Fuzzy cognitive map for optimizing solutions for ... - CyberLeninka

Available online at www.sciencedirect.com

Procedia - Social and Behavioral Sciences 57 (2012) 47 – 52

International Conference on Asia Pacific Business Innovation and Technology Management

Fuzzy cognitive map for optimizing solutions for retaining full-service restaurant customer Shiu-Chun Chen Chang Jung Christian University

Abstract This study aims to use the fuzzy cognitive map (FCM) to identify the decision factors most relevant in increasing repurchase rate for a full-service restaurant. FCM knowledge causality based on Structural Equation Model is represented by adjacency matrix where the enhancement certain factors affect other factors. To provide restaurant operators obtains the optimal solutions for activating the customer retention program.

© 2012 2012Published Publishedby byElsevier ElsevierLtd. Ltd.Selection Selection and/or peer-review under responsibility ofAsia the Asia © and/or peer-review under responsibility of the Pacific Business Innovation and Technology Management Society (APBITM) Pacific Business Innovation and Technology Management Society (APBITM).” Keywords: Fuzzy cognitive map, structural equation model, full-service restaurant

1. Introduction Service is produced and consumed simultaneously, and consumers frequently experience the service entirely within the physical facility of the restaurateur. [1] discussed three categories of indicators of present service experience: functional clues, mechanical clues, and human clues. Functional clues refer to the food itself (such as food quality) and to service accuracy and efficiency (such as waiting time for service) in a restaurant. Mechanical clues are nonhuman elements in the service environment, and comprise design and ambient factors. Human clues comprise service employee behavior. Based on [1], this study proposes three key restaurant attributes (i.e. food quality, servicescape, and waiting experience). The highly competitive environment means customers have little incentive to be loyal, and so restaurateurs must maintain low operating margins. Thus restaurant operators must understand how to provide key attributes in order to increase repurchase rate under the constraint of limited resource. Generally, restaurateurs use cognitive and context-sensitive judgments to design restaurants with specific attributes based on limitations of their understanding and situation. Restaurateurs cannot easily simultaneously analyze large volumes of different types of information. Fuzzy cognitive map (FCM) is useful for modeling complex systems [2], and can improve system performance. Structural Equation Modeling (SEM) can holistically explain the causal relationship of latent variables. But SEM remains insufficiently distinct to enable decision makers to identify the optimal solution for customer repurchase rate. In this study, adopting the FCM approach based on SEM designs a model of customer repurchase intention and establishes guideline that develop customer repurchase rate. * Corresponding author. Tel.: 886-8-7338241; fax: 886-8-7348663 E-mail address: [email protected].

1877-0428 © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the Asia Pacific Business Innovation and Technology Management Society (APBITM) doi:10.1016/j.sbspro.2012.09.1156

48

Shiu-Chun Chen / Procedia - Social and Behavioral Sciences 57 (2012) 47 – 52

2. Literature Review 2.1 Fuzzy Cognitive Map (FCM) FCM is a graphical representation based on the work of Axelrod on cognitive mapping [3], and originates from the combination of fuzzy logic and neural networks. The goal of FCM is to provide an effective mechanism for forecasting outcomes by letting relevant issues interact with one another. FCM (shows in Figure1) consists of nodes and weighted arcs. Nodes represent concepts or problems. Meanwhile, weighted arcs represent causality between pairs of concepts. If each concept is characterized by a number Bi, the edge eij is formed by the influence of causal concept Bi on concept Bj, and measures the degree to which Bi causes Bj. A positive weight indicates a positive relationship between the two nodes, zero indicates no relationship exists, and a negative weight indicates a negative relationship. Positive or negative signs and weights derive from the experts [4]. If specific nodes are stimulated, the resulting activities can resonate through other nodes on the map along positively or negatively weighted connections [5]. ġ

e12

B2

e23

e52 B1

B3 e53

e15

e43

e41 B5

B4 e54

Figure1 A typical FCM FCM is described using connection matrix, and the activation levels of its nodes can be represented as a state vector [4]. The matrix comprises row and column factors, and the corresponding causality coefficients between them are called the adjacency matrix. Row factors are cause factors and column factors are effect factors. The values of nodes B 1, B2,….Bn together represent state vector B, called ‘What-if’ performed based on decision-maker intention. The value of each element of the input vector can be 1 or 0 depending on whether a specific element is enhanced. For example, vector B (1101) means that the four nodes form FCM, with the 1st, 2nd, and 4th nodes being activated, while the 3rd node is inactive. Therefore, through what-if simulations, decision makers can identify a set of relevant decision variables and values. To compute an FCM state vector B at time step (t+1), the adjacency matrix F is multiplied by the state vector B (t). [4][6] addressed that a threshold function was applied as B (t+1) = S [B (t). F]. Where B (t) denotes the state vector (1×n) of the concept at some time step t. Meanwhile, F represents the adjacency matrix (n×n). FCM is then constructed by combining knowledge from numerous experts. The FCM of each expert is cumulatively superimposed, whereby [4][7] address that the equation thus is as seen below, where Fi represents the augmented FCM matrix for expert i, n denotes the number of experts, and wi is the credibility weight of expert i. n

F˙

¦w F i

i

t 1

FCM resembles human reasoning and the human decision-making process [8], and is easily adjusted to incorporate new phenomena [9]. FCM can be used to perform analysis, test the influence of parameters, and predict system behavior [5][9], identify fuzzy causal relations among factors [5]. FCM has also been applied to knowledge management [10], the development of behavioral models complex systems [11], relationship management among organization in airline services [12], design of controls in business-to-consumer e-commerce web-based systems [5]. The advantages of applying the formalism of FCM can be used in business decision-making applications, and can assist managers in


simultaneously and systematically analyzing the relationships among factors, and in understanding and easily quantifying the strength and direction of the interrelationships. 2.2 Restaurant Attributes Servicescape consists of three dimensions, including 1) ambient conditions, 2) spatial layout and functionality, and 3) signs, symbols and artifacts [13]. Employees are the main contact between restaurants and their customers, and cognitive responses and repurchase intentions [14]. Therefore, Servicescape comprises interior design, ambient factor, spatial layout and human elements in this study. Physical environment influenced customer evaluations of service quality and behavioral responses [15][16]. Customers with positive experience of restaurant servicescape are likely to evaluate service quality of a full-service restaurant positively. Waiting negatively affects consumer perceptions of quality, service evaluation, purchase intention and satisfaction [17][18]. The psychological aspects of waiting are very important, as is the associated experience [19][20]. Some practitioners also try to fill wait time or to influence perceived wait by providing entertainment facilities such as magazines, video games, and electric massage chairs entertain customers [21]. Waiting time is also known as a “time price”[22], and is a non-monetary price component associated with service acquisition [23]. Positive waiting experience does not create negative perceptions, but rather positively affects customer perceptions of overall service quality or value. Consumers form expectations towards food product attributes based on intrinsic or extrinsic stimuli. Consumers derive their perceptions of food quality from tastiness of food, menu variety, variety of food, food presentation, serving size, safety, appeal, dietary acceptability, healthy options, food freshness, temperature and cleanliness [24][25][26]. Food quality is a significant determinant of customer assessments of restaurants [27], and is essential to satisfy consumer needs and expectations [28]. 2.3 The relationships among perceived overall service quality, perceived value and customer behavior intention Customer behavior generally involves dynamic interactions and exchanges [29]. Perceived value is one of the most important and best understood customer behavior in the service industry [30][31]. However, customer choice results from multiple value perceptions [32][33]. The literature suggests that perceived value can be conceptualized as a multidimensional construct [32][33], for example through value as price, value as want fulfillment, value as a price-quality trade-off, and value as the culmination of what is obtained versus what is given up [23]. Service quality is specified as a complex, abstract, and multidimensional nature constructs [34][32]. [35] adopted DINESERV and modified service quality items to measure overall service quality, includes serving food exactly, providing prompt and quick service, and employees answering customer questions. Behavioral intention describes an affirmed likelihood of engaging in a certain behavior [36], and is closely correlated with the behavior itself [37]. As customers have positive attitudes towards products they recommend to others, repeat purchases, spending more and paying premium prices [38], this can provide practical guidance helping restaurant practitioners to understand customer behavioral intention. Perceived value is a better predictor of intentions than either satisfaction or quality [39][40]. Quality is fundamental to perceived value [39]. Restaurant service quality directly and significantly influences perceived value [39][40][23], and customer behavioral intentions [41]. 3. Methodolology 3.1. Research approaches FCM forms an adjacency matrix by a set of identified causality coefficients and yields a simulation [42]. The simulation enables decision makers have a clear picture of between factors to identify the most relevant factors and to enhance outcomes. But the causalities of FCM are derived from based on the experts. To accurately pre-specify the causalities between the concepts for various practical problems is hard for experts, which has less accuracy and reliably that will lead to the results not precisely described [43][5]. To objectively require quantify the causality coefficients and indicate the significance of causal links performs a FCM. This study adopts Structural Equation Modeling (SEM) to

49

50


understand the causality between variables by the questionnaire, and further to examine reliably and validity. A two-step structural equation modelling was used to test the research model. Maximum likelihood was used for all parameter estimation with Amos 16. The first confirmatory factor analysis (CFA) is conducted to evaluate the measurement model for modelled constructs. CFA enables performance of tests regarding the reliability, convergent validity and discriminate validity of the measurement model. To assess reliability and internal validity of the measurement model is examined by calculating the composite reliability (CR) and average variance extracted (AVE). The AVE of each measure, we hope that is more than 50% of the variance as suggested by [44] to indicate that the variance captured by the construct is greater than the variance due to measurement error [45]. Convergent validity is a measure of the degree which two observed variables to measure the same construct correlated and is expected when each measurement’s estimated pattern coefficient on its underlying construct factor is significant. Items have a factor loading over 0.45 [46]. Discriminate validity was assessed according to [45] suggested approach. By examining AVE for each of the latent constructs and comparing this to the squared correlations among the constructs, the shared variance among any two constructs (i.e., the square of their inter-correlation) was always less than the average variance explained by the construct, which suggests that discriminate validity has been achieved. The second are used to examine the hypothesized relationships among servicescape, waiting experience, food quality, perceived value, perceived overall service quality, and customer behavioural intention. We expect that the structural model exhibit a good fit with the data, with fit indices of Root Mean Square Error of Approximation (RMSEA), Goodness-of-Fit Index (GFI), Adjusted Goodness-ofFit Index (AGFI), and Comparative Fit Index (CFI) fulfilling the respective benchmarks [44][45]. In addition to, we also gained to quantify the causality coefficients for more objective method and build an adjacency matrix to perform a FCM simulation. 3.2. Development of Measures The development of the instruments was adapted from previous literature review [13] [35] [25][48][26][38]. Table 1 lists the constructs, definitions and sources of scales. The questionnaire was first developed in English, but as the survey was conducted in Chinese. The study invited industrial practice experts and academicians to aid in the process of translation. The wording and interpretation of items and the extent which respondents would feel them possess the necessary knowledge to provide appropriate responses scrutinized until a final draft of the questionnaire. Table 1 Construct Measurement

Construct

Construct Definition

Servicescape

The physical settings and environment of full-service restaurant comprises interior design, ambient factor, spatial layout and [13] [35] human elements.

Waiting experience

The degree to which customers are satisfied with the waiting time associated with a desired service.

[26]

Food quality

The restaurant provides in terms of food tasty, food presentation, menu variety and healthy food options.

[25]

Perceived overall service quality

The restaurant provides food exactly, prompt and quick service, [35] and well response employee.

Perceived value

Overall consumer assessment of consumption experiences is based on perceptions of what is received versus what is provided.

[48]

Customer behavioral intention

The intention of a customer returns to a full-service restaurant, recommend it to others, and provide positive word of mouth.

[38]

Construct Sources

The draft questionnaire was developed, used respondent anonymity, meaning anonymity of the measurement items and pilot-tested by 50 full-service restaurant diners. The result of pilot-test is that all variables’ reliability is greater than [45] suggested standard value 0.7 (see Table 2). Items that do not significantly contribute to the reliability and have lower reliability are eliminated. Finally, the servicescape has 14 items to reflect four dimensions such as interior design, ambient, spatial layout, and human elements. Waiting experiences are measured with five items. Food quality has four items. Perceived overall service quality has three items and perceived value is four items, customer behavioral


intention is measured in terms of repeat patronage, recommendation and saying positive word-ofmouth. All items were assessed via a 7-point Likert-scale, ranging from extremely disagree (1) to extremely agree (7). Table 2 Construct Cronbach’ α Construct

Cronbach’ α

Servicescape

0.748

Waiting experience

0.823

Food quality

0.718

Perceived overall service quality

0.712

Perceived value

0.771

Customer behavioral intention

0.809

4. Anticipated Achievement In this paper, we first use questionnaire survey. For eliminating the Common Method Variance, we also use respondent anonymity, meaning anonymity of the measurement items and Harman’s singlefactor to test it. Beside we adopt SEM to understand the causality between variables or among multiple variables. This way first uses CFA to test the construct validity, convergent validity, and discriminate validity of the questionnaire. To show has not any problem with the measures of the questionnaire, and the items are able to measure adequately the latent variables and to gain the causality coefficients for more objective method. We adopt the fuzzy cognitive map simulation make decision makers to perform a lot of situations which might occur in real relation among servicescape, waiting experience, food quality, perceived value, perceived overall service quality, and customer behavioral intention by performing in a way of (1) creating several stimuli vectors which denote input conditions, (2) multiplying with adjacency matrix from questionnaires. It shows objective method to gain quantify the causality coefficients and build an adjacency matrix to perform a FCM simulation, and help the decision maker has a clear picture of affecting factors and their relation in the complex interactions among factors, and can also find or choice which combinations of changes in design factors that would lead to identify the decision factors most relevant in increasing repurchase rate for a full-service restaurant. References [1] Berry LL, Carbone LP, Haeckel SH. Managing the total customer experience. Sloan Management Review 2002; 43 (3): 85-9. [2] Stylios CD, Groumpos P.P. Modeling complex systems using fuzzy cognitive maps. IEEE Trans Syst Man Cybern Syst Hum 2004; 34(1): 155-162. [3] Axelrod R. Structure of decision: The Cognitivem Maps of political elites. Princeton University Press, New Jersey; 1976. [4] Kosko B. Neural networks and fuzzy systems: A dynamical systems approach to machine intelligence. New York: PrenticeHall; 1992. [5] Lee S. Ahnm H. Fuzzy cognitive map based on structural equation modeling for the design of controls in business-toconsumer e-commerce web-based systems. Expert Syst Appl 2009; 36(7): 10447-60. [6] Kosko B. The new science of fuzzy Logic: Fuzzy thinking. London: HarperCollins; 1994. [7] Kosko B. Fuzzy engineering. New Jersey: Prentic-Hall; 1997. [8] Stylios CD. Georgopoulos VC. Malandraki GA. Chouliara S. Fuzzy cognitive map architectures for medical decision support systems. Applied Soft Computing 2008; 8(3): 1243-1251. [9] Rodriguez-Repiso L. Setchi, R. Salmeron JL. Modeling IT projects success with Fuzzy Cognitive Maps. Expert Syst Appl 2007; 32(2): 543-559. [10] Noh JB. Lee KC. Kim JK. Lee JK. Kim SH. A case-based reasoning approach to cognitive map-driven tacit knowledge management. Expert Syst Appl 2000; 19(4): 249-59. [11] Stylios CD. Groumpos PP. Fuzzy cognitive maps in modeling supervisory control systems. J Intell Fuzzy Syst 2000; 8(2): 83-98. [12] Kang I. Lee S. Choi J. Using fuzzy cognitive map for the relationship management in airline service. Expert Syst Appl 2004; 26(4): 545-555. [13] Bitner MJ. Servicescapes: the impact of physical surrounding on customer and employees. J Market 1992; 56(2): 57-71. [14] Tombs A. McColl-Kennedy J. Social-services cape conceptual model. Market Theor 2003; 3 (4): 447-475. [15] Jang SS. Namkung Y. Perceived quality, emotions, and behavioral intentions: Application of an extended Mehrabian– Russell model to restaurants. J Bus Res 2009; 62(4): 451-460. [16] Kim WG. Moon YJ. Customer cognitive, emotional, and actionable response to the servicescape: A test of the moderating effect of the restaurant type. Int J Hospit Manag 2009; 28(1): 144-156. [17] Berry LL. Seiders K. Grewal D. Understanding service convenience. J Market 2002; 66(3): 1-17.

51

52


[18] Cameron MA. Baker J. Peterson M. Braunsberger K. The effects of music, wait-length evaluation, and mood on a low-cost wait experience. J Bus Res 2003; 56(6): 421-430. [19] Fitzsimmons JA. Fitzsimmons MJ. Service management: Operations, strategy, and information technology. New York: McGraw Hill; 2004. [20] Nie W. Waiting: Integrating social and psychological perspectives in operations management. Omega 2000; 28(6): 611-629. [21] Jen W. Tu R. Lu T. Managing passenger behavioral intention: An integrated framework for service quality, satisfaction, perceived value, and switching barriers. Transportation 2011; 38(2): 321-342. [22] Baker J. Parasuraman A. Grewal D. Voss GB. The influence of multiple store environment cues on perceived merchandise value and patronage intentions. J Market 2002; 66(2): 120-141. [23] Zeithaml VA. Consumer perceptions of price, quality and value: A means-end model and synthesis of evidence. J Market 1988; 52(3): 2-22. [24] Kim WG. Ng CYN. Kim YS. Influence of institutional DINESERV on customer satisfaction, return intention, and word-ofmouth. Int J Hospit Manag 2009; 28 (1): 10-17. [25] Namkung Y. Jang S. Does food quality really matter in restaurant: its impact of customer satisfaction and behavioral intention? J Hospit Tourism Res 2007; 31(3): 387-410. [26] Sulek JM. Hensley RL. The relative importance of food, atmosphere, and fairness of wait: the case of a full-service restaurant. Cornell Hotel and Restaurant Administration Quarterly 2004; 45(3): 235-247. [27]Susskind AM. Chan EK. How restaurant features affect check averages. Cornell Hotel and Restaurant Administration Quarterly 2000; 41(6): 56-63. [28] Peri C. The universe of food quality. Food Quality and Preference 2006; 17(1-2): 3-8. [29] Mattila AS. Consumer behavior research in hospitality and tourism journals. Int J Hospit Manag 2004; 23(5): 449-57. [30] Jensen HR. The interrelationship between customer and consumer value. Asia Pacific Advances in Consumer Research 1996; 2: 60-63. [31] Woodruff RB. Gardial SF. Know your customer: New approaches to understanding customer value and satisfaction. Cambridge: Blackwell Publishers; 1996. [32] Petrick JF. Development of a multi-dimensional scale for measuring the perceived value of a service. J Leisure Res 2002; 34 (2): 119-134. [33] Sweeney JC. Soutar G. Consumer perceived value: the development of multiple item scale. J Retailing 2001; 77 (2): 203220. [34] Parasuraman A. Zeithaml VA. Berry LL. SERVQUAL: a multiple-item scale for measuring consumer perceptions of service quality. J Retailing 1988; 64(1): 12-37. [35] Ha J. Jang SCS. Perceived values, satisfaction, and behavioral intention: the role of familiarity in Korean restaurants. Int J Hospit Manag 2010; 29(1): 2-13. [36] Oliver RL. Customer Satisfaction: A Behavioral Perspective on the Consumer. New York: McGraw-Hill; 1997. [37] Jang SC. Feng R. Temporal destination revisit intention: the effects of novelty seeking and satisfaction. Int J Hospit Manag 2007; 28 (2): 580-590. [38] Zeithaml VA. Berry LL. Parasuraman A. The behavioral consequences of service quality. J Market 1996; 60(2): 31-46. [39] Cronin JJJr. Bready MK. Hult GT. Assessing the effects of quality, value, and customer satisfaction on consumer behavioral intention in service environment. J Retailing 2000; 76(2): 193-218. [40] Petrick JF. Backman SJ. An examination of the construct of perceived value for the prediction of golf travelers’ intentions to revisit. J Trav Res 2002; 41(1): 38-45. [41] Liu YH. Jang SCS. The effects of dining atmospherics: an extended Mehrabian-Russell model. Int J Hospit Manag 2009; 28(4): 494-503. [42] Huff A. Mapping Strategic Thought. In, Huff, A.S.,ed. Mapping strategic thought, pp.11-49. London: John Wiley; 1990. [43] Kang I. Lee S. Choi J. Using fuzzy cognitive map for the relationship management in airline service. Expert Syst Appl 2004; 26(4): 545-555. [44] Bagozzi R P. Yi Y. On the evaluation for structural equation models. J Acad Market Sci 1988; 16(1): 74-94. [45] Fornell C. Larcker DF. Evaluating structural equation models with unobservable variables and measurement error. J Market Res 1981; 18(1): 39-50. [46] Jöreskog K G.. Sörbom D. LISREL 8: User’s reference guide. Chicago: Scientific Software; 1996. [47] Hair JF. Anderson RE. Tatham RL. Black WC. Multivariate Data Analysis with Readings. New York: Macmillan Publishing; 1998. [48] Ryu K. Han H. Kim TH. The relationships among overall quick-casual restaurant image, perceived value, customer satisfaction, and behavioral intention. Int J Hospit Manag 2008; 27(3): 459-469.