOM is a much-discussed topic in marketing literature. However, it has not lost its popularity; quite to the contrary, it keeps offering researchers new areas to investigate. It is generally assumed that marketing is building up an intended perception (i.e. an illusion) for the targeted customers. In contrast WOM gives the facts more realistically. From the customer's perspective, fact means an experiment with a product or service (Silverman, 2001, p.24). However, probability of disappointment after using the product or service causes confusion and high risk perception. Customer's need to know what to expect from their purchases is high, especially in services since they have no visible features to examine. However, messages delivered by other customers or independent experts can affect and guide customer's purchase decisions more reliably than commercial information, and shorten their decision process by eliminating the perceived risk and confusion. Messages delivered by other personal sources contain more qualified, dependable, need oriented, and important features that customers actively search. Information which serves to simplify customer decision making also increases sales (Silverman, 2001, p.24).

    We define WOM as “effects of customers to each other in any way regarding a product, a service, or a brand with non-commercial concerns.” Consumers communicate to each other about their experiences of purchases. These communications are valuable for both customers and companies; for customers, when making purchase decisions; for companies, to understand which WOM are successful in attracting customers.


  3. Conceptual Framework

    Because of its persuasiveness on consumers' choice, WOM is investigated extensively by generated models (Gilly et al. 1998, pp.88-90; Bansal and Voyer, 2000, p.172; Sweeney et al. 2008, p.358; Wangenheim and Bayon, 2004, p.1179; Gatignon and Robertson, 1986, p.535). This study is expected to broaden the scope of WOM model and to present some results useful for both researchers and managers.

    One of the important models was generated by Gilly and others. This study included both sender and receiver characteristics. Homophily was categorized as demographic and attitudinal (Gilly et al. 1998, pp.84-85). A more complex model was produced by Gatignon and Robertson. The model pictured a big scenario about interpersonal communication. Again, researchers considered both receiver and sender and effect of this process (Gatignon and Robertson, 1986, p.535). Another model was generated by Bansal and Voyer. They put into the model interpersonal and non-interpersonal elements. This model may be considered more powerful and explicative model about interpersonal communication (Bansal and Voyer, 2000, p.172). The model of this study was based on initial models. The following part explains how WOM model was broadened according to the literature review.


    1. Model for Study

      WOM is a concept which provides consumers information they need to make rational purchase decisions. This is a much-discussed topic in marketing literature. However, new areas which enter marketing agenda open new discussions for this essential component of marketing literature.

      To understand consumer behavior, it is important to understand interpersonal communications among consumers. Messages which comprise experimental transmission is an essential part of such communications. While many researchers emphasized the importance and the effectiveness of WOM in consumers purchase decisions, in the literature, there is no research directly focused on message content and message properties of WOM. This study tries to exhibit the components of effective WOM messages, and to show its effectiveness in after WOM decisions.

      This study both fulfills gaps in the WOM literature and offers an opportunity to use WOM more effectively in the marketplace. The developed model for the study is given above in Figure 1. This study’s aim is to investigate how consumers evaluate cues of messages while purchasing a service in a broadened WOM model. This model includes receiver and sender characteristics, perceived risk, and tie strength. Previous studies handled these factors as salient in this area. Further, customer’s willingness to search and evaluate information from personal sources was examined in the service context.


      image

      Figure 1. The Broadened Model for testing WOM effect in Different Services

      First contribution of this study is related to message characteristics. The main goal of consumers who are less knowledgeable is to obtain some useful cues, either affirmative or intrinsic, regarding products or services. Consumers who are willing to search for information want to reach at cues which may help them make the right decision regarding services. In this study, two services were used. They were selected according to a classification from the literature (Mitra et. al. 1999, p.209; Krishnan and Hartline, 2001, p.334). Choice of a movie theatre represented a search-based, and choice of a repair and maintenance shop an experience- based service (Incirkus, 2014, p.85). Predicting effects of message characteristics on consumers’ choice according to service categories may help managers to make appropriate strategies to mobilize their consumers’ network. This gap is called some researchers’ attention (Sweeney et al. 2008, p.358; Mazzarol et al. 2007, p.1489; Gatignon and Robertson, 1986, pp.536-537; Duhan et al. 1997, p.284); however, to our knowledge, there is no empirical research in the literature regarding the role of message characteristics either in product or service context. This study attempts to reveal the role of the message in a WOM process in the context of services.

      Second contribution of the study is related to how active search of customer for WOM affects customer's decision making. Customer's active search for WOM is examined empirically by Bansal and Harvir. It is found “that when WOM information is actively sought, it will have a greater influence on the receiver’s purchase decision than if it was not actively sought” and, though its statistical support is weak, active search is found to be triggered by strong tie. Although active engagement of receiver in WOM is expected to increase the effects of recommendation, statistical findings were based on a single item (Bansal and Harvir, 2000, pp.174-175). Thus, it was considered that the effects of active search on purchase decision needed more investigation. Studies conducted by Flynn and others, and by Reynolds and Darden revealed some items useful in this area (Flynn et al. 1996, p.146; Reynolds and Darden, 1971, p.453). Through these studies, a scale which contains five items was adapted. The scale includes behavioral tendency of consumer and how consumer is motivated for active search. Thereby, it is expected to investigate relationship between active search and purchase decision.

      Thirdly, perceptual homophily was combined with sender characteristics in this study. Because homophily dimensions are essential properties of individuals, they should be examined together. Additionally, although homophily is a trigger to interpersonal communication, it does not entirely overlap with strong tie. Thus, this dimension should be examined independently from tie strength. Additionally, same effects were empirically observed on purchase decision (Wangenheim and Bayon, 2004, p.1179); thus, homophily and expertise will be examined together.

      Additionally, classification of services was considered in this study. Mitra and others investigated information search activity in services in a triangle search, experience, and credence-based dimensions (Mitra et al. 1999, p.209). Krishnan and Hartline developed a selection process for predicting services according to their categories (Krishnan and Hartline, 2001, p.334). Incirkus used this process with several changes and selected four services for three service categories (Incirkus, 2014, p.85). Movie theatre and repair and maintenance shop were selected according to the results of the mentioned study.


    2. Sender Characteristics (SCs)

      One of the premises of this study is that SCs affect both receiver’s information seeking activity and his subsequent purchase decision. The major SCs investigated in earlier research encompass expertise (Bansal and Voyer 2000, 169; Gilly et al. 1998, p.85; Brown et al. 2007, p.6; Wangenheim and Bayon, 2004, p.1180), credibility (Wu and Wang, 2011, p.452), similarity (Gilly et al. 1998, p.85; Wangenheim and Bayon, 2004, p.1175; Sweeney et al. 2008, p.353), experience (Silverman, 2001, p.27; Braunsberger and Munch, 1998, p.25), and opinion leadership (Hawkins et al. 1995, p.165; Kotler and Keller, 2006, p.178; Gilly et al. 1998, p.85).

      This study proposes a new hypothesis and a new contribution to SCs construct. Firstly, it is hypothesized that SCs affect perception of the message; in other words, the investigation of the effects of SCs on message is an important contribution to WOM research. In their qualitative research, Sweeney et al. found that perceived credibility of the WOM sender is important for WOM outcome (Sweeney et al. 2008, p.356). While their emphasis was primarily on the effectiveness of message characteristics, they drew attention to this point nevertheless. Secondly, the theoretical model of this study incorporates homophily (similarity) into the SCs construct. A scale which contains seven items was developed based on aforementioned studies.

      Based on the aforementioned SCs examination, the following hypotheses were generated:

      H1: The stronger the SCs, the greater the customer's active search for WOM (CASWOM).

      H2: The stronger the SCs, the stronger the influence of the sender’s WOM on the receiver’s purchase decision.

      H3: The stronger the SCs, the stronger the persuasiveness of message characteristics (MCs).


    3. Receiver Characteristics (RCs)

      Factors which determine perceived risk (PR) are customer’s performance expectation from the product/service, uncertainty, and the level of familiarity. Consumers find an opportunity to reduce risk perception via information search. Factors of RCs emerge as expertise or prior knowledge (Bansal and Voyer, 2001, p.170; Gilly et al. 1998, p.86; Duhan et al. 1997, p.286) and preference of WOM source for risk reduction (Gilly et al. 1998, p.86; Duhan et al. 1997, p.284).


      WOM literature drew little attention to what happens when a receiver accepts the WOM message and to the conditions that enhance receiver’s perceptions or actions. In other words, receiver outcomes which follow WOM message delivery and dimensions that are likely to enhance the probability of receiver actions on such WOM should be examined (Sweeney et al. 2008, p.344). This study posits that RCs have positive relationship with customer's active search for WOM and WOM influence on customer decision making, and negative relationship with perceived risk and with perception of received message.

      Developed hypotheses for RCs are given below:

      H4: The stronger the RCs, the greater the CASWOM.

      H5: The stronger the RCs, the stronger the WOM influence on purchase decision (WIN).

      H6: The stronger the RCs, the weaker the perception of MCs.

      H7: The stronger the RCs, the weaker the receiver’s perceived risk (PR).


    4. Perceived Risk (PR)

      PR indicates negative feelings when buying and consuming products/services because of the uncertainty or/and undesirable consequences; i.e., overall risk. One definition of perceived risk: “The uncertainty that consumers face when they cannot foresee the consequences of their purchase decisions” (Schiffman and Kanuk, 2000, p.153). Another perspective considers “perceived risk as representing the anxieties felt because the consumer cannot anticipate the outcomes of a purchase, but believes that there may be negative consequences” (Berkovitz et al. 2000, p.158). These definitions refer to uncertainty and consequences as reasons of perceived risks; however, there are various types of perceived risks. The types of risks that consumers perceive while making product decisions comprise physical risk, financial risk, functional risk, social risk, psychological risk, and time risk.

      Perceived risk presents conflicting results. This problem originates from the existence of many types of risk whose effects vary for different products/services, conditions, and consumers. Focusing on certain specific risks or on the overall risk could be a solution. Bansal and Voyer were obliged to break their perceived risk construct and to consider overall risk with a single item which was included as battery to their perceived risk construct (Bansal and Voyer, 2000, pp.172-173). We too will use, as battery to the perceived risk construct, two items which include uncertainty and undesirable consequences of service purchase.


      In this study, we will examine relationship of PR with MCs, CASWOM, and WIN. Developed hypotheses for PR examination are given below:

      H8: The higher the PR, the more extensive the CASWOM.

      H9: The higher the PR, the stronger the perception of MCs.

      H10: The higher the PR, the stronger the WIN.

    5. Tie Strength (TS)

      TS is an important construct for the WOM process. There are many studies indicating how relationships among people affect marketing knowledge flow from one source to receiver. There are findings in relevant literature that TS is an effective facilitator of knowledge dissemination in the marketplace. Nonetheless, sometimes some inconsistencies on the measurement of TS types may be observed.

      Granovetter defined TS as “a combination of the amount of time, the emotional intensity, the intimacy (mutual confiding), and the reciprocal services which characterize the tie” (Granovetter, 1973, p.1361). Masden and Campbell in their study, modeled strength:

      “Strength is a one-dimensional unobserved concept or "point variable" intervening between its predictors (neighbor, co-worker, and kinship statuses, overlapping organizational memberships, and measures of social distance), and its indicators (closeness, duration, frequency, breadth of discussion topics, and confiding)” (1984, p.490).

      In their study, indicators and predictors are used as two measurement tools; indicators as actual components of TS and predictors as aspects of relationships that are related to but not components of TS (Marsden and Campbell, 1984, pp.485-488). Frenzen and Davis, used closeness, intimacy, support, and association as the indicators of TS (Frenzen and Davis, 1990, p.6).

      We will investigate, in the present study, the relationship between TS and MCs, CASWOM, WIN by using the indicators emerged in the literature.

      H11: The stronger the TS, the more powerful the appreciation of MCs.

      H12: The stronger the TS, the higher the probability of benefiting from for CASWOM.

      H13: The stronger the TS, the higher the probability of WIN.


    6. Customer's Active Search for WOM (CASWOM)

      Factors increasing CASWOM and CASWOM’s effect on purchase decisions are the foci of interest in this part. It could define the activity as “Actively seeking WOM is construed as the process of vigorously seeking and ultimately attaining a message” (Bansal and Harvir, 2000, p.167). Whether SCs, RCs, MCs, PR, and TS have impact on how eagerly WOM information is searched are the issues to be examined. If customers are actively involved in information search from external sources, it is expected that WOM related message will be more effective on customer's purchase decision.

      Sender characteristics are accepted as triggers of CASWOM. While Bansal and Harvir found the relation between sender expertise and active information search to be less significant (Bansal and Harvir, 2000, p.174), source (sender) trustworthiness is found to be a factor which affects opinion change of receiver by Hovland and Weiss (Hovland and Weiss, 1951, p.647).

      Perceived risk is another construct which is interconnected to CASWOM. Services can be categorized according to their easiness-difficulty in purchase decision: namely, search, experience, and credence. Experience and credence services are more difficult to consider than search ones. Lack of prior knowledge increases as one moves from search services to credence services. These conditions cause an increase in perceived risk, and thus complicate problem solving (Mitra et al. 1999, pp.210-212). Mitra et al., in their study, showed that perceived risk increases as decision making becomes more difficult. However, there is little support for increased search intentions and search time as decision making gets difficult (Mitra et al. 1999, p.222). Another study lent support to the argument that perceived risk and customer's active search for WOM are positively correlated (Bansal and Harvir, 2000, p.174).

      For both expert and non-expert customers, search for information exhibits differences. Selnes and Troye compared customers according to their level of expertise. They investigated how customers search information and solve their problems. They found that experts tended to acquire larger amounts of and more sensory information than non-experts. Moreover, experts utilize such information more successfully than others (Selnes and Troye, 1989, p.425). Receiver expertise correlates consistently with information search, if they have greater prior knowledge, this will encourage them to seek information because they are confident that they will find the best solution (Bansal and Harvir, 2000, p.174). Tie strength is asserted as influential on the amount of information sought in Bansal and Harvir’s study. Though its significance is barely acceptable (0.10), statistical support is attained. Also, actively sought information is found to be influential on customer's purchase decision (Bansal and Harvir, 2000, pp.174-175). Thus the hypothesis:

      H14: The higher the CASWOM, the greater the impact of WIN.


    7. Message Characteristics (MCs)

      Although message is essential in the WOM process, since its internal dimensions increase effectiveness of a given WOM, it has not been investigated comprehensively. WOM, as peer-to-peer communication, has every feature of a two-sided communication process. Encoding-decoding relationship strongly takes place in a WOM process. Therefore, understanding the MCs can help marketers to produce more insider information and the flow of the information in question.

      In the WOM literature, there is no research which directly focuses on the message or information per se. Sweeney et al. emphasize that the MCs of the WOM itself have not been extensively considered (Sweeney et al. 2008, p.348). We anticipate seeing and showing the direct effects of MCs to CASWOM and WOM influence on purchase decision. This study may not expose every aspect of the effects of message in WOM context; nevertheless, it will help to develop a view about what their real roles are in a WOM process. Especially, as WOM is regaining popularity, the message deserves more attention as a complementary construct of WOM process. Though this study’s focus is not online WOM, it also makes some contributions to e-WOM context.

      As a result of the preceding review, the following hypotheses are established:

      H15: The stronger the MCs, the greater the CASWOM

      H16: The stronger the MCs, the more powerful the WOM influence on purchase decision.


    8. WOM Influence on Customer Decision Making

      Finally, a scale was generated for understanding the influence of WOM factors' effects on actual purchase decision of customers. Thus, it will be shown how receiver may be affected by the message delivered, characteristics of senders who deliver the messages, PR felt by receivers, the characteristics and search proneness of receivers.


  4. Research Design and Results

    This study was conducted on the European side of Istanbul. For the study, a convenience sample was used. It was possible to conduct the fieldwork by institutions which offered these services. However, this would create a risk that the study might be perceived as a satisfaction survey and institution dependent. This might cause a strong participant bias. Thus, a convenience sample was preferred. For the survey, suitable subjects for Movie Theater were found around movie theaters by pollsters. For choice of place of Repair and Maintenance Shop, fieldworkers focused around areas with repair and maintenance shop concentration. People found according to our research criteria were interviewed. The final sample included 201 data per sector, or a total of 402 data.

    To determine how many observations will be sufficient for a good SEM model, size of observations may be used. However, sample size is not independent from other parts of the research design. MacCallum et al. showed that “it is not possible to make blanket recommendations regarding this issue without considering other important aspects of design” (MacCallum et al. 2001, p.636).

    Sample size was determined based on the discussion in the literature. 20:1 subjects-to-variables ratio and other suggestions were examined. At the end, 250 observations per sector or a total of 500, was found appropriate. This choice was made based on two suggestions for the sample size. For factor analysis a rating scale for sample size: 100=poor, 200=fair, 300=good, 500=very good (Comray and Howard 1992, 217). For SEM, Hair and others suggest at least 150 cases (minimum sample size) for models with seven or fewer constructs, modest communalities (0.5), and no underidentified constructs (Hair et al. 2010, p.662).


    1. Demographics of Respondents by Sectors

      In determining the demographics of participants, four questions were asked. These questions contained gender, education, age, and family income. Distributions of demographics are given in Table 1. Gender distribution of participants differed according to service type. Obtained distribution for Movie Theater and Repair and Maintenance Shop male participants were in the majority; 157 for Movie Theater (78.1%), and 127 for Repair and Maintenance Shop (63.2%). Participants who have high school and college degree were in the majority in two sectors. Those under 34 years of age constituted 82.00% for movie theatre and 68.7% for repair and maintenance shop.


      Table 1. Demographics of Participants (n=201 per sector)



      Movie Theater

      Repair & Maintenance Shop

      Gender

      Number

      Percentage

      Number

      Percentage

      Males

      157

      78.1

      127

      63.2

      Female

      44

      21.9

      74

      36.8

      Total

      201

      100

      201

      100

      Education



      Primary & Secondary

      29

      14.4

      57

      28.4

      High School

      127

      63.2

      101

      50.2

      College / University

      45

      22.4

      43

      21.4

      Graduate Education

      0

      0

      0

      0

      Total

      201

      100

      201

      100

      Age



      18-24

      104

      51.7

      53

      26.4

      25-34

      61

      30.3

      85

      42.3

      35-44

      20

      10.0

      40

      19.9

      45-54

      11

      5.5

      13

      6.5

      55 +

      5

      2.5

      10

      5.0

      Total

      201

      100

      201

      100

      Family Income



      1000 TL or lower

      21

      10.4

      4

      2.0

      1001 – 2000 TL

      73

      36.3

      74

      36.8

      2001– 3000 TL

      68

      33.8

      109

      54.2

      3001- 4000 TL

      28

      13.9

      13

      6.5

      4001- 5000 TL

      7

      3.5

      0

      0

      5001 and more

      4

      2.0

      1

      0.5

      Total

      201

      100

      201

      100


    2. Missing Value

      For the obtained data, firstly, missing value analysis was conducted for each sector. A systematic mistake happened while copying questionnaires. For RCs, 4 items were arranged suitable for Repair and Maintenance Shop sector. However, in the process of preparing for fieldwork, one question was coded as explanation of study. Unfortunately, fieldworkers did not ask this question to the participants. Thus, RCs 4 was excluded directly from analysis for Repair and Maintenance Shop.

      All other results showed that the data had very few missing values. Researchers are concerned with whether the cause of missing value comes from the research design. According to this approach, missing data are named as "known" if they result from the research design, and "unknown" if not. Known missing values can occur due to procedural factors. Thus, remedies don't solve this problem overwhelmingly. Unknown missing values are related to the respondent, generally (Hair et al. 2010, pp.45-47). Our missing values could be labeled as unknown missing values, and the amount of missing could be neglected.


  5. Exploratory Factor Analysis and Testing for Reliability

    Two types of validity were achieved for the WOM model in two service sectors of the study; namely, content validity and construct validity. In construct validity, convergent and discriminant validity were checked. Satisfactory results were reached in the validity checks for all services. A two-stage analysis was preferred as measurement model. First, exploratory factor analysis was applied. This stage contained principal component analysis and reliability analysis by using Statistical Package for the Social Sciences (SPSS) 20.0. This stage helped purify scales of the study. In the second stage, Confirmatory Factor Analysis was applied by AMOS 20.0. In this analysis, scales obtained and composed from retained items were used.


    1. Movie Theater, Exploratory Factor Analysis and Test for Reliability

      An exploratory factor analysis was applied for 38 items which compose seven constructs. After exploratory analysis, all of the seven constructs, except SCs, were kept intact for the second stage. Only one item, SCs 7, from SCs construct was excluded, because this item loaded on two constructs; it loaded on SCs construct and on eighth factor as a single item at the same time. This item was related to homophily. A summary of the results is presented in Table 2 according to their eigenvalues.

      For PR, an alpha 0.948 was attained and all items were loaded perfectly on one construct. General five items of PR (Financial, Functional, Physical harm, Psychological, and Social) and two items which related to difficulties in decision making and level of general risk regarding this service composed this construct. PR is most influential construct in this model for Movie Theater. It covers 26.312% of the total variance explained. 7 items of the SCs construct loaded on one construct except SCs 7. It loaded on both factors. One was SCs factor, and other was eighth factor which SCs 7 loaded solely. This item was excluded from construct. Alpha value of SCs factor is 0.886.


      Table 2. Factor Analysis (EFA) Results for Movie Theater


      Constructs & Measurement Items

      Factor Loadings

      Means

      St. Deviations

      Perceived Risk

      PR01 Financial

      0.875

      2.701

      1.331

      PR02 Functional

      0.922

      2.736

      1.275

      PR03 Physical

      0.859

      2.731

      1.326

      PR04 Psychological

      0.906

      2.665

      1.289

      PR05 Social

      0.843

      2.680

      1.248

      PR06 Decision Hardship

      0.881

      2.690

      1.230

      PR07 Risky Decision

      0.803

      2.700

      1.257

      Sender Characteristics




      SCs01 Knowledge

      0.582

      3.755

      0.880

      SCs02 Perfect utilization

      0.679

      3.811

      0.924

      SCs03 Trustable

      0.650

      3.781

      1.016

      SCs04 Affecting others

      0.802

      3.781

      0.991

      SCs05 Strong experience

      0.827

      3.840

      0.908

      SCs06 Homophily; Life-style

      0.733

      3.755

      0.987

      Customer's Active Search for WOM




      CAS01 Intention

      0.731

      3.622

      1.028

      CAS02 Mutual consideration

      0.803

      3.875

      0.888

      CAS03 Decided approach

      0.774

      3.726

      0.980

      CAS04 Eagerness

      0.853

      3.692

      0.919

      CAS05 Planned asking

      0.820

      3.706

      0.910

      Message Characteristics




      MCs01 Vividness

      0.620

      3.785

      0.882

      MCs02 Confidence of sender

      0.624

      3.945

      0.850

      MCs03 Affective activation

      0.778

      3.884

      0.861

      MCs04 Richness

      0.791

      3.960

      0.830

      MCs05 Clearness

      0.637

      3.846

      0.917

      MCs06 Gain and loss

      0.643

      3.881

      0.875

      WOM Influence on Purchase Decision




      WIN01 Different ideas

      0.720

      3.761

      0.996

      WIN02 Changed my decision

      0.794

      3.856

      0.839

      WIN03 How to utilize

      0.738

      3.876

      0.836

      WIN04 More quality service

      0.669

      3.945

      0.716

      WIN05 Helped for right decision

      0.710

      3.891

      0.915

      Tie Strength

      TS01 Closeness

      0.767

      3.647

      1.100

      TS02 Confidence

      0.855

      3.731

      0.973

      TS03 Sharing free time

      0.838

      3.776

      0.956

      TS04 Mutualization

      0.809

      3.811

      0.987

      Receiver Characteristis




      RCs01 Knowledge

      0.869

      3.841

      0.946

      RCs02 Knowledge by Vendor

      0.854

      3.836

      0.893

      RCs03 Purchase experience

      0.853

      3.760

      0.934

      RCs04 Usage experience

      0.852

      3.825

      0.946


      CASWOM contains 5 items. Items were covering intentionality, readiness, motivation, and planned activation regarding customer's active search. In the exploratory analysis, these items loaded perfectly on their construct with an alpha 0.889 which is high. Thus, a construct was built for this model and for future studies. Variance explained by CASWOM is 8%.

      6 MCs items also loaded on their construct perfectly with an alpha 0.857. Thus all MCs items were retained for the second stage analysis. WIN contains 5 items which were indicating WOM influence on consumer purchase decision. They were loaded perfectly on WIN factor with an alpha 0.854. 4 items for TS are retained for second stage of analysis. Alpha attained for TS is 0. 902. Similarly, 4 items were used to measure RCs. All items loaded on their factor with an alpha 0.896.

      Table 3. Reliability, Eigen Value, Explained Variance for Movie Theater


      Movie Theater

      Cronbach's Alfa

      Eigen Value

      Explained Variance (%)

      Perceived Risk (PR)

      0.948

      9.735

      26.312

      Sender Characteristics (SCs)

      0.886

      5.663

      15.306

      Customer's Active Search for WOM (CASWOM)

      0.889

      2.960

      8.000

      Message Characteristics (MCs)

      0.857

      2.538

      6.859

      WOM Influence on Purchase Decision (WIN)

      0.854

      2.312

      6.249

      Tie Strength (TS)

      0.902

      1.614

      4.362

      Receiver Characteristics (RCs)

      0.896

      1.464

      3.956

      Measures




      Total Variance Explained



      71.043

      KMO Measure of Sampling Adequacy



      0.818

      Barlett's Test of Sphericity (sig.)



      0.000


      Total variance explained presents the common factors computed by principal component analysis. Further, the eigenvalues associated with related factors, the percentage of total variance computed for each factors, and the cumulative percentage of the total variance computed for each factor were presented (Ho, 2006, p.219). In this model 71.043% of the total variance was explained by 7 factors. Barlett's test of sphericity was used for confirming the appropriateness of data for factor analysis. Barlett's test serves to see that the variables were uncorrelated in the population. Relevant results presented confirmed the adequacy of the correlation matrix (sign:0.000). As a useful statistic, Kaiser- Meyer-Olkin (KMO) measure of sampling adequacy was reported. If KMO statistic values are smaller than 0.5, this points out that the correlations between pairs of variables cannot be clarified by other variables. Therefore, this condition may imply that factor analysis may not be appropriate (Hair et al. 2010, 104). From the data, 0.818 KMO statistics was computed. Thus, result can be interpreted as commendable. The results are presented in Table 3.


    2. Repair and Maintenance Shop, Exploratory Factor Analysis and Test for Reliability

      A Principal Component Analysis was conducted for the items measuring seven constructs for this stage of the analysis. Only three items were excluded from two different constructs. One was from RCs construct, and other two were from SCs construct. The results of conducted Exploratory Factor Analysis and other important measures are reported in Tables 4 and 5.

      Table 4. Reliability, Eigen Value, Explained Variance for Repair and Maintenance Shop


      Movie Theater

      Cronbach's Alfa

      Eigen Value

      Explained Variance (%)

      Perceived Risk (PR)

      0.974

      6.651

      19.004

      Customer's Active Search for WOM (CASWOM)

      0.919

      6.435

      18.386

      Tie Strength (TS)

      0.975

      3.980

      11.373

      Message Characteristics (MCs)

      0.868

      2.955

      8.441

      WOM Influence on Purchase Decision (WIN)

      0.903

      2.727

      7.790

      Receiver Characteristics (RCs)

      0.959

      2.228

      6.365

      Sender Characteristics (SCs)

      0.793

      1.731

      4.944

      Measures




      Total Variance Explained



      76.304

      KMO Measure of Sampling Adequacy



      0.846

      Barlett's Test of Sphericity (sig.)



      0.000


      Constructs retained without lost items are PR with an alpha 0.974, CASWOM with an alpha 0.919, TS with an alpha 0.975, MCs with an alpha 0.868, WIN with an alpha 0.903. For RCs 4 items were arranged. However, in the process of preparing for fieldwork, one question was coded as explanation and instruction of the study. Unfortunately, fieldworkers did not ask this question to the participants. Thus, RCs 4 was excluded directly from analysis for Repair and Maintenance Shop. The other 3 items loaded on their constructs perfectly with an alpha 0.959.


      Table 5. Factor Analysis (EFA) Results for Repair and Maintenance Shop


      Constructs & Measurement Items

      Factor Loadings

      Means

      St. Dev.

      Perceived Risk

      PR01 Financial

      0.896

      2.950

      1.121

      PR02 Functional

      0.921

      2.771

      1.161

      PR03 Physical

      0.932

      2.746

      1.179

      PR04 Psychological

      0.911

      2.856

      1.168

      PR05 Social

      0.913

      2.816

      1.141

      PR06 Decision Hardship

      0.951

      2.871

      1.142

      PR07 Risky Decision

      0.960

      2.905

      1.130

      Customer's Active Search for WOM




      CAS01 Intention

      0.813

      3.751

      0.792

      CAS02 Mutual consideration

      0.854

      3.841

      0.821

      CAS03 Decided approach

      0.836

      3.836

      0.740

      CAS04 Eagerness

      0.872

      3.886

      0.750

      CAS05 Planned asking

      0.890

      3.861

      0.735

      Tie Strength

      TS01 Closeness

      0.935

      3.159

      1.037

      TS02 Confidence

      0.954

      3.100

      1.058

      TS03 Sharing free time

      0.956

      3.060

      1.052

      TS04 Mutualization

      0.961

      3.085

      1.038

      Message Characteristics




      MCs01 Vividness

      0.725

      3.995

      0.418

      MCs02 Confidence of sender

      0.740

      4.154

      0.575

      MCs03 Affective activation

      0.670

      4.040

      0.582

      MCs04 Richness

      0.737

      4.144

      0.524

      MCs05 Clearness

      0.828

      4.129

      0.560

      MCs06 Gain and loss

      0.814

      4.159

      0.543

      WOM Influence on Purchase Decision




      WIN01 Different ideas

      0.760

      4.020

      0.469

      WIN02 Changed my decision

      0.809

      4.065

      0.557

      WIN03 How to utilize

      0.828

      4.030

      0.457

      WIN04 More quality service

      0.881

      4.045

      0.493

      WIN05 Helped for right decision

      0.849

      4.055

      0.482

      Receiver Characteristis




      RCs01 Knowledge

      0.963

      2.706

      0.989

      RCs02 Knowledge by Vendor

      0.945

      2.806

      0.937

      RCs03 Purchase experience

      0.951

      2.806

      0.926

      Sender Characteristics




      SCs01 Knowledge

      0.631

      3.925

      0.547

      SCs02 Perfect utilization

      0.706

      4.050

      0.555

      SCs04 Affecting others

      0.648

      4.035

      0.569

      SCs06 Homophily; Life-style

      0.767

      3.940

      0.630

      SCs07 Homophily; Like or Dislike

      0.760

      3.990

      0.600


      SCs contained 7 items. In this stage SCs 3 and SCs 5 were excluded. After eliminating these items, one factor was reached from 5 items with an alpha 0.793. Total variance explained was high: 76.304%. Barlett's test for sphericity was significant (0.000), and KMO statistic computed for this analysis was 0.846 can be interpreted as commendable. Consequently, 35 items and seven constructs were retained for the second stage of the analysis.


  6. CFA Results for Study

    Second stage of the analysis comprised two steps. First step was a CFA which was conducted according to precedent EFA. The second step was testing study hypotheses by SEM.


    1. Movie Theater CFA results

      A CFA was conducted on the scales involving the measures that were retained for the analysis of EFA in the first stage. The confirmatory factor analysis was conducted with AMOS. Multiple fit indexes were reported for the assessment of the model. Chi- square, chi-square/degrees of freedom (χ2/df), Goodness-of-Fit Index (GFI), root mean square error of approximation (RMSEA), and Comparative Fit Index (CFI) were presented as fit indices. Fit indexes before analysis were needed to be developed (χ2=1581.464, p< 0.05, df= 608, χ2/df= 2.601, GFI = 0.708, AGFI=0.662, IFI = 0.825, TLI = 0.807, CFI = 0.824, and RMSEA = 0.089).

      After analysis, generally, the reported fit indexes pointed out that the model was reasonably consistent with the data. According to the recommended values, all the fit indexes were close to, or better than, these values (χ2=319.073, p< 0.05, df= 166, χ2/df= 1.922, GFI = 0.867, AGFI = 0.814, IFI = 0.941, TLI = 0.924, CFI =0.940, and RMSEA = 0.068). Final measure was about Hoelter index. The Hoelter index calculates a sample size, at which models were considered meaningfully significant. Specifically, it presents an estimation of sample size which is sufficient for yielding an adequate model fit for a χ2 test. "This fit statistic differs substantially from those previously discussed in that it focuses it focuses directly on the adequacy of sample size, rather than on model fit" (Byrne 2009, 83). AMOS gives Hoelter index which sign the critical N for accepting model at level 0.05 and 0.01. Hoelter index for Movie Theater were 124 for 0.05 and 133 for 0.01.

      In addition to assessing the validity of measures, results of the analysis are shown in Table 6 with factor loadings. According to the performed analysis, to increase CFA indexes, 21 items were retained for SEM testing.


      Table 6. CFA Results for Movie Theater Scales (A-Z)


      Construct

      Item

      Standardized Loading (λ)

      t-value (p<0.05)

      Customer's Active Search for WOM

      CAS05

      0.837

      Scaling

      CAS04

      0.881

      13.166

      CAS02

      0.753

      11.534


      MCs04

      0.645

      Scaling

      Message Characteristics

      MCs02

      0.685

      7.906


      MCs01

      0.842

      8.883


      PR04

      0.918

      Scaling

      Perceived Risk

      PR03

      0.874

      18.620


      PR02

      0.927

      20.906


      RCs04

      0.829

      Scaling

      Receiver Characteristics

      RCs03

      0.803

      11.732


      RCs02

      0.827

      11.946


      SCs05

      0.660

      Scaling

      Sender Characteristics

      SCs04

      0.752

      10.273


      SCs03

      0.920

      8.530


      TS03

      0.754

      Scaling

      Tie Strength

      TS02

      0.961

      13.950


      TS01

      0.875

      13.194

      WOM Influence on Purchase Decision

      WIN03

      0.733

      Scaling

      WIN02

      0.863

      10.913

      WIN01

      0.777

      10.219


    2. Repair and Maintenance Shop CFA Results

      CFA performed for Repair and Maintenance Shop presented satisfying results regarding model fit. Reported results before analysis were needed to develop (χ2 = 828.493, p<0 .05, df= 539, χ2/df= 11.537, GFI = 0.809, AGFI = 0.776, IFI =0.955, TLI = 0.950, CFI = 0.954, and RMSEA = 0.052).

      At this step, some of the items from different constructs were excluded according to their factor loadings and unnecessary contribution to the model. Therefore, CAS 1 from CASWOM construct, MCs 1, 3, and 4 from MCs construct, SCs 1 and 2 from SCs construct, and WIN 1 from WIN construct were excluded. After smoothing by modification indices results were increased according to first results (χ2 = 437.441, p< 0.05, df= 303, χ2/df= 1.444, GFI = 0.861, AGFI = 0.826, IFI= 0.976, TLI = 0.972, CFI = 0.976, and RMSEA = 0.047).

      Table 7. CFA Results for Repair and Maintenance Shop Scales (A-Z)


      Construct

      Item

      Standardized

      Loading (λ)

      t-value

      (p<0.05)


      CAS05

      0.921

      Scaling

      Customer's Active Search for WOM

      CAS04

      0.905

      19.603

      CAS03

      0.783

      14.675


      CAS02

      0.787

      14.842


      MCs06

      0.829

      Scaling

      Message Characteristics

      MCs05

      0.909

      12.979


      MCs04

      0.691

      10.411


      PR07

      0.977

      Scaling


      PR06

      0.969

      41.277


      PR05

      0.919

      28.966

      Perceived Risk

      PR04

      0.882

      24.289


      PR03

      0.916

      28.572


      PR02

      0.892

      25.403


      PR01

      0.852

      21.465


      RCs03

      0.952

      Scaling

      Receiver Characteristics

      RCs02

      0.918

      25.13


      RCs01

      0.955

      29.072

      Sender Characteristics

      SCs07

      0.891

      Scaling


      SCs06

      0.801

      7.326


      TS04

      0.968

      Scaling

      Tie Strength

      TS03

      0.956

      34.075

      TS02

      0.961

      35.32


      TS01

      0.929

      29.161


      WIN05

      0.841

      Scaling

      WOM Influence on Purchase Decision

      WIN04

      0.895

      15.35

      WIN03

      0.809

      13.415


      WIN02

      0.765

      12.395


      Hoelter index for Repair and Maintenance Shop were 158 for 0.05 and 167 for 0.01. Analysis was terminated because new attempts were inefficient and/or distorting some results occasionally, for instance RMSEA, CFI. Thus, measurement study was ended and results were recorded and presented in Table 7. These results were sufficient for the next step of the analysis.


  7. Findings by Hypotheses Testing

    SEM was used to test the hypotheses. This statistical methodology enables a confirmatory approach to the analysis of relevant structural theory (Byrne 2010, 3). Similar to multiple regression analysis, this method serves to reveal interrelationship among expressed series of equations (Hair et al. 2010, p.634). The term SEM conveys two important aspects of the procedures: (a) that the causal processes under study are represented by a series of structural (i.e., regression) equations, and (b) that these structural relations can be modeled pictorially to enable a clearer conceptualization of theory under study (Byrne, 2010, p.3). To determine the extent to which structured model is consistent with the data, models were tested by SEM.


    1. Movie Theater Testing of Hypotheses with SEM

      Sixteen hypothesized paths of the model were tested by using AMOS. Overall fit, analytical power, and the significance of the paths were measured for this model. All fit indexes present reasonable consistency with the data and the model. (χ2 = 376.810, p<0.05, df = 172, χ2/df= 2.191, GFI = 0.850, AGFI = 0.798, IFI= 0.921, TLI = 0.902, CFI = 0.919, and RMSEA = 0.077). Table 8 presents the path loadings with t values for each path. Figure 2 illustrates the model used in testing the hypotheses. According to the results of path analysis, it is observed that while some of the model hypotheses were supported strongly, others weren't found to be significant.

      H1, h3, and H3 were related to SCs and their effects on three endogenous variables of the model; CASWOM, WIN, and MCs. H1 was stated to observe the effects of sender characteristics on willingness of customers while searching for information. According to the results of path analysis, it was found that SCs affect CASWOM positively and significantly. Thus, H1 was supported. h3 was for investigation of relationship between SCs and WIN. Result was compatible with hypothesis. H3 was drawn to investigate effects of SCs on MCs. SCs, characteristics of message deliverer were found positively significant on the way messages were perceived.

      H4, H5, H6, and H7 comprised the relationship of RCs with other model variables. H4, H5, and H6 were found insignificant. They were drawn for understanding how characteristics of the message receiver affect customer's search willingness, purchase decision, and perception of message characteristics.

      H7 proposed that there was a negative relationship between RCs and risk perception. Results indicated a consistency with the hypothesis. Thus H7 was supported. H8, H9, and H10 were hypotheses for investigating the role played by perceived risk in information search, decision making, and persuasion of message. All three proposals were found insignificant. Thus, H8, H9, and H10 were not supported. Hypotheses for seeking relationships between TS and CASWOM, WIN, and MCs, respectively, were H11, H12, and H13. All three were significant in contrast to PR related hypotheses. H11 which states that perception of message characteristics is positively linear with tie strength was supported strongly. Similarly, H12 was supported because an expected effect of TS on WIN was found to be significant and positively related. H13 which draw relationship with WIN was found significant; however, the direction of effect was found to be negative. This was contrary to expectations.


      image

      Figure 2. Path Coefficients for the Model of Movie Theater

      H14 involved the investigation of CASWOM-WIN relationship. CASWOM effect on WIN was found significant and positive. This presents appropriateness with purpose of hypothesis. Thus, H14 was supported. H15 and H16 were hypotheses to understand effects of MCs on CASWOM and WIN. It was found that MCs don't trigger information seeking activities of customers. The relationship between MCs and CASWOM was not significant. Thus, H15 was not supported. H16 was supported strongly. It was found that MCs affects WIN positively and significantly.


      Table 8. Path Analysis Results for Movie Theater


      Hypothesized Relationship

      Coefficient

      t-Stat.

      Results

      H1

      Sender Characteristics →

      Customer's Active Search for WOM

      0.217**

      2.339

      Supported

      h3

      Sender Characteristics →

      WOM Influence on Purchase Decision

      0.340***

      4.394

      Supported

      H3

      Sender Characteristics →

      Message Characteristics

      0.355**

      5.696

      Supported

      H4

      Receiver Characteristics →

      Customer's Active Search for WOM

      0.062

      0.838

      Not

      supported

      H5

      Receiver Characteristics →

      WOM Influence on Purchase Decision

      0.020

      0.370

      Not

      supported

      H6(-)

      Receiver Characteristics →

      Message Characteristics

      0.035

      0.762

      Not

      supported

      H7(-)

      Receiver Characteristics →

      Perceived Risk

      -0.297**

      -2.521

      Supported

      H8

      Perceived Risk →

      Customer's Active Search for WOM

      0.016

      0.339

      Not

      supported

      H9

      Perceived Risk →

      Message Characteristics

      -0.012

      -0.421

      Not

      supported

      H10

      Perceived Risk →

      WOM Influence on Purchase Decision

      0.015

      0.449

      Not

      supported

      H11

      Tie Strength →

      Message Characteristics

      0.308***

      5.262

      Supported

      H12

      Tie Strength →

      Customer's Active Search for WOM

      0.221**

      2.375

      Supported

      H13

      Tie Strength →

      WOM Influence on Purchase Decision

      -0.215***

      -2.980

      Supported

      H14

      Customer's Active Search for WOM →

      WOM Influence on Purchase Decision

      0.125**

      2.011

      Supported

      H15

      Message Characteristics →

      Customer's Active Search for WOM

      0.196

      1.159

      Not

      supported

      H16

      Message Characteristics →

      WOM Influence on Purchase Decision

      0.487***

      3.431

      Supported


      Goodness-of-Fit Statistics




      Χ2


      376.810. p< 0.05


      df


      172



      Χ2/df


      2.191



      Goodness of fit index (GFI)


      0.850



      Normed fit index (NFI)


      0.863



      Tucker-Lewis Index(TLI)


      0.902



      Comparative fit index (CFI)


      0.921



      Root mean square error of approximation (RMSEA)

      0.077


      Notes: *p<0.10; ** p<0.05; ***p<0.01


    2. Repair and Maintenance Shop Testing of Hypotheses with SEM

      Hypothesized paths of the model were tested for Repair and Maintenance Shop by using AMOS. Overall fit, analytical power and the significance of the paths were measured for this model. All fit indexes present reasonable consistency with the data and the model. All other measures surpassed recommended values. (χ2 = 354.021, p< 0.05, df = 301, χ2/df= 1.176, GFI = 0.891, AGFI = 0.863, IFI = 0.990, TLI= 0.989, CFI = 0.990, and RMSEA = 0.030). Table 9 present the path loadings with t values for each path. And Figure 3 illustrates the model used while testing the hypotheses.

      Fit indexes of the model for Repair and Maintenance Shop showed that data and model were consistent. However, for tested hypotheses, results don't exhibit expected indications. H1 and H3, from the three SCs related hypotheses, were supported. Positive effect of SCs on CASWOM was supported by significance lower than 0.05 values (sign. 0.047). Similarly, SCs effects on MCs were observed strongly in the analysis. Thus H3 was supported. However, SCs effect on WIN was not significant. Thus, h3 was rejected.

      From H4, H5, H6, and H7 which were related to RCs, only, H5 was partially supported. For the other three hypotheses, observed effects weren't significant.


      image

      Figure 3. Path Coefficients for the Model of Repair and Maintenance Shop


      Table 9. Path Analysis Results for Repair and Maintenance Shop


      Hypothesized Relationship

      Coefficient

      t-Stat.

      Results

      H1

      Sender Characteristics →

      Customer's Active Search for WOM

      0.208**

      1.982

      Supported

      h3

      Sender Characteristics →

      WOM Influence on Purchase Decision

      0.091

      1.385

      Not

      supported

      H3

      Sender Characteristics →

      Message Characteristics

      0.367***

      4.964

      Supported

      H4

      Receiver Characteristics →

      Customer's Active Search for WOM

      0.012

      0.224

      Not

      supported

      H5

      Receiver Characteristics →

      WOM Influence on Purchase Decision

      -0.06*

      -1.835

      Partially

      supported

      H6(-)

      Receiver Characteristics →

      Message Characteristics

      -0.032

      -0.892

      Not

      supported

      H7(-)

      Receiver Characteristics →

      Perceived Risk

      0.108

      1.198

      Not

      supported

      H8

      Perceived Risk →

      Customer's Active Search for WOM

      0.074*

      1.806

      Partially

      supported

      H9

      Perceived Risk →

      Message Characteristics

      0.009

      0.313

      Not

      supported

      H10

      Perceived Risk →

      WOM Influence on Purchase Decision

      -0.026

      -1.001

      Not

      supported

      H11

      Tie Strength →

      Message Characteristics

      -0.038

      -1.211

      Not

      supported

      H12

      Tie Strength →

      Customer's Active Search for WOM

      0.128***

      2.828

      Supported

      H13

      Tie Strength →

      WOM Influence on Purchase Decision

      0.045

      1.544

      Not

      supported

      H14

      Customer's Active Search for WOM →

      WOM Influence on Purchase Decision

      0.03

      0.642

      Not

      supported

      H15

      Message Characteristics →

      Customer's Active Search for WOM

      0.238**

      1.948

      Supported

      H16

      Message Characteristics →

      WOM Influence on Purchase Decision

      0.256***

      3.256

      Supported


      Goodness-of-Fit Statistics




      Χ2


      354.021, p<.05


      df


      321



      Χ2/df


      1.176



      Goodness of fit index (GFI)


      0.891



      Normed fit index (NFI)


      0.940



      Tucker-Lewis Index(TLI)


      0.989



      Comparative fit index (CFI)


      0.990



      Root mean square error of approximation (RMSEA)

      0.030


      Notes: *p<0.10; ** p<0.05; ***p<0.01


      Similar to the preceding group of hypotheses, from among the repeated for measuring PR effects in model, only H8 was partially supported. These results indicated that PR has positive effects on CASWOM, although significance of results was weak.

      H10, H11, and H12 were measuring effect of TS on MCs, CASWOM, and WIN. Only H12 was supported strongly. For the other two hypotheses, results were not significant. CASWOM effect on WIN was found to be insignificant. Thus H14 was rejected. H15 and H16 were supported. Two hypotheses indicate that MCs had positive and significant effect on CASWOM and WIN.


  8. Discussion and Managerial Implication

    1. Discussion

      Receiver's behaviors after message delivery or WOM occurrences were investigated via a broadened model. The classified two services were subjected to investigation. The results were compared according to service types. Movie Theater represented research-based services and repair and maintenance shop represented experience-based services.

      Current study allowed us to examine and compare MCs effects on service sectors. On the whole, MCs affected CASWOM for experience service and WIN for both. Inefficiency of MCs on CASWOM in search service associated with logical design. Since customers need less information for search service than experience service, fewer attempts to obtain new information is understandable. Mitra and others have found that behavioral intention for experience service was stronger than for search service. Nevertheless, search time and use of personal and impersonal sources, for gathering information in search services were higher than in experience services (Mitra et al. 1999, p.222). For support in CASWOM of experience service, our study showed that customers need more attempts when confusion prone. This may lead to a need of legitimization of his decision or verification of previously gathered information. It is clear that, according to the results, WOM messages drive customers to strong and profitable purchase decision.

      SCs were redesigned for this research. Perceptual homophily was added to SCs as an essential component. For search service, trustworthiness, experience, and influence of others' ideas were salient according to literature (Bansal and Voyer, 2000, p.169; Gilly et al. 1998, p.85; Brown et al. 2007, p.6; Wangenheim and Bayon, 2004, p.1180; Wu and Wang, 2011, p.452; Silverman, 2001, p.27; Braunsberger and Munch, 1998, p.25; Hawkins et al. 1995, p.165; Kotler and


      Keller, 2006, p.178), however, for experience service perceptual homophily was the salient factor (Gilly et al. 1998, p.85; Wangenheim and Bayon, 2004, p.1175; Sweeney et al. 2008, p.353). Gilly and others underline that effects of homophily depends on the product category. In services, from the point of SCs, it can be articulated that the results showed parallelism with product categories.

      The scale of CASWOM was redesigned and presented high solidity according to results of analyses. For search service, effects were found to be positive and significant, however, for experience services, H15 wasn't supported. Bansal and Harvir found that the greater the information searching, the greater the effect on purchase decision (Bansal and Harvir, 2000, p.174). On the light of findings, we can note that, new attempts didn't eliminate confusion in customer's mind.

      MCs were found effective for both sectors. Therefore, the role of MCs in purchase decision was verified empirically. The message emerged as a problem solving tool especially in confusion. However, results supported that customers relied on delivering information for both service.

      For the Movie Theater SCs play a key role. Because of knowledge level held by parties in interrelationship are approximately equal, in another words, they communicate easily. Senders give new information to receiver which lead him to evaluate and make new search. Customers may predict the results of a search based service purchases. Thus, new information doesn't cause confusion. As another view, customer may not reach new information, however, his relationship with sender can lead to reevaluate existing information. RCs didn't observed effects on WOM influence, and further information search, and message perception. This result may depend on initial knowledge of customer. Customer who has high level of knowledge, according to study results, cannot find new information. Thus, message, information, and further searching activities don't affect the attitude of receiver themselves.

      For the Repair and Maintenance Shop it can be deduced that tie strength doesn't affects decision of customer. Rather than customer evaluate sender’s personality and experience, and message itself. Despite moderate risk perception and low initial knowledge they don't attempt to generate more information. We can say that these customers are highly focused on MCs while shaping their decisions.


    2. Managerial Implications

      This study built on previous studies in the area of interpersonal communications. It attempted to clear recommendation-based decision making by using message characteristics from the receiver’s perspective in actual purchases. The presented models, influenced by studies from WOM area, provided additional insights.

      Shaped roles for message characteristics were observed clearly. Measuring message effects, through actual purchases and via an empirical study may give practitioners several clues to notice and to fulfill the gap regarding what customers want to hear. The study revealed that customers may be oriented according to cues they found in the messages, even though there may change for different services. For experience-based service rational components may attract customers. Clear, satisfying, and profitable solutions suggested by sender may drive customers to prefer a service provider.

      Especially for Repair and Maintenance Shop as an experience service, highest degree of confusion was observed. To overcome confusion, first experimental transmission and secondly tie strength was found to be effective. It was found that receiver characteristics affect perceived risk for research-based service but not, for experience service. This might be explained by confusion proneness. Results indicate that customers are eager to collect information. However, to affect such willingness, marketers can generate several reasons to talk about their services or use some tools which make visible their services. These types of applications help previous users to make explanations about and show usefulness of provided services.

      In experience based services, users can confirm that service was presented fairly, only after usage experience. Service providers may help satisfied customers to express and communicate their experience. Intrinsic cues regarding this type of services may trigger experience transmission; such as speed of the services, price applications, guarantees, free periodic controls for previous customers etc.

      A search-based service's performance may be estimated by the help of some useful explanations regarding physical evidence or service generation process. According to the results, an inexperienced customer may shape his decision on the basis of advice concerning rational evaluation of presented cues. Eagerness while delivering messages, self-confidence and content of message may help the choice of a search service. In practice, negative correlations between risk and receiver characteristics may be useful innovative applications. New versus old comparison may attract customer's attention, and may trigger trial intention.

      This study conceptualized WOM on the perspective of receiver. The changes of his decision were investigated. It was seen that messages had a key role in decision making. Especially in confusion proneness it is more effective than tie strength. Indirectly, this study reinforced the importance of experiment transmission. Further researches may show the appropriateness of developed scales and models.


    3. Limitations

The study had several limitations; albeit pointing out several avenues for further investigation. First of all, there was a systematic limitation in this study for experience-based service. One of the questions for receiver characteristics (RCs 4) was written as headline by mistake. Hence, this question wasn't asked by the fieldworkers, and receiver characteristics were measured by only three items.

Service selection was made after an intensive study. Two services seemed appropriate to their classification. However, high confusion proneness which exists in repair and maintenance shop service may cause a pressure in results. This is partially normal; nevertheless, new investigations may be more explanatory.

The role of MCs for different type of customers and for various services should be investigated. This may enhance message oriented word-of-mouth marketing approach. Therefore, business may find out different ways to deliver true side of reality.

Some scales in the WOM model were reshaped. These scales were tested, and they showed solidity in targeted measures. However, further researches may increase the strength of the scales' validity.


References

Bansal, S.H., and Voyer, P.A. (2000). Word-of-Mouth Processes within a Services Purchase Decision Context. Journal of Service Research, 3(2), 166-177.

Berkowitz, E.N., Kerin, R.A., Hartley, S.W., and Rudelius, W. (2000). Marketing. 6th edition, USA: Irwin McGraw-Hill.

Brown, J., Broderick, A.J., and Lee, N. (2007). Word of Mouth Communication within Online Communities Conceptualizing the Online Social Network. Journal of Interactive Marketing, 21(3), 2-20.

Byrne, B.M. (2009). Structural Equation Modeling with AMOS: Basic Concepts, Applications, and Programming. New York: Routledge.

Darby, R.M., and Karni, E. (1973). Free Competition and the Optimal Amount of Fraud. Journal of Law and Economics, 16(1), 67-88.

Duhan, D.R., Jhonson, S.D., Wilcox, J.B., and Harrell, G.D. (1997). Influences on Consumer Use of Word-of-Mouth Recommendation Sources. Journal of the Academy of Marketing Science, 25(4), 283-295.

Flynn, L.R., Goldsmith, R.E., and Eastman, J.K. (1996). Opinion Leaders and Opinion Seekers: Two New Measurement Scales. Journal of the Academy of Marketing Science, 24(2), 137-147.

Frenzen, J.K., and Davis, H.L. (1990). Purchasing Behavior in Embedded Markets. The Journal of Consumer Research, 17(1), 1-12.

Gatignon, H., and Robertson, T.S. (1986). An Exchange Theory Model of Interpersonal Communication. Advances in Consumer Research, 13, 534-538.

Gilly, C.M., Graham, J.L., Wolfinbarger, M.F., and Yale, L.J. (1998). A Dyadic Study of Interpersonal Information Search. Journal of the Academy of Marketing Science, 26(2), 83-100.

Granovetter, M.S. (1973). The Strength of Weak Ties. American Journal of Sociology, 78(6), 1360-1380.

Hair, J.F., William, C.B., and Barry, J.B. (2010). Multivariate Data Analysis: A Global Perspective. Pearson Education.

Hawkins, I.D., Best, R.J., and Kenneth, A.C. (1995). Consumer Behavior: Implication for Marketing Strategy. 6th edition, USA: Irwin.

Ho, Robert. 2006. Handbook of Univariate and Multivariate Data Analysis and Interpretation with SPSS. Boca Raton: Chapman &Hall/CRC.

Hovland, C.I., and Weiss, W. (1951). The Influence of Source Credibility on Communication Effectiveness. The Public Opinion Quarterly, 15(4), 635-650.

Incirkus, L. (2014). WOM Influence on Purchase Decisions in Search, Experience, and Credence Based Service Industries: A Message Oriented Approach. Unpublished doctoral thesis.

Krishnan, C.B., and Hartline, M.D. (2001). Brand Equity: Is It More Important in Services? Journal of Services Marketing,15(5), 328-342.

Kotler, P., and Keller, K.L. (2006). Marketing Management. 12th edition, New Jersey: Pearson Prentice Hall.

Marsden, P.V., and Karen, E.C. (1984). Measuring Tie Strength. Social Forces, 63(2), 482-501.

Mazzarol, T., Sweeney, J.C., and Soutar, G.N. (2007). Conceptualizing Word-Of- Mouth Activity, Triggers and Conditions: An Exploratory Study. European Journal of Marketing, 41(11/12), 1475-1494.

Mitra, K., Reiss, M.C., and Capella, L.M. (1999). An Examination of Perceived Risk, Information Search and Behavioral Intentions in Search, Experience and Credence Services. The Journal of Services Marketing, 13(3), 208-228.

Reynolds, F.D., and Darden, W.R. (1971). Mutually Adaptive Effects of Interpersonal Communication. Journal of Marketing Research, 8, 449-454.

Schiffman, L.G., and Kanuk, L.L. (2000). Consumer Behavior. 7th edition, New Jersey: Prentice Hall.

Schmitt, P., Skiera, B., and Bulte, C.V. (2011). Referral Programs and Customer Value. Journal of Marketing, 75, 46-59.

Selnes, F., and Troye, S.V. (1989). Buying Expertise, Information Search, and Problem Solving. Journal of Economic Psychology, 10, 411-428.

Silverman, G. (2001). The Secrets of Word of Mouth Marketing. New York: AMACOM books.

Sweeney, C.J., and Soutar, G.N. (2001). Consumer Perceived Value: The Development of a Multiple Item Scale. Journal of Retailing, 77, 203–220.

Sweeney, J.C., Soutar, G.N., and Mazzarol, T. (2008). Factors Influencing Word of Mouth Effectiveness: Receiver Perspectives. European Journal of Marketing, 42(3/4), 344-364.

Wu, P.C., and Wang, Y. (2011). The Influences of Electronic Word-of-Mouth Message Appeal and Message Source Credibility on Brand Attitude. Asia Pacific Journal of Marketing and Logistics, 23(4), 448-472.



Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.