Description
This study is based to identify the factors that influence the consumer buying behavior
of the two wheeler Bikes at Allahabad, Lucknow and Varanasi cities of Uttar Pradesh. The
survey is mainly focused on the buying behavior of the consumer that motivates them to
purchase the two wheeler bikes.
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis!
Subhadip Roy*
The Indian motorbike industry is in a very healthy state today. With foreign players like Honda operating independently, the market is becoming extremely competitive. This is an exploratory study which identifies the consumer choice patterns regarding the four attributes of a motorbike: Fuel efficiency, color, type of brakes and pick-up. The objective is to find out the consumer preference ordering of the four attributes mentioned. Conjoint Analysis was used to find out the different aspects of consumer choice regarding motorbikes. This study is based on a field survey conducted on 88 respondents in and around Hyderabad.
Motivation and Objective
The Indian two-wheeler industry is set to become more competitive as most of the strategic alliances have broken and the foreign are operating independently. In such a scenario, the players need to know the importance of the attributes of a motorbike and whether there are any differences of choice pattern in different segments of the consumers. This provided the motivation for this study. The objective was to find out the consumer preference ordering of various attributes of a motorbike and applicability of conjoint analysis in finding that out and the probable use of the tool in new product development and market segmentation. This study was based on a field survey conducted in and around Hyderabad.
!
The Indian Two-wheeler Industry
The Indian two-wheeler industry has come up a long way from being a tightly government controlled industry to a highly competitive one encouraging foreign players’ participation. Today, the Indian consumer has a wide variety of brands and models to choose from when he/she thinks of buying a two-wheeler. Approximately 5.4 million units of two-wheelers are sold every year in India, making it one of the largest two-wheeler markets in the world. Not only that, India has an average of 27 two-wheelers per 1000 people, making it the highest in Asia.1 The origin of Indian two-wheeler industry can be traced back to 1948, when Bajaj
A part of this paper was presented at the 17th AIMS Annual Convention held in Hyderabad during August 28-30, 2005.
About the Author
*
Research Scholar, The ICFAI Institute for Management Teachers (IIMT), Hyderabad. E-mail: [email protected]
1
According to data published by the World Bank.
© 2006 The ICFAI University Press. All Rights Reserved.
48 The ICFAI Journal of Marketing Management, February 2006
Auto Ltd. started importing and marketing Vespa scooters in India. Shortly afterwards, Enfield India Ltd., (manufacturer of the then famous “Bullet” motorbike) started its manufacturing operations in India. This was followed by Ideal Jawa and Escorts Ltd. in the 1960s. The motorcycle segment registered a healthy upward trend during the 1960s and in the early 1970s, it accounted for 36% of the entire two-wheeler market. However, the motorcycle market in the 1960s and 1970s was a largely seller-dominated one and consumers had very little freedom of their own, all because of the “License Raj”. De-licensing of the sector in the early 1980s completely changed the structure of the two-wheeler industr y. With the new players entering the market, the choice of the consumers started to widen. Almost all the leading players in the market entered into strategic alliances with the Japanese two-wheeler manufacturers such as Hero (alliance with Honda), TVS Motors (alliance with Suzuki), Escorts (alliance with Yamaha) and Bajaj (alliance with Kawasaki). During the period from 1993 to 1999, the two-wheeler industry grew at a Compounded Annual Growth Rate (CAGR) of 14.6%, which was largely due to the contribution of the motorcycle segment, which grew at a CAGR of 24.3% compared to that of 11% for scooters. Fuel efficiency and stylized body were the two most important for the shift of demand from the scooters to motorbikes. Presently, the two-wheeler market in India is a more or less oligopolistic market with three major players controlling more than 80% market share. The leader among them is Hero Honda, a joint venture between the Hero Group of India and Honda Motors of Japan. Hero Honda is the largest two-wheeler manufacturer in the world
and has 37.9% market share in India (Presently the Hero Group and Honda are operating independently in India) (see Figure 1). The second largest player is Bajaj Auto Ltd. which is an indigenous company with a market share of 22.3%. The third player is TVS Motor Company with a market share of 20.9% and most of it coming from South India. Besides this, there are other players like Yamaha, Kinetic India Ltd. Royal Enfield Motors, etc.
Figure 1: Market Share of Players in Two-wheeler Market in India
0.9% 3.9%
Hero Honda Bajaj Auto
4.3% 5.8%
37.9% 20.9%
TVS Motors Honda Scooters Yamaha Majestic Auto Others
22.3%
Source: SIAM, 2003-04.
Methodology
Why Conjoint Analysis?
The origin of conjoint analysis dates back to the 1960s when a mathematical psychologist named Luce and a statistician named Tukey developed the concept in a joint paper in 1964. Thereafter conjoint analysis has been developed in the subsequent years by scholars like Kruskal, Carroll and Young in the late 1960s, Srinivasan and Green in the 1970s and Johnson in the 1980s. Today conjoint analysis is used extensively in marketing research for measuring and analyzing consumer preferences. Conjoint analysis can be defined as “a decompositional method that estimates the structure of a
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
49
consumer’s preferences, given his/her overall evaluations of a set of alternatives that are pre-specified in terms of levels of different attributes” (Green and Srinivasan, 1978). The advantage of conjoint analysis is that it can decompose the consumer’s overall preferences ordering into the preferences for any specific attribute and its different levels. It is this quality that makes it different from other methods like Factor Analysis and Discriminant Analysis, which are compositional methods. The benefit of using conjoint analysis is that it can be carried out at the individual level as well as group level. Lastly, an important advantage of conjoint analysis is its ability to simulate realistic situations as pointed out by Wittink and Cattin (1989). Though, a study similar to that done by the author was not found, there are studies which have tried to find out the relative importance of product attributes with the help of conjoint analysis. For example, a study by Lonial, Menezes and Ziam, in 2000, tried to find out the importance of attributes of a PC, which guides purchase behavior among students. Another study by Moskowitz (2001) tried to find out the consumer choice pattern on various attributes of fast food (burger). Conjoint analysis has been helpful in the services industr y as well. A study by Ross et al . (2003) applied conjoint analysis to identify the preference patterns regarding a recreational facility.
flexibility compared to other preference models like “ideal point” or “vector” model (Green and Srinivasan, 1978). In fact, the part-worth model can be converted into other models like the ideal point or vector model using suitable assumptions. In the given study, a part-worth model has been chosen because the study is an exploratory one. For the sake of simplicity, interaction effects have been assumed to be not present.
Conjoint Design
The design chosen was the full-profile conjoint, where hypothetical products have been developed using the complete set of attributes and levels. The “trade off ” or “two-factor-at-a-time” method is not selected here because of the problem of routinized response. Also, since the number of attributes or factors are low (four factors), it will not make the consumers face the problem of information overload. The basic advantage of the full profile method is that it gives a realistic picture of the stimuli (the hypothetical product) by defining the levels of each factor. Scholars such as Wittink and Cattin (1989), Cattin et al. have given favorable opinion about the full-profile conjoint method because of its flexibility and friendliness for numerical improvements. Green and Srinivasan (1978) have also recommended the use of full-profile conjoint if the number of attributes is six or less.
Choice of Preference Model
The preference model chosen here was the “part-worth” model. The “part-worth” model suggests that the consumer attaches a specific utility to one level of an attribute. The total utility of a multi-attribute product is some additive function of the part-worths of the attribute levels present in that particular product. The part-worth model is simple to explain and has greater
Development of the Stimulus Set
Having selected the design methodology, the question of preparing the stimulus set arises. Three important issues arise in this part: i. The decision about the number of attributes and their levels.
50
The ICFAI Journal of Marketing Management, February 2006
ii. The total number of stimuli to be used. iii.The construction of the stimuli. The decision about the number of attributes and their levels: Since the study was to find out the consumer preference patterns of motorbikes, an online search was conducted to identify the important attributes of a motorbike. After going through the specifications of 37 different brands, four attributes were finalized as: • Fuel efficiency • Color • Type of brake • Pickup Some attributes were not considered like displacement and power, since they had relationships with either fuel efficiency or pickup. Initially the levels of each attribute were as:
But the stimulus set it was generating consisted of minimum 25 stimuli which can lead to the problem of routinized response. So, some of the levels of the attributes were reduced without the loss of generality. Finally the attributes and their levels were:
• • • •
Fuel efficiency (in kmpl) 55 Color Conventional Brake type Disc + Drum Only Drum Customized 60 70
Pickup (0-60 kmph) in Sec. 6 7 8
The final stimulus set consisted of nine stimuli or hypothetical motorbike models as:
• • • •
Fuel efficiency Color Type of Brake Pickup :
: : :
4 levels 5 levels 2 levels 4 levels
Presentation of the Stimulus
The stimuli were presented in a structured questionnaire (see Table 1) to the respondents. The motorbike names mentioned as Bike A, Bike B etc. in the table are known as cards in conjoint
Table 1: The Full Profile Conjoint
Fuel Efficiency Bike Name (in kilometers per liter) Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I 55 kmpl 60 kmpl 70 kmpl 55 kmpl 60 kmpl 70 kmpl 55 kmpl 60 kmpl 70 kmpl Color Pick-up (0-60 km/hour) Brake Type Your Rank
Conventional Of your Choice Conventional Conventional Of your Choice Conventional Conventional Of your Choice Conventional
6 Seconds 7 Seconds 8 Seconds 7 Seconds 8 Seconds 6 Seconds 8 Seconds 6 Seconds 7 Seconds
Disc and Drum Only Drum Disc and Drum Only Drum Disc and Drum Disc and Drum Disc and Drum Disc and Drum Only Drum
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
51
terminology. Each card stands for a particular combination of attributes and is expected to evoke a particular utility (or a sense of utility) in the respondent’s mind. In case of small products or FMCGs, the researchers can have the option of providing the respondent with sample products with a particular set of attribute levels. This was not possible given the scale and scope of the study. The other method of presenting the stimulus is in a pictorial format which was not possible in this case because, except color, the other attributes cannot be represented pictorially. Thus the method adapted was one of the verbal representations. However, the interviewers were sufficiently equipped to answer queries of the respondents.
to identify the part-worth estimates. LINMAP is one of the traditional and most used methods of conjoint analysis and stands next to the Conjoint Analyzer algorithm developed by Bretton and Clarke. According to Green and Srinivasan (1978): “In LINMAP, attribute weights can be constrained to be non-negative, partworth functions can be constrained to be monotone or of the ideal point types, while s u c h constraints cannot be imposed for the other approaches.” (pg. 113) Various new methods of Hybrid Conjoint Analysis have been developed of late and one of the ver y popular methods used nowadays is the “Adaptive Conjoint Analysis” developed and maintained by the Sawtooth Software Corp. However, due to resource problem, the option of using adaptive conjoint analysis was negated.
Scale of Measurement
The scale of measurement is always an issue in case of conjoint analysis. In case of full profile conjoint, the scale may be a rating or ranking scale. Both rating scale and ranking scale have their own merits and demerits. While rating scale may give the respondent the freedom of giving the same rating to two combinations which he equally prefers, in case of ranking, he may be forced to rank them differently. However, the ranking scale forces a consumer to think about his/her preferences deeply and then come to a decision of ranking. This helps in avoiding the problem of routinized responses. This is one of the reasons why ranked data are assumed to be more reliable (Green and Srinivasan, 1978). Thus, the respondents were asked to rank the stimuli/combinations in their order of preference, where rank 1 means most preferred and rank 9 means least preferred.
Data Collection
The targeted sample size was 100, which can be considered as a sufficient sample size for the study. The author, along with some help from his students, selected the respondents randomly but with the help of convenient sampling. The total number of completely filled up questionnaires received was 88. The sample consisted of 56 students and 32 nonstudents.
Results
The overall weights assigned to the different attributes for the total group, student group and nonstudent group follow a similar pattern i.e., Fuel Efficiency > Type of Brakes > Pick-up > Color (> stands for ‘preferred to) But it was found out that the Nonstudent group gave much more importance to Fuel
Estimation Method
The estimation method used here is LINMAP which uses linear programming
52
The ICFAI Journal of Marketing Management, February 2006
efficiency and relatively less to color and brakes as compared to the student group. (see Table 2)
• •
This is followed by type of brakes which is more of a safety purpose. There exists significant difference in preference pattern of the students and the nonstudents.
Table 2: Weight Assigned to Different Attributes (in %)
Group Overall Student Nonstudent Fuel Color Brakes Efficiency 37.53 34.11 44.68 16.64 18.75 11.93 25.70 27.50 23.36 Pick-up Total 20.13 19.64 20.04
On the basis of the findings we can find out the total utility of each combination for each group. The model used here is 100 an additive model which can be 100 mathematically represented as:
100
Ui= PFj + PCk + PBl + PPm where, = Utility of the ith Combination = Part-worth of j th level of Fuel Efficiency
This indicates that style (which can be defined by color) and Safety (which can be defined by the type of brakes) are more important to the students rather than mileage (defined by fuel efficiency). The part-worth estimates of the different levels of each attribute are given in Table 3. When it came to the part-worth of different levels of attributes, it was found out that, there existed significant differences in the part-worth assigned to the most preferred
Ui PF j
PCk = Part-worth of kth level of Color PBl = Part-worth of the lth level of Brake Type PPm = Part-worth of the mth level of Pickup Using the part worth estimates given in Table 3 the utilities for the different
Table 3: The Part-worth Estimates
Group Fuel Efficiency (kmpl)
55 60 70
Color
Conven Custotional mized
Brakes
Disc+ Drum Only Drum
Pick-up (0-60) in seconds
6 7 8
Overall -17.849 -1.827 Student -16.024 -2.058 Non-21.601 -1.481 Student
19.675 -8.318 8.318 18.082 -9.376 9.376 23.082 -5.963 5.963
12.852 -12.852
7.490 5.513 -12.643
13.752 -13.752 7.137 5.363 -12.500 11.678 -11.678 7.810 4.415 -12.225
level of fuel efficiency (70 kmpl) and the most preferred level of color (customized) among the students and nonstudents.
combinations for different groups were calculated which are represented in Table 4. (see Appendix for details of each group). One observation is made here that the expected ranks of the combinations are similar for overall and the student group, but the nonstudent group has two combinations with different ranks to those of the students. This validates (to some extent) the point that
Summary of Findings
The summary of findings from the study is:
•
Fuel efficiency is the most important attribute of consumer choices regarding motorbike.
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
53
Table 4: Total Utility and Expected Ranks
Group Bike Name Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I Total Utility -5.825 -0.848 11.566 -33.506 6.7 31.699 -25.958 26.833 4.018 Overall Expected Rank 7 6 3 9 4 1 8 2 5 Student Total Utility -4.511 -1.071 9.958 -33.789 8.57 29.595 -24.148 28.207 0.317 Expected Rank 7 6 3 9 4 1 8 2 5 Non-student Total Utility -8.076 -2.781 16.572 -34.827 3.935 36.607 -28.111 23.97 9.856 Expected Rank 7 6 3 9 5 1 8 2 4
there are differences in preferences among the students and nonstudents.
Limitations
Before concluding the study, the limitations need to be discussed. First, the study was conducted in Hyderabad only and that may lead to some regional bias towards any attribute. Second, the part-worth model was used only, whereas in case of some attributes like fuel efficiency and pickup, vector model could have been used. Third, individual choice patterns were not studied in detail.
awareness about safety and people want safer vehicles. Lastly it can be concluded from the study that the motorbike market can be segmented into student and nonstudent section because there is a difference in the preference pattern of the students and nonstudents. Thus, any company which wants to target a particular segment should change its marketing plan to suit the needs of its target group. %
Reference # 03J-2006-02-06-01
References
1. Carroll Douglas J and Green Paul E (1995). “Psychometric Methods in Marketing Research: Part I, Conjoint Analysis.” Journal of Marketing Research, November, Vol. 32, Issue 4, p. 385. 2. Cattin Philippe; Gelfand Alan E and Danes Jeffrey (1983). “A Simple Bayesian Procedure for Estimation in a Conjoint Model”. Journal of Marketing Research , February, Vol. 20, Issue 1, p. 29.
Conclusion
In spite of the limitations, the study can act as a stepping stone of a study with more detail and rigor. Because the findings suggest that any company in the motorbike business should concentrate on fuel efficiency as that is most important to any consumer. Then it should focus on the safety aspect as there is an increasing
54
The ICFAI Journal of Marketing Management, February 2006
3. Dey Abhijeet (2000). “Models on Two Wheels.” www.indiainfoline.com, Januar y, collected from http:// w w w. i n d i a i n f o l i n e . c o m / n e v i / twoo.html on 03.03.2005. 4. Green Paul E and Krieger Abba M, (1991). “Adaptive Conjoint Analysis: Some Caveats and Suggestions.” Journal of Marketing Research , May, Vol. 28, Issue 2, p. 215. 5. Green Paul E and Krieger Abba M, (1991). “Segmenting Markets with Conjoint Analysis.” Journal of Marketing, October, Vol. 55, Issue 4, p. 20. 6. Green Paul E and Srinivasan V (1978). “Conjoint Analysis in Consumer Research: Issues and Outlook”. Journal of Consumer Research, September, Vol. 5, Issue 2, p. 103. 7. Green Paul E and Srinivasan V (1990). “Conjoint Analysis in Marketing: New Developments with Implications for Research and Practice”. Journal of Marketing , October, Vol. 54, Issue 4, p. 3. 8. Lonial Subhash, Menezes Dennis and Zaim Selim (2 0 0 0 ).
“Identifying Purchase Driving Attributes and Market Segments fo r P C s U s i n g C o n j o i n t a n d Cluster Analysis.” Journal of Economic and Social Research , July, Vol. 2, Issue 2, p. 19. 9. Moskowitz Howard (2001). “Creating New Product Concepts for Food Service–The Role of Conjoint Measurement to Identify Promising Product Features”. Food Service Technology, Spring, Vol. 1, Issue 1, p. 35. 10. Ross Stephen D, Norman William C and Dorsch Michael J (2003). “The Use of Conjoint Analysis in the Development of a New Recreation Facility”. Managing Leisure, October, Vol. 8, Issue 4, p. 227. 11. Wittink Dick R and Cattin Philippe (1989). “Commercial Use of Conjoint Analysis: An Update”, Journal of Marketing, July, Vol. 53, Issue 3.
Web References
1. 2. 3. www.autodrive.com www.indiacar.com www.siam.com
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
55
Appendix
Total Utility and Part-worth Estimates for the Overall Group
Bike Name Fuel Efficiency Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -17.849 -1.827 19.675 -17.849 -1.827 19.675 -17.849 -1.827 19.675 Color -8.318 8.318 -8.318 -8.318 8.318 -8.318 -8.318 8.318 -8.318 Brake Type 12.852 -12.852 12.852 -12.852 12.852 12.852 12.852 12.852 -12.852 Pick-up 7.49 5.513 -12.643 5.513 -12.643 7.49 -12.643 7.49 5.513 Total -5.825 -0.848 11.566 -33.506 6.7 31.699 -25.958 26.833 4.018 Expected Rank 7 6 3 9 4 1 8 2 5
Total Utility and Part-worth Estimates for the Student Group
Bike Name Fuel Efficiency Colour Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -16.024 -2.058 18.082 -16.024 -2.058 18.082 -16.024 -2.058 18.082 -9.376 9.376 -9.376 -9.376 9.376 -9.376 -9.376 9.376 -9.376 Brake Type 13.752 -13.752 13.752 -13.752 13.752 13.752 13.752 13.752 -13.752 Pick-up 7.137 5.363 -12.5 5.363 -12.5 7.137 -12.5 7.137 5.363 Total -4.511 -1.071 9.958 -33.789 8.57 29.595 -24.148 28.207 0.317 Expected Rank 7 6 3 9 4 1 8 2 5
Total Utility and Part-worth Estimates for the Non-student Group
Bike Name Fuel Efficiency Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -21.601 -1.481 23.082 -21.601 -1.481 23.082 -21.601 -1.481 23.082 Color -5.963 5.963 -5.963 -5.963 5.963 -5.963 -5.963 5.963 -5.963 Brake Type 11.678 -11.678 11.678 -11.678 11.678 11.678 11.678 11.678 -11.678 Pick-up 7.81 4.415 -12.225 4.415 -12.225 7.81 -12.225 7.81 4.415 Total -8.076 -2.781 16.572 -34.827 3.935 36.607 -28.111 23.97 9.856 Expected Rank 7 6 3 9 5 1 8 2 4
56
The ICFAI Journal of Marketing Management, February 2006
doc_782619300.pdf
This study is based to identify the factors that influence the consumer buying behavior
of the two wheeler Bikes at Allahabad, Lucknow and Varanasi cities of Uttar Pradesh. The
survey is mainly focused on the buying behavior of the consumer that motivates them to
purchase the two wheeler bikes.
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis!
Subhadip Roy*
The Indian motorbike industry is in a very healthy state today. With foreign players like Honda operating independently, the market is becoming extremely competitive. This is an exploratory study which identifies the consumer choice patterns regarding the four attributes of a motorbike: Fuel efficiency, color, type of brakes and pick-up. The objective is to find out the consumer preference ordering of the four attributes mentioned. Conjoint Analysis was used to find out the different aspects of consumer choice regarding motorbikes. This study is based on a field survey conducted on 88 respondents in and around Hyderabad.
Motivation and Objective
The Indian two-wheeler industry is set to become more competitive as most of the strategic alliances have broken and the foreign are operating independently. In such a scenario, the players need to know the importance of the attributes of a motorbike and whether there are any differences of choice pattern in different segments of the consumers. This provided the motivation for this study. The objective was to find out the consumer preference ordering of various attributes of a motorbike and applicability of conjoint analysis in finding that out and the probable use of the tool in new product development and market segmentation. This study was based on a field survey conducted in and around Hyderabad.
!
The Indian Two-wheeler Industry
The Indian two-wheeler industry has come up a long way from being a tightly government controlled industry to a highly competitive one encouraging foreign players’ participation. Today, the Indian consumer has a wide variety of brands and models to choose from when he/she thinks of buying a two-wheeler. Approximately 5.4 million units of two-wheelers are sold every year in India, making it one of the largest two-wheeler markets in the world. Not only that, India has an average of 27 two-wheelers per 1000 people, making it the highest in Asia.1 The origin of Indian two-wheeler industry can be traced back to 1948, when Bajaj
A part of this paper was presented at the 17th AIMS Annual Convention held in Hyderabad during August 28-30, 2005.
About the Author
*
Research Scholar, The ICFAI Institute for Management Teachers (IIMT), Hyderabad. E-mail: [email protected]
1
According to data published by the World Bank.
© 2006 The ICFAI University Press. All Rights Reserved.
48 The ICFAI Journal of Marketing Management, February 2006
Auto Ltd. started importing and marketing Vespa scooters in India. Shortly afterwards, Enfield India Ltd., (manufacturer of the then famous “Bullet” motorbike) started its manufacturing operations in India. This was followed by Ideal Jawa and Escorts Ltd. in the 1960s. The motorcycle segment registered a healthy upward trend during the 1960s and in the early 1970s, it accounted for 36% of the entire two-wheeler market. However, the motorcycle market in the 1960s and 1970s was a largely seller-dominated one and consumers had very little freedom of their own, all because of the “License Raj”. De-licensing of the sector in the early 1980s completely changed the structure of the two-wheeler industr y. With the new players entering the market, the choice of the consumers started to widen. Almost all the leading players in the market entered into strategic alliances with the Japanese two-wheeler manufacturers such as Hero (alliance with Honda), TVS Motors (alliance with Suzuki), Escorts (alliance with Yamaha) and Bajaj (alliance with Kawasaki). During the period from 1993 to 1999, the two-wheeler industry grew at a Compounded Annual Growth Rate (CAGR) of 14.6%, which was largely due to the contribution of the motorcycle segment, which grew at a CAGR of 24.3% compared to that of 11% for scooters. Fuel efficiency and stylized body were the two most important for the shift of demand from the scooters to motorbikes. Presently, the two-wheeler market in India is a more or less oligopolistic market with three major players controlling more than 80% market share. The leader among them is Hero Honda, a joint venture between the Hero Group of India and Honda Motors of Japan. Hero Honda is the largest two-wheeler manufacturer in the world
and has 37.9% market share in India (Presently the Hero Group and Honda are operating independently in India) (see Figure 1). The second largest player is Bajaj Auto Ltd. which is an indigenous company with a market share of 22.3%. The third player is TVS Motor Company with a market share of 20.9% and most of it coming from South India. Besides this, there are other players like Yamaha, Kinetic India Ltd. Royal Enfield Motors, etc.
Figure 1: Market Share of Players in Two-wheeler Market in India
0.9% 3.9%
Hero Honda Bajaj Auto
4.3% 5.8%
37.9% 20.9%
TVS Motors Honda Scooters Yamaha Majestic Auto Others
22.3%
Source: SIAM, 2003-04.
Methodology
Why Conjoint Analysis?
The origin of conjoint analysis dates back to the 1960s when a mathematical psychologist named Luce and a statistician named Tukey developed the concept in a joint paper in 1964. Thereafter conjoint analysis has been developed in the subsequent years by scholars like Kruskal, Carroll and Young in the late 1960s, Srinivasan and Green in the 1970s and Johnson in the 1980s. Today conjoint analysis is used extensively in marketing research for measuring and analyzing consumer preferences. Conjoint analysis can be defined as “a decompositional method that estimates the structure of a
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
49
consumer’s preferences, given his/her overall evaluations of a set of alternatives that are pre-specified in terms of levels of different attributes” (Green and Srinivasan, 1978). The advantage of conjoint analysis is that it can decompose the consumer’s overall preferences ordering into the preferences for any specific attribute and its different levels. It is this quality that makes it different from other methods like Factor Analysis and Discriminant Analysis, which are compositional methods. The benefit of using conjoint analysis is that it can be carried out at the individual level as well as group level. Lastly, an important advantage of conjoint analysis is its ability to simulate realistic situations as pointed out by Wittink and Cattin (1989). Though, a study similar to that done by the author was not found, there are studies which have tried to find out the relative importance of product attributes with the help of conjoint analysis. For example, a study by Lonial, Menezes and Ziam, in 2000, tried to find out the importance of attributes of a PC, which guides purchase behavior among students. Another study by Moskowitz (2001) tried to find out the consumer choice pattern on various attributes of fast food (burger). Conjoint analysis has been helpful in the services industr y as well. A study by Ross et al . (2003) applied conjoint analysis to identify the preference patterns regarding a recreational facility.
flexibility compared to other preference models like “ideal point” or “vector” model (Green and Srinivasan, 1978). In fact, the part-worth model can be converted into other models like the ideal point or vector model using suitable assumptions. In the given study, a part-worth model has been chosen because the study is an exploratory one. For the sake of simplicity, interaction effects have been assumed to be not present.
Conjoint Design
The design chosen was the full-profile conjoint, where hypothetical products have been developed using the complete set of attributes and levels. The “trade off ” or “two-factor-at-a-time” method is not selected here because of the problem of routinized response. Also, since the number of attributes or factors are low (four factors), it will not make the consumers face the problem of information overload. The basic advantage of the full profile method is that it gives a realistic picture of the stimuli (the hypothetical product) by defining the levels of each factor. Scholars such as Wittink and Cattin (1989), Cattin et al. have given favorable opinion about the full-profile conjoint method because of its flexibility and friendliness for numerical improvements. Green and Srinivasan (1978) have also recommended the use of full-profile conjoint if the number of attributes is six or less.
Choice of Preference Model
The preference model chosen here was the “part-worth” model. The “part-worth” model suggests that the consumer attaches a specific utility to one level of an attribute. The total utility of a multi-attribute product is some additive function of the part-worths of the attribute levels present in that particular product. The part-worth model is simple to explain and has greater
Development of the Stimulus Set
Having selected the design methodology, the question of preparing the stimulus set arises. Three important issues arise in this part: i. The decision about the number of attributes and their levels.
50
The ICFAI Journal of Marketing Management, February 2006
ii. The total number of stimuli to be used. iii.The construction of the stimuli. The decision about the number of attributes and their levels: Since the study was to find out the consumer preference patterns of motorbikes, an online search was conducted to identify the important attributes of a motorbike. After going through the specifications of 37 different brands, four attributes were finalized as: • Fuel efficiency • Color • Type of brake • Pickup Some attributes were not considered like displacement and power, since they had relationships with either fuel efficiency or pickup. Initially the levels of each attribute were as:
But the stimulus set it was generating consisted of minimum 25 stimuli which can lead to the problem of routinized response. So, some of the levels of the attributes were reduced without the loss of generality. Finally the attributes and their levels were:
• • • •
Fuel efficiency (in kmpl) 55 Color Conventional Brake type Disc + Drum Only Drum Customized 60 70
Pickup (0-60 kmph) in Sec. 6 7 8
The final stimulus set consisted of nine stimuli or hypothetical motorbike models as:
• • • •
Fuel efficiency Color Type of Brake Pickup :
: : :
4 levels 5 levels 2 levels 4 levels
Presentation of the Stimulus
The stimuli were presented in a structured questionnaire (see Table 1) to the respondents. The motorbike names mentioned as Bike A, Bike B etc. in the table are known as cards in conjoint
Table 1: The Full Profile Conjoint
Fuel Efficiency Bike Name (in kilometers per liter) Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I 55 kmpl 60 kmpl 70 kmpl 55 kmpl 60 kmpl 70 kmpl 55 kmpl 60 kmpl 70 kmpl Color Pick-up (0-60 km/hour) Brake Type Your Rank
Conventional Of your Choice Conventional Conventional Of your Choice Conventional Conventional Of your Choice Conventional
6 Seconds 7 Seconds 8 Seconds 7 Seconds 8 Seconds 6 Seconds 8 Seconds 6 Seconds 7 Seconds
Disc and Drum Only Drum Disc and Drum Only Drum Disc and Drum Disc and Drum Disc and Drum Disc and Drum Only Drum
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
51
terminology. Each card stands for a particular combination of attributes and is expected to evoke a particular utility (or a sense of utility) in the respondent’s mind. In case of small products or FMCGs, the researchers can have the option of providing the respondent with sample products with a particular set of attribute levels. This was not possible given the scale and scope of the study. The other method of presenting the stimulus is in a pictorial format which was not possible in this case because, except color, the other attributes cannot be represented pictorially. Thus the method adapted was one of the verbal representations. However, the interviewers were sufficiently equipped to answer queries of the respondents.
to identify the part-worth estimates. LINMAP is one of the traditional and most used methods of conjoint analysis and stands next to the Conjoint Analyzer algorithm developed by Bretton and Clarke. According to Green and Srinivasan (1978): “In LINMAP, attribute weights can be constrained to be non-negative, partworth functions can be constrained to be monotone or of the ideal point types, while s u c h constraints cannot be imposed for the other approaches.” (pg. 113) Various new methods of Hybrid Conjoint Analysis have been developed of late and one of the ver y popular methods used nowadays is the “Adaptive Conjoint Analysis” developed and maintained by the Sawtooth Software Corp. However, due to resource problem, the option of using adaptive conjoint analysis was negated.
Scale of Measurement
The scale of measurement is always an issue in case of conjoint analysis. In case of full profile conjoint, the scale may be a rating or ranking scale. Both rating scale and ranking scale have their own merits and demerits. While rating scale may give the respondent the freedom of giving the same rating to two combinations which he equally prefers, in case of ranking, he may be forced to rank them differently. However, the ranking scale forces a consumer to think about his/her preferences deeply and then come to a decision of ranking. This helps in avoiding the problem of routinized responses. This is one of the reasons why ranked data are assumed to be more reliable (Green and Srinivasan, 1978). Thus, the respondents were asked to rank the stimuli/combinations in their order of preference, where rank 1 means most preferred and rank 9 means least preferred.
Data Collection
The targeted sample size was 100, which can be considered as a sufficient sample size for the study. The author, along with some help from his students, selected the respondents randomly but with the help of convenient sampling. The total number of completely filled up questionnaires received was 88. The sample consisted of 56 students and 32 nonstudents.
Results
The overall weights assigned to the different attributes for the total group, student group and nonstudent group follow a similar pattern i.e., Fuel Efficiency > Type of Brakes > Pick-up > Color (> stands for ‘preferred to) But it was found out that the Nonstudent group gave much more importance to Fuel
Estimation Method
The estimation method used here is LINMAP which uses linear programming
52
The ICFAI Journal of Marketing Management, February 2006
efficiency and relatively less to color and brakes as compared to the student group. (see Table 2)
• •
This is followed by type of brakes which is more of a safety purpose. There exists significant difference in preference pattern of the students and the nonstudents.
Table 2: Weight Assigned to Different Attributes (in %)
Group Overall Student Nonstudent Fuel Color Brakes Efficiency 37.53 34.11 44.68 16.64 18.75 11.93 25.70 27.50 23.36 Pick-up Total 20.13 19.64 20.04
On the basis of the findings we can find out the total utility of each combination for each group. The model used here is 100 an additive model which can be 100 mathematically represented as:
100
Ui= PFj + PCk + PBl + PPm where, = Utility of the ith Combination = Part-worth of j th level of Fuel Efficiency
This indicates that style (which can be defined by color) and Safety (which can be defined by the type of brakes) are more important to the students rather than mileage (defined by fuel efficiency). The part-worth estimates of the different levels of each attribute are given in Table 3. When it came to the part-worth of different levels of attributes, it was found out that, there existed significant differences in the part-worth assigned to the most preferred
Ui PF j
PCk = Part-worth of kth level of Color PBl = Part-worth of the lth level of Brake Type PPm = Part-worth of the mth level of Pickup Using the part worth estimates given in Table 3 the utilities for the different
Table 3: The Part-worth Estimates
Group Fuel Efficiency (kmpl)
55 60 70
Color
Conven Custotional mized
Brakes
Disc+ Drum Only Drum
Pick-up (0-60) in seconds
6 7 8
Overall -17.849 -1.827 Student -16.024 -2.058 Non-21.601 -1.481 Student
19.675 -8.318 8.318 18.082 -9.376 9.376 23.082 -5.963 5.963
12.852 -12.852
7.490 5.513 -12.643
13.752 -13.752 7.137 5.363 -12.500 11.678 -11.678 7.810 4.415 -12.225
level of fuel efficiency (70 kmpl) and the most preferred level of color (customized) among the students and nonstudents.
combinations for different groups were calculated which are represented in Table 4. (see Appendix for details of each group). One observation is made here that the expected ranks of the combinations are similar for overall and the student group, but the nonstudent group has two combinations with different ranks to those of the students. This validates (to some extent) the point that
Summary of Findings
The summary of findings from the study is:
•
Fuel efficiency is the most important attribute of consumer choices regarding motorbike.
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
53
Table 4: Total Utility and Expected Ranks
Group Bike Name Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I Total Utility -5.825 -0.848 11.566 -33.506 6.7 31.699 -25.958 26.833 4.018 Overall Expected Rank 7 6 3 9 4 1 8 2 5 Student Total Utility -4.511 -1.071 9.958 -33.789 8.57 29.595 -24.148 28.207 0.317 Expected Rank 7 6 3 9 4 1 8 2 5 Non-student Total Utility -8.076 -2.781 16.572 -34.827 3.935 36.607 -28.111 23.97 9.856 Expected Rank 7 6 3 9 5 1 8 2 4
there are differences in preferences among the students and nonstudents.
Limitations
Before concluding the study, the limitations need to be discussed. First, the study was conducted in Hyderabad only and that may lead to some regional bias towards any attribute. Second, the part-worth model was used only, whereas in case of some attributes like fuel efficiency and pickup, vector model could have been used. Third, individual choice patterns were not studied in detail.
awareness about safety and people want safer vehicles. Lastly it can be concluded from the study that the motorbike market can be segmented into student and nonstudent section because there is a difference in the preference pattern of the students and nonstudents. Thus, any company which wants to target a particular segment should change its marketing plan to suit the needs of its target group. %
Reference # 03J-2006-02-06-01
References
1. Carroll Douglas J and Green Paul E (1995). “Psychometric Methods in Marketing Research: Part I, Conjoint Analysis.” Journal of Marketing Research, November, Vol. 32, Issue 4, p. 385. 2. Cattin Philippe; Gelfand Alan E and Danes Jeffrey (1983). “A Simple Bayesian Procedure for Estimation in a Conjoint Model”. Journal of Marketing Research , February, Vol. 20, Issue 1, p. 29.
Conclusion
In spite of the limitations, the study can act as a stepping stone of a study with more detail and rigor. Because the findings suggest that any company in the motorbike business should concentrate on fuel efficiency as that is most important to any consumer. Then it should focus on the safety aspect as there is an increasing
54
The ICFAI Journal of Marketing Management, February 2006
3. Dey Abhijeet (2000). “Models on Two Wheels.” www.indiainfoline.com, Januar y, collected from http:// w w w. i n d i a i n f o l i n e . c o m / n e v i / twoo.html on 03.03.2005. 4. Green Paul E and Krieger Abba M, (1991). “Adaptive Conjoint Analysis: Some Caveats and Suggestions.” Journal of Marketing Research , May, Vol. 28, Issue 2, p. 215. 5. Green Paul E and Krieger Abba M, (1991). “Segmenting Markets with Conjoint Analysis.” Journal of Marketing, October, Vol. 55, Issue 4, p. 20. 6. Green Paul E and Srinivasan V (1978). “Conjoint Analysis in Consumer Research: Issues and Outlook”. Journal of Consumer Research, September, Vol. 5, Issue 2, p. 103. 7. Green Paul E and Srinivasan V (1990). “Conjoint Analysis in Marketing: New Developments with Implications for Research and Practice”. Journal of Marketing , October, Vol. 54, Issue 4, p. 3. 8. Lonial Subhash, Menezes Dennis and Zaim Selim (2 0 0 0 ).
“Identifying Purchase Driving Attributes and Market Segments fo r P C s U s i n g C o n j o i n t a n d Cluster Analysis.” Journal of Economic and Social Research , July, Vol. 2, Issue 2, p. 19. 9. Moskowitz Howard (2001). “Creating New Product Concepts for Food Service–The Role of Conjoint Measurement to Identify Promising Product Features”. Food Service Technology, Spring, Vol. 1, Issue 1, p. 35. 10. Ross Stephen D, Norman William C and Dorsch Michael J (2003). “The Use of Conjoint Analysis in the Development of a New Recreation Facility”. Managing Leisure, October, Vol. 8, Issue 4, p. 227. 11. Wittink Dick R and Cattin Philippe (1989). “Commercial Use of Conjoint Analysis: An Update”, Journal of Marketing, July, Vol. 53, Issue 3.
Web References
1. 2. 3. www.autodrive.com www.indiacar.com www.siam.com
Consumer Choice of Motorbike Attributes: An Application of Conjoint Analysis
55
Appendix
Total Utility and Part-worth Estimates for the Overall Group
Bike Name Fuel Efficiency Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -17.849 -1.827 19.675 -17.849 -1.827 19.675 -17.849 -1.827 19.675 Color -8.318 8.318 -8.318 -8.318 8.318 -8.318 -8.318 8.318 -8.318 Brake Type 12.852 -12.852 12.852 -12.852 12.852 12.852 12.852 12.852 -12.852 Pick-up 7.49 5.513 -12.643 5.513 -12.643 7.49 -12.643 7.49 5.513 Total -5.825 -0.848 11.566 -33.506 6.7 31.699 -25.958 26.833 4.018 Expected Rank 7 6 3 9 4 1 8 2 5
Total Utility and Part-worth Estimates for the Student Group
Bike Name Fuel Efficiency Colour Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -16.024 -2.058 18.082 -16.024 -2.058 18.082 -16.024 -2.058 18.082 -9.376 9.376 -9.376 -9.376 9.376 -9.376 -9.376 9.376 -9.376 Brake Type 13.752 -13.752 13.752 -13.752 13.752 13.752 13.752 13.752 -13.752 Pick-up 7.137 5.363 -12.5 5.363 -12.5 7.137 -12.5 7.137 5.363 Total -4.511 -1.071 9.958 -33.789 8.57 29.595 -24.148 28.207 0.317 Expected Rank 7 6 3 9 4 1 8 2 5
Total Utility and Part-worth Estimates for the Non-student Group
Bike Name Fuel Efficiency Bike A Bike B Bike C Bike D Bike E Bike F Bike G Bike H Bike I -21.601 -1.481 23.082 -21.601 -1.481 23.082 -21.601 -1.481 23.082 Color -5.963 5.963 -5.963 -5.963 5.963 -5.963 -5.963 5.963 -5.963 Brake Type 11.678 -11.678 11.678 -11.678 11.678 11.678 11.678 11.678 -11.678 Pick-up 7.81 4.415 -12.225 4.415 -12.225 7.81 -12.225 7.81 4.415 Total -8.076 -2.781 16.572 -34.827 3.935 36.607 -28.111 23.97 9.856 Expected Rank 7 6 3 9 5 1 8 2 4
56
The ICFAI Journal of Marketing Management, February 2006
doc_782619300.pdf