The CAM is an advanced multivariate market research technique that deals with the understanding of consumer preferences for products and services. Companies design new products or services that will meet customers’ needs by means of the CAM. This method appeared in mid-1970 and the first marketing journal paper on Conjoint Analysis was published by Green and Rao (1971). CAM has been widely used since 1980. According to a survey made in 1989, of all the CAM based researches, 59% are related to consumption goods, 18% are related to industrial goods, 9% are related to financial services and the remaining 14% are related to other areas (Wittink and Cattin, 1989).
One benefit of the CAM is its ability to produce market models that will enable companies to improve their market share and the other benefit is to examine the fluctuations in consumer preferences. It is possible to measure the effects of different levels of controllable attributes by the use of the CAM.
In this paper, section two introduces the CAM. Section three develops an application, which deals with the determination of the students’ MBA preferences. In section four, a simulation and simulator study is introduced. Conclusions are given in section five.
1. Conjoint Analysis Method
Understanding customers’ needs and wants is of great importance in business planning, decision-making and strategy development. The CAM is a multivariate technique used to understand customers’ needs and wants and their decisions about how they really value products or services and to develop preferences for these products or services. Breaking down a product or service into a set of attributes and levels is the core of the CAM. For example, colour is the attribute of a car; red, blue and white are the levels related to the colour attribute of the car. When evaluating the value of a product or a service, the different values provided by each attribute are combined. Utility, which is the conceptual basis for measuring value in CAM, is a subjective judgment of preference unique to each individual. It encompasses all product or service features and as such is a measure of overall preference. In CAM, utility is assumed to be based on the value placed on each of the levels of the attributes. One of the utility estimation methods is part-worth utility estimation method. The utilities of attributes are represented with the linear combination of the preferences of each level. Part-worth gives the importance level of each attribute. It is estimated as in (1).
where is the desirability of the jth level of the pth attribute and is the function related to the part-worth.
In CBC, part-worth of the each attribute is estimated by using Multinomial Logit Analysis (MNL). MNL is used since the dependent and the independent variables are categorical. Estimation of the part-worths is provided by the maximum likelihood method.
Utility is expressed in a relationship reflecting the manner in which the utility is formulated for any combination of attributes with the use of an additive model (Hair et al., 1998). Additive model is the basic model used in the CAM. In additive model, a consumer adds up the utilities for each attribute in order to get a total utility for a combination of attributes. The larger the utility, the more preferred is the level. The sum of the utilities is equal to “0” within each attribute. Products or services with the highest utility can be regarded as the most preferred product or service.
In the CAM, the first step is to break down a product or a service into important characteristics as attribute and levels. After evaluating the consumer preferences for a product or a service, a utility associated with each level of each attribute is computed. At the evaluation stage, the separate amounts of value provided by each attribute are put together and the most preferred product or service is determined accordingly.
In the CAM, a set of real or hypothetical products or services in combination of selected levels of each attribute is constructed. These combinations, known as treatments or stimuli, are presented to consumers to receive their overall evaluations about these products or services. The influence of each attribute and the value of each level of each attribute are evaluated by means of the consumer overall evaluations. At this stage, each attribute has to be independent, that is to say, attributes can appear together. Attributes such as “price” and “brand” have to be carefully examined (http://www.dobney.com./Conjoint/conjoint design.htm). Each level of each attribute has to be easily read and understood by customers. Another feature is that attributes have to cover all those parts that are important for the customer; in other words, levels have to cover all possibilities of an attribute. These features are vital when performing the CAM. Evaluation of the contribution of each attribute is of great importance when defining a product or a service with optimum combination of attribute levels. These evaluations can be used in estimating consumer preferences and in identifying marketing opportunities for future products and services.
When performing the CAM, different conjoint methodologies can be used according to different features. One of them is the Adaptive Conjoint Analysis (ACA). It allows for a large number of attributes (up to 30) and levels (up to 7 for each attribute). It is also possible to perform computer-based interview by using the ACA. In the mid-’80s, Johnson (1987) introduced the ACA program. The other most common Conjoint Analysis type is the Choice-based Conjoint (CBC) Analysis. The first paper related to CBC (using the MNL) was written by McFadden (1974). Punj and Staelin (1978) adopted this technique after McFadden (1974). Gensch and Recker (1979) also used this model. Batsell and Lodish (1981) used MNL to model individual choices. They indicated that the model can be applied to market segments. Theil’s logit model to a CBC problem is applied by Mahajan et al. (1982). Louviere and Woodworth (1983) dealt with more complex experimental designs needed for CBC Analysis. The CBC Analysis and related calculations are completely different from ACA.
In this study, CBC is used as the CAM. The main characteristic of CBC is that the consumer expresses his preferences by choosing products or services from a set of products or services, rather than by ranking or rating them. The CBC has several advantages. Choosing a preferred product or a service from a group of products is a simple task that consumers can do in markets. It is possible to have a “none” option through which consumers can express their lack of interest. CBC can use all the attributes available when expressing a service or a product that a consumer can choose from a set of products or services. By the random designs used by the CBC system that allow study of all interactions, the CAM can be easily analyzed at an aggregate level. In the CBC, the values that each consumer assigns to attribute levels can not be examined; instead data are aggregated for the analysis from the consumers by examining the “utility values” produced from the group of consumer preferences. The utility values can be predicted for a product or a service that cannot actually appear in choice tasks.
2. An Application of Conjoint Analysis Method
All over the world and in Turkey, there is a great competition among MBA programs. In this study, MBA preferences of Başkent University students are aimed to be determined with the help of the CAM. Universities in Turkey which have MBA programs are regarded as markets and the students of Başkent University are taken as customers. The CBC Sawtooth Software Program is used as the analysis tool.
The first step of the analysis is to define the problem by means of attributes and their levels given in Appendix 1. In order to select the methodology and to design the stimuli, attributes and their levels are determined with the help of specialists, related articles and web sites of the universities. The critical factor in specifying attributes and their levels is that they should have all positive and negative characteristics of the product or service; they should be perceiveble, discriminative and practible. Too many attributes can greatly affect the reliability of the model thus only the most important characteristics are included to the analysis. Secondly, the stimuli, which are combinations of specific set of levels of attributes, are designed. Each combination of this specific plan is called stimulus. Choice cards, which are evaluated by respondents, are arranged to form each stimulus. For this purpose, 6Í3Í4Í2Í4Í2Í2Í2Í3Í7=96768 different combinations are obtained by multiplying each level of each attribute. But it is almost impossible for a consumer to evaluate these combinations. For this reason, orthogonal arrays and fractional factorial designs, the most common method for defining the subset of stimuli, are used (Addelman, 1962, Montgomery, 1991). The orthogonality and balance among attributes are achieved by randomly constructed designs of CBC Sawtooth Software Program. Such designs have the advantage of examining all interaction effects.
In this study, 12 different choice cards are obtained from the CBC Sawthooth Software Program. One example of these choice cards is given in Appendix 2. Each choice card contains 7 different options that contain several attributes at different levels. For reliability of the CAM, the consumers need to have information about a service or a product. If the consumers have enough information about attributes they can choose the best alternatives from the sets of different alternatives.
The stratified simple random sampling plan is used for selecting the samples. This technique is regarded as the appropriate sampling technique. For this reason, junior and senior students from the Faculties of Economics and Administrative Sciences, Science and Letters and Engineering are selected as the target population in order to evaluate the MBA preferences of Başkent University students. When determining the sample size, what is taken into consideration is the proportion of the students who wish to study MBA, equation (2) is used, and the appropriate sample size is estimated to be 251 with a 95% confidence, by assuming p=0.5 in order to have the maximum variance at the most heteregenous state as there is no information about the number of students who wish to study MBA, and with d=0.05, the sensitivity level.
where is the standard normal distribution table value, is the estimate for proportion, is the sensitivity level, N is the population size and n is the sample size.
Since there is no information related to the cost of project, homogenity proportional sharing method is used to determine the number of students to be interviewed in each stratum. 251 students are distributed to the faculties then to the departments by multinomial staging. While collecting the data, it was not possible to use target population, so sampling frame is used, and the students for this study are chosen by means of random numbers from the sampling frame. The choice tasks are given to the students through face-to-face interview. To generate data for the determination of the MBA preferences based on tuition and fees, a survey is conducted with 251 students. About 39% of the 251 students want to have an MBA degree. The numbers and percentages of students, willing to pursue an MBA degree are given in Appendix 3. Since the Conjoint Analysis results are more accurate when the consumers’ have information about the product or service then for that reason, the Conjoint Analysis results are evaluated from these 98 students. These students were asked to evaluate 12 different choice cards in which each choice card has 7 different options. The students then chose one of the options.
4. Conducting Simulation and Simulator
The CAM can develop market simulation models that can predict the consumers’ attitude to product changes. The purpose of conducting simulation is to estimate the proportion of consumers’ choice for specific attribute profiles entered into the simulator (Orme, 2003). A market simulator may help us to investigate new product designs, to develop pricing strategies and to estimate a choice share from the current market. By using the simulator, utilities that reflect consumers’ preferences for each level of each attribute can be converted into a form that helps us to predict share of preference for different products and services.
In this section, the part-worths found from the MNL, computed for each attribute level, are used in market simulation to compute share of preferences among MBA programs. To estimate market share of each scenario, total effects are estimated as in (3).
After estimating , the market shares are estimated as in (4).
where s is the total number of scenarios.
In this study, the market simulator is developed to show current market share of the universities and to investigate alternative MBA program scenarios instead of the current MBA programs in the universities (Appendix 4). With the help of the simulator, market shares of the universities can be determined when these universities have changes in their MBA programs. The vlookup (vertical look up) function of the Microsoft Excel program is used to develop the simulator. Simulator consists of forms that include each level of each attribute. Changing the attribute levels in the form can change MBA program scenarios, and the simulator can give the market share of the universities according to the new MBA program scenarios. According to the Appendix 4, Boğaziçi University has the greatest market share with 28.5%, Bilkent and Sabancı Universities follow Boğaziçi University with market shares of 26.7% and 15.3%, respectively. Bilgi University is the least preferred university with a market share of 7.4%. Başkent University has a market share of 8.2%. If Başkent University decides to drop its tuition and fees from $12,500 to $10,000, its market share goes up to 10.1% as seen in Appendix 5. It is concluded that a decrease in the fee has a positive effect, as expected, on the market share of the universities.
Many other scenarios can be tried in the simulator used in Appendix 4. Universities can try new MBA programs to increase their market share by using this simulator, and can determine which scenario is the best one to attract students.
The survey results show that “University Name” plays the most important role in MBA preferences and level of importance for the attribute of “University Name” is 41.3%. The second important attribute is the “Tuition and Fees” according to 18.95% of the students. “Specialization Possibility” follows “Tuition and Fees” with 9.12%. The least important attribute is the “Academic Staff” according to Başkent University students. In Turkey, the most preferred university for the MBA program is Boğaziçi University and the second most preferred one is Bilkent University. Contrary to the expectations, “2-year full-time MBA program without thesis” has greater appeal for students than “1-year distance learning e-MBA program without thesis”. It can be concluded that establishing a new e-MBA program will not increase the demand for the universities. Thus, it can be said that students believe in the importance of attending the courses rather than studying at home or at work.
According to the results, students prefer programs which offer “International Connection” and which do not require a “Work Experience Condition”. Offering a “Specialization Possibility”, using English as the “Medium of Instruction” and possessing an “Academic Staff” which offers a counselling service have greater influence on students than those universities that do not offer these opportunities.
Executive MBA program, in which a student simultaneously works and studies, is the most preferred type of MBA program. It can be concluded that, students prefer to earn money and to improve themselves simultaneously when studying at the university. And the least preferred type is the Academic MBA program. That is because to have a successful academic career is not an easy way to proceed. A student who wants to be successful in his academic career has to study a lot.
Another important finding is that “higher tuition and fees” makes the MBA less attractive. Thus, universities have to hold the fees at the lowest possible values.
Addelman, S. (1962) Orthogonal main-effect plans for asymmetrical factorial experiments, Technometrics, 4, 21-46.
Batsell, R. R. and Lodish, L. M. (1981) A model and measurement methodology for predicting individual consumer choice, Journal of Marketing Research, 18, 1-12.
CBC User Manual, Version 2.0, Sawtooth Software, Inc.
Gensch, D. H. and Recker, W. W. (1979) The multinomial multiattribute logit choice model, Journal of Marketing Research, 16, 124-132.
Green, P. E. and Rao, V. R. (1971) Conjoint measurement for quantifying judgmental data, Journal of Marketing Research, 8, 355-363.
Hair, J. F., Anderson, R. E., Tatham, R. L. and Black, W. C. (1998) Multivariate Data Analysis, 5th edn., New Jersey: Prentice Hall.
Johnson, R. M. (1987) Adaptive conjoint analysis, in Sawtooth Software Conference on Perceptual Mapping, Conjoint Analysis, and Computer Interviewing. Sawtooth Software, Ketchum, ID, pp. 253-265.
Louviere, J. and Woodworth, G. (1983) Design and analysis of simulated consumer choice or allocation experiments, Journal of Marketing Research, 20, 350-367.
Mahajan, V., Green, P. E. and Goldberg, S. M. (1982) A conjoint model for measuring self and cross-price demand relationships, Journal of Marketing Research, 19, 334-342.
McFadden, D. (1974) Conditional logit analysis of qualitative choice behavior, Frontiers on Econometrics, edn. P. Zarembka, New York: Academic Press, pp. 105-421.
Montgomery, D.C. (1991) Design and Analysis of Experiments, Chichester: John Wiley.
Orme, B. (2003) Introduction to Market Simulators for Conjoint Analysis, Sawtooth Software, Inc.
Punj, G. N. and Staelin, R. (1978) The choice process for graduate business schools, Journal of Marketing Research, 15, 588-598.
Wittink, D. And Cattin, P. (1989) Commercial use of conjoint analysis: An update, Journal of Marketing, 53, 91-96.
Appendix 1 Attributes and Levels
Appendix 2 An Example of Choice Card
Appendix 3 The Number of Students Interviewed and Percentages of Students Willing to Pursue MBA Degree
Appendix 4 Market Simulator
Appendix 5 Market Simulator (for Başkent University)
The authors would like to thank Prof. Ismail ERDEM and Sahir KÖKSOY for their careful reading of the manuscript and their constructive comments. In addition, authors are grateful to Ceren TOPGÜL, Berk ÖZDEMIR, the referees and associate editor for contributions.
AUTHORS / SOBRE LOS AUTORES
Assistant Professor at University of Baskent, Faculty of
Science and Letters, Department of Statistics and Computer Sciences,
Baglica Campus, 06530, Ankara, TURKEY. Main areas of study:
Multivariate Analysis, Mathematical Statistics, Nonparametric
Statistics, Poisson RegressionAssistant Professor at University of Başkent (Ankara-Turkey).
Search for this author's articles in Scholar Google
Assistant at University of Baskent, Faculty of Science and Letters,
Department of Statistics and Computer Sciences, Baglica Campus,
06530, Ankara, TURKEY. MSc, Economy (Middle East Technical
Search for this author's articles in Scholar Google
Research Assistant at University of Baskent, Faculty of
Science and Letters, Department of Statistics and Computer Sciences,
Baglica Campus, 06530, Ankara, TURKEY. MSc, Financial
Mathematics (Middle East Technical University, Ankara- TURKEY).
Search for this author's articles in Scholar Google
ABOUT THE AUTHORS / SOBRE LOS AUTORES
İlknur Özmen (firstname.lastname@example.org). Assistant Professor at University of Baskent, Faculty of Science and Letters, Department of Statistics and Computer Sciences, Baglica Campus, 06530, Ankara, TURKEY. Main areas of study: Multivariate Analysis, Mathematical Statistics, Nonparametric Statistics, Poisson RegressionAssistant Professor at University of Başkent (Ankara-Turkey). Search for this author's articles in Scholar Google
Bilge Yaşıt. Research Assistant at University of Baskent, Faculty of Science and Letters, Department of Statistics and Computer Sciences, Baglica Campus, 06530, Ankara, TURKEY. MSc, Economy (Middle East Technical University, Ankara-TURKEY). Search for this author's articles in Scholar Google
Özge Sezgin. Research Assistant at University of Baskent, Faculty of Science and Letters, Department of Statistics and Computer Sciences, Baglica Campus, 06530, Ankara, TURKEY. MSc, Financial Mathematics (Middle East Technical University, Ankara- TURKEY). Search for this author's articles in Scholar Google
ARTICLE RECORD / FICHA DEL ARTÍCULO
Özmen, İlknur, Yaşıt, Bilge & Sezgin, Özge (2006). A Conjoint Analysis to Determine the Preferences for Some Selected MBA Programs. Revista ELectrónica de Investigación y EValuación Educativa, v. 12, n. 1. http://www.uv.es/RELIEVE/v12n1/RELIEVEv12n1_7.htm. Consultado en (poner fecha).
Title / Título
A Conjoint Analysis to Determine the Preferences for Some Selected MBA Programs. [Análisis Conjunto para determinar las preferencias para algunos programas seleccionados de MBA]]
Authors / Autores
İlknur Özmen, Bilge Yaşıt, & Özge Sezgin
Revista ELectrónica de Investigación y EValuación Educativa (RELIEVE), v. 12, n. 1
Publication date /
Fecha de publicación
2006 (Reception Date: 2006 Feb. 22; Approval Date: 2006 April 29; Publication Date: 2006 May 02)
This paper reviews the Conjoint Analysis Method (CAM), which is a multivariate marketing research technique used to determine consumer behaviours and preferences for products or services. One aim of this study is to demonstrate that the CAM can be used in “Service Sector” as well as in “Product Sector” and the other aim is to utilize CBC Sawtooth Software Program, which is a special program for CAM. A usage of CBC Sawtooth Software Program is demonstrated in the analysis of Management Business Administration (MBA) program preferences of Başkent University students. This study includes those MBA programs that require substantial tuition and fee payments. According to the results of the study, “University Name” plays the most important role in MBA preferences. The Conjoint Analysis found that, most preferred university is the Boğaziçi University and the most preferred type of MBA program is the “Executive MBA Program”. Another important finding is that “Higher Tuition and Fees” makes the MBA less attractive..
Este trabajo examina el método de análisis conjunto (CAM), que es una técnica multivariante de investigación de mercados utilizada para determinar los comportamientos y preferencias del consumidor por productos o servicios. Un objetivo de este estudio es demostrar que el CAM puede ser aplicado en el "sector servicios" al igual que en el "sector de producción". El otro objetivo principal es utilizar programa informático CBC de Sawtooth, que es un programa especial para CAM. El CBC de Sawtooth se aplica a las preferencias de estudiantes por los programas de Administración y Dirección de Empresas de (MBA) de la universidad de Baskent. Este estudio incluye sólo los programas de MBA que requieren una cuota de matrícula cuantiosa y pago de honorarios. Los resultados del estudio muestran que la variable "nombre de la universidad" es la más importante en las preferencias que los alumnos tienen por los MBA. El Análisis Conjunto encontró que los preferidos son como universidad de Bogaziçi y como tipo de programa el "MBA Ejecutivo" (que simultanea estudio y trabajo). Otra conclusión importante es que los mayores costes hacen el MBA menos atractivo..
Conjoint Analysis; Stimuli; Utility; Multinomial Logit Analysis; Consumer Preferences; Executive MBA; Sampling Plan; Simulator
Análisis Conjunto; estímulos; servicio público; Análisis Logit Multinomial; preferencias del consumidor; MBA ejecutivo; muestreo; simulación
Başkent University, Ankara (Turkey)
Publication site /
Language / Idioma
English (Title, abstract and keywords in Spanish )
Volumen 12, n. 1
© Copyright, RELIEVE. Reproduction and distribution of this article is authorized if the content is no modified and its origin is indicated (RELIEVE Journal, volume, number and electronic address of the document).
© Copyright, RELIEVE. Se autoriza la reproducción y distribución de este artículo siempre que no se modifique el contenido y se indique su origen (RELIEVE, volumen, número y dirección electrónica del documento).
Revista ELectrónica de Investigación y EValuación Educativa