Agent-based modeling of consumer decision making process based on power distance and personality
Introduction
Agent-based modeling is a new analytical and computational method envisaged as important in many fields of study that have multi-level system properties, since it gives a better understanding of micro processes and their emergent consequences at macro level [18], [49]. This applied method must create a simplified representation of what occurs in reality so that each agent plays the role of an individual as if it is happening in social reality. Agent-based modeling is a new way of analyzing of dynamic complex systems [18]. Hence, it is used in consumer behavior modeling to give a better understanding of, as well as predicting, the consumer decision making processes [18], [97]. The consumers are represented as autonomous agents with individual characteristics as well as an independent internal decision making process [18], and sellers as agents who enter their products with different characteristics into the market.
Consumer decision making involves research in the fields of sociology, psychology, consumer behavior and marketing, computer science and artificial intelligence which is related to the research of the computational modeling of complex social systems [97]. Modeling this behavior in an agent-based simulation requires a pragmatic integration of findings from all these fields. What is of prime importance for the scientists today is the presentation of a computational agent-based model for the consumers’ decision making processes which is the closest to the real consumer’s decision. Hence, what can be of help to researchers is taking human factors influencing consumer behaviors into account.
Consumer behavior includes five important stages: need recognition, information search, evaluation of alternatives, purchase, and post purchase behaviors [7], [9], [18], [47], [62], [75], [83]. The consumer decision making process starts with the recognition of a need. The main factor here is ‘need’. According the nature of the present study, we use the definition given by Max-Neef [55] who defines a need as an ‘underlying internal forces that drive our actions’. Other important concepts in this stage include actual state, desired state and tolerance threshold of each need for the consumer. Actual state refers to the extent of the consumer’s present satisfaction with the product he (or she, but for simplicity’s sake we shall use ‘he’) possesses. The desired state is the consumer’s desired satisfaction. A need is felt by a consumer only when the discrepancy between the desired and the actual state is larger than the tolerance threshold of the given need and accordingly forces the consumer to take an action to satisfy the aroused need. Then, the other stages of consumer decision making processes come into play to search, evaluate and find the best option.
Research confirms that culture and personality influence the consumer behavior at different stages of decision making process [6], [8], [11], [17], [20], [50], [54], [61], [63], [73], [83], [86], [96]. Considered as groups, people of different countries have their own average purchasing behavior. For example, the product parameters which are important for an average American are different from those that matter to an average Japanese. Culture, then, in its macro level will explain the difference in purchasing behavior common among a group of people. What makes this interesting is that at the same time, different purchasing behavior can be found among people born and bred in the same cultural background or even belonging to the same family. Such disparities are related to personal differences and can be explained in terms of personality and situational characteristics.
What has motivated the current study is the influence of the stratification of society on the purchasing behavior of its members. This is because that attribute seemed to have an empirical relationship with car purchasing behavior. We can hypothesize that most people’s purchasing behavior is affected by their social status, personal traits, and the cultural values of the society in which they live. This is a complex set of dependencies and we wished to start where there was a clear trend. There was also theoretical justification. People in one society might aim at buying the products that keep them in the same level of social status so that others are able to construe their social position at a glance. In contrast, displaying status symbols is less important, or even frowned upon, However, environmental and social responsibility is of high importance in other societies [83].
Culture is able to explain these differences at group level. Scientists hold that among the numerous effective factors in purchasing behavior, the two factors of culture and personality influence people’s purchasing behavior most besides their wealth [6], [50], [61], [63], [73], [83], [86].
Culture entails the unwritten rules of a society shared among its members. Among different perspectives towards culture and taking into account the objectives of this study which takes culture modeling as an effective factor in purchasing behavior, only those perspectives on culture can be applied to the study which present culture by different dimensions as well as in numbers [33]. One of the best known perspectives on culture is [23] which takes culture as having been formed of five dimensions which are also measurable in numbers. These dimensions include Power Distance, Individualism vs. Collectivism, Masculinity, Uncertainty Avoidance, and Long Term Orientation. Power Distance is defined by Hofstede et al. [25] as ‘the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally’ (p. 61). In this definition, institutions include informal institutions such as families and communities. The definition posits Power Distance as a society-level phenomenon, not as a personal attribute. There are consequences for behavior though. Societies of large power distance are stratified, and people take their own class as a reference for their behaviors. There is likely to be a class of very rich people, a large working class, and not much of a middle class. Societies of small power distance are egalitarian, and people take everybody as their point of reference. There is likely to be a large middle class with few outliers. Most countries are in between these extremes.
Hofstede [23, p. 111] asserts that societies with large power distance support status consistency, which means that visible status symbols consistent with one’s class membership are expected. Jager [42] emphasizes the importance of hierarchical societies in purchasing behavior. De Mooij [6, p. 145] found that car consumers in societies of large power distance are more likely to value ‘design’ and ‘international image’ in a purchased car. Large power distance as defined by Hofstede [23] is in fact the description of hierarchical societies, where everybody is supposed to have his or her rightful place in society. An important need in societies with large power distance is that of showing appropriate social status of oneself and one’s family so that other people can pay proper respect. This is consistent with taking one’s own class as a point of reference; one would not like to be mistaken for somebody of lower class. In societies with small power distance important needs are preserving autonomy, civic or social responsibility for the common good of the larger community, social equality and not showing off. De Mooij [6] found that environmental friendliness is a car buying motive in societies of small power distance. We call the latter need ‘social responsibility’. It is consistent with taking the entire population, rather than one’s own class, as a reference for buying behavior.
So, power distance is not about personal status seeking, but about maintaining proper stratification in society. Because of its proven real-world relation with car purchasing behavior [6], this dimension has been chosen to be analyzed in the present study. This is not to say that other dimensions could not also be studied and might not yield a richer picture. All the other dimensions in Hofstede’s model have probable, or proven, relationships with purchasing behavior. Selecting one dimension is a way of making a start on a case with good data, and checking whether the design approach taken here is any good in terms of reproducing the real-world behavior demonstrated by that data.
At the individual level, personality is another important factor influencing the individual’s behavior [44], [59]. Personality can generally be taken as a set of unique, almost permanent characteristics particular to a person [13], [74]. Like what was stated about selecting a culture model, among the presented models of personality, only those can be used which present personality in terms of characteristics with measurable numbers. Therefore, such models as OCEAN [59] which describe personality in five traits or Cattell et al. [4] which does this with 16 traits can be used. Yet, since in scientific discussions the OCEAN model is preferred for it has divided the characteristics closely related to personality better than other models and has shown validity across countries [59], [64], [65], we apply the same model. What is important is how we are able to define and model culture and personality in the process of decision making so that it is more believable. Therefore, other researchers will be able to use other models of culture and personality in their research according to their own concerns.
A factor that cannot be ignored in studying purchasing behavior is the consumer’s wealth. According to De Mooij [6], when consumer’s wealth increases, the effects of culture on purchasing behavior show up. In fact, culture and personality, as well as other personal characteristics may enhance the tendency of having a specific product; yet, it goes without saying that if one is not financially able, his internal tendencies cannot be fully traced by analyzing his purchases. Thus, the unignorable factor of budget is also included in the study.
To address the increasing demand for richer models in multi-agent systems, the MASQ meta-model was proposed by Ferber et al. [14]. The MASQ meta-model provides a proper framework for such systems. It separates the physical from the non-physical and the individual from the shared. It includes mind, body/object, space, and culture. MASQ meta model has been implemented in different areas of agent-based modeling [67], [68].
The above briefly presents some of the central concepts in this paper; they will be treated in more detail later. We need all these elements to attain the paper’s main purpose: proposing a conceptual and computational model for agent-based simulation of the consumer decision making processes. We are more interested in formalizing culture and personality in agent to make the consumer agent decision making process more human-like with regard to their internal needs. The conceptual model is based on the idea of MASQ [14] for multi-agent systems. We define the society and individuals based on this perspective. Furthermore, we propose a framework for agent mind which includes necessary elements in consumer decision making process. The car-buying motive is selected as a case to test and validate the proposed model. The model is validated in eleven European countries.
The rest of the paper is organized as follows: Section 2 describes the consumer decision making process in detail. Section 3 illustrates culture in sociology and personality in psychology. We describe Hofstede’s model of national culture [21], [22], [23], [24], [25] and McCrae and Costa’s theory of personality [52], [57], [58], [59] in detail in this section. The effects of culture and personality on consumer decision making processes are described in Section 4. Section 5 studies the literatures of agent-based simulation and agent architectures of consumer behavior with and without culture and personality. In Section 6, we propose our conceptual and computational model for consumer decision making processes. Section 7 includes the experimental results which confirm the validity of the proposed model. Finally in Section 8 we discuss about the proficiency of our model and in Section 9 we conclude the paper and propose future work.
Section snippets
Consumer decision making process
The consumer decision making process involves five stages to fulfill purchasing a product: Need Recognition, Information Search, Evaluation of Alternatives, Purchase, and Post-Purchase Behavior [7], [9], [47], [62], [83]. Fig. 1 shows the consumer decision making process.
The first stage in consumer decision making process is need recognition. It plays the central role in this process [19].
According to Max-Neef [55] the concept of need has to do with desires and wants as internal forces which
Culture in sociology and social psychology
The field of culture is a contested one in the social sciences. There is hardly an issue about which all disciplines agree. Some would hold that culture is a personal attribute much like personality [79], while others see it as shared, even as a prerequisite for the existence of a group [21]. It is therefore necessary to make a choice of model to use. When comparing individuals in teams, one might conceptualize their individual cultural backgrounds; when comparing organizations, one might take
Culture, personality and consumer behavior
In this section, the key concepts of consumer decision making process such as need, desired and actual state, need recognition, and purchase are described based on culture and personality.
Agent cognitive architectures
What is important in agent modeling is presenting an architecture which functions like the human mind. To this end, agent’s architecture and its internal decision making process should take the human cognitive factors into account to suitably represent the ‘cognitive mental model of a human’ [16]. A number of models have been presented such as ACT-R [2], SOAR [51], CLARION [85] and BDI [69]. The presented cognitive architectures are general ones and can be used in different areas where agents
Proposed model
In this section we propose an agent-based conceptual and computational model for consumer decision making process based on culture and personality. According to the narrative description about culture and personality and their effects on consumer behavior, we propose a conceptual model for consumer agents in Section 6.1. This conceptual model is based on the MASQ meta model for multi-agent systems [14]. The process of decision making of consumers is considered in this model. In Section 6.2, we
Experimental results
The model is implemented in Repast Symphony, which is free and open-source software for agent based simulation [66]. Since this model is based on the MASQ meta-model, different packages have been developed for Mind, Culture, Space, and Body. Other packages have been considered for the management of the communication among agents, I/O management, and representation of the products. R software was used to generate input data for the agents and some statistical operations.
In this section, we
Discussion
Many models have been presented for modeling purchasing behavior. Some of which have benefited a utility function for purchasing a product [3], [71], [72], [87], [88], [89], [90], [91], [92], [93], [97]. These consumer agents have been modeled based on a variety of factors such as quality and price [87], [88], [89], [90], [91], [92], [93] any of which is already of a specific value to the consumer and in fact the purchase is done based on that particular factor and value. In Roozmand et al. [72]
Conclusion and future work
In this paper we proposed a conceptual and computational agent-based model for the consumer decision making process. We were inspired by the MASQ meta model for designing and implementing the model. MASQ includes four parts: mind, culture, body and space. The main part of the model is the mind. It makes decision and contains four modules, three input vector and one state variable. In the proposed model we focused on consumer needs which drive the consumer toward a specific product. Actually,
Acknowledgements
The authors highly appreciate Dr. Marieke De Mooij for providing the data on car-buying motives from EMS 1999, which validation of the model was impossible without her help. Also, the authors thank Dr. John Tranier who helped us to develop the model based on MASQ meta model and writing agents with Repast Simphony; and Ynte Van Dam, assistant professor of Wageningen University, Marketing and Consumer Behavior Department, whose useful help has greatly improved our knowledge about influential
References (97)
- et al.
Intelligent human interface based on mental cloning-based software
Knowledge-Based Systems
(2009) - et al.
Understanding consumer heterogeneity: a business intelligence application of neural networks
Knowledge-Based Systems
(2010) - et al.
Agent-based analysis and simulation of the consumer airline market share for Frontier Airlines
Knowledge-Based Systems
(2010) - et al.
Joint factors in self-reports and ratings: neuroticism, extraversion and openness to experience
Personality and Individual Differences
(1983) - et al.
A heuristic personality-based bilateral multi-issue bargaining model in electronic commerce
Journal of Human-Computer Studies
(2009) - et al.
Agent-based simulation of consumer purchase decision-making and the decoy effect
Journal of Business Research
(2007) - A. Adler, Individual psychology, in: C. Murchinson (Ed.), Psychologies of Worcester,...
- et al.
An integrated theory of the mind
Psychological Review
(2004) - et al.
Openness, conscientiousness and stingy for buyer and seller agents in electronic marketplace
International Journal of Computer Theory and Engineering
(2010) - et al.
The Handbook for the Sixteen References 239 Personality Factor Questionnaire
(1970)