YEAR 2006

By C. V. Phani Kumar and Bhargab Maitra


Road users perceived values associated with various attributes of travel are essential inputs for judicious improvement planning of roads and transportation systems. In the present paper choice based conjoint analysis is used to model perceived values associates with various attributes of travel with reference to a bus route. Using ‘effects codes’, an attempt is also made to understand the variations in perceived values with different levels of attributes. The methodology presented in the paper can be applied for improvements in planning of transportation facilities like introduction of new transport services, highways, parking lots etc.


Transportation sector in India has gained great attention in the last few years. Government and private investors are investing substantial amounts of money for improvements of transport infrastructure. These improvements bring benefits to the end users. Benefits may be in the form of savings in travel time, improved comfort, less fuel consumption due to free flow conditions and improved road surfaces. Investments on transportation infrastructures are normally recovered by charging the end users either in the form of toll or higher fares. While charging the end users, it is essential to know users willingness to pay for the improvements. Therefore, it is necessary to understand user’s perceived values associated with various attributes of service, their influence on choice behaviour of trip makers, and utilise the same for judicious improvement planning of transportation service. Many researchers have attempted to model people’s perceptions about various attributes of travel using Revealed Preference (RP) and/or Stated Preference (SP) data (Bates 1982; Kroes and Sheldon 1988; Louviere 1988; Hensher 1994; Hunt 2001; Jose Holguin-Veras 2002; Praveen and Rao, 2002; TEMS 2001).

Generally attributes’ influence is assumed to be linear but it may not be so in reality. The objectives of the present paper are, modeling the trip maker’s perceptions about travel attributes with the application of stated choice method, and understand the variation or non-linearity in the perceived values associated with travel attributes in choice decisions of services by “effects coding” the attribute levels.

In the present paper, modeling the trip maker’s perceptions about travel attributes has been demonstrated with reference to a rural bus route, which connects the district head quarter (Midnapur) and a recreation place (Digha) and passes through around 30 habitations of which some have business importance (Contai, Belda) and covers a distance of 142 km. So it was felt that all kinds of trips viz. job, business, recreation, education and other trips can be captured and also the effect of distance on attribute values can also be studied for the route. The travel demand along this route is largely served by the bus service. The bus service takes about 5 hours to cover the distance of 142 km. and serves about 35 intermediate stops.


The study involves design of experiment, data collection technique, preparation of database, and analysis of the data. The methodologies followed for each of these steps are mentioned below.

3.1. Stated Preference Approach

Conjoint analysis is a marketing research technique where products are decomposed into their component parts to analyse how decisions are made and how decisions are influenced by the inclusion or exclusion of an attribute or its level. This technique is also used in transportation studies. One of the important approaches is Choice-based conjoint or Stated Choice Method. Stated preference techniques have largely been used in wide range of disciplines like transportation (Hensher 1994; Ka-hung Lai and Wing-gun Wong 2000), environmental (Opaluch et al. 1993; Adamowicz et al. 1998) and product marketing (food, home appliances etc.). Stated Choice Method (SCM) having strong theoretical foundation based on economic theory is used to model the behavior of individuals. The SCM facilitates estimation of importance of each attribute from peoples’ responses as they do trade-off among the alternatives, represented by various attributes and their levels, in the form of choice sets. These methods also facilitate to analyze how decisions vary with variations in the magnitude of the attributes to model consumer surplus. In this study, different profiles are generated using various attributes with different levels and presented to the respondent in the form of choice set. The responses in the form of ‘choice’ among the presented choice alternatives are used for modeling. Attribute levels are effects coded (-1,+1) so that it can produce the coefficients for each level of attribute to estimate part-worth values for attribute levels which help to understand the nonlinearity associate with the levels of attributes.

3.2. Theoretical Framework

Random Utility Theory (RUT) (Thurstone 1927; McFadden 1974), the basis for several models and theories of decision-making in psychology and economics, states that the utility of each element consists of an observed (deterministic) component denoted by V and a random (disturbance) component denoted by e,
U = V + e
The deterministic part V is again a function of the observed attributes (z) of the choice as faced by the individual, the observed socioeconomic attributes of the individual (S) and a vector of parameters (b), then
V = V (z, S, b)

A probabilistic statement can be made (due to presence of the random component) as, when an individual ‘n’ is facing a choice set, Cn, consisting of Jn choices, the choice probability of alternative i is equal to the probability that the utility of alternative ‘i’, Uin, is greater than or equal to the utilities of all other alternatives in the choice set. i.e.

3.3. Stated Choice Experiment Design

For the study of the hypothetical choice decisions by the trip makers along the study route, a set of important service characteristics/attributes is identified in consultation with trip makers and concerned experts. Attributes considered initially were Travel time, Fare, Headway and Comfort, Reliability, Noise level inside the bus, and Appearance of the bus. But, it is found from the response from trip makers that appearance and noise level do not really have impact on their choice. The attributes, then identified as having major influence on decision, are

Travel time, Fare, Headway and Comfort for the purpose of development of choice options. Each attribute is specified at three levels. Attributes, fare and travel time are defined as percentage deviation from existing fare and travel time respectively. The decision attributes and their levels are shown in Table 1.

A set of four attributes each at three levels produce 34 full factorial design of 81 combinations. Full fractional designs are those where each level of each attribute is combined with every level of all other attributes. However, full factorial designs produce large number of options, which are difficult to study, and not all options are competitive to each other. Fractional factorial technique on the other hand reduces the number of sets without much loss of statistical information given by full factorial designs. Fractional factorial orthogonal design (SPSS 7.5) has produced 9 combinations from levels of attributes presented in Table 1. The orthogonal main effects only design imposes independence between the attributes and assumes that interaction effects are negligible. Three sets each containing three randomly drawn combinations from the total of 9 combinations, are prepared. Measures were taken to make sure that there is no obvious option in the choice set. Each respondent is shown all the three sets in the survey form to ‘state’ his/her ‘choice’ for each set. A sample choice set is shown in figure 1.


Enumerators are trained and used to collect data from various locations along the study route by interviewing trip makers. Twelve locations are identified for conducting interviews of the trip makers. Primary data includes the socio-economic information such as age, gender, income, social status etc. of the respondents, Trip characteristics such as origin-destination, purpose of the trip, fare paid, time spent in the bus etc. and the ‘choice’ from the choice set. Secondary data related to mode specific and route specific is collected from sources like Regional Transport Authority (RTA) and bus operating agencies.


Normally, during behavioural surveys all respondents do not give consistent responses. Therefore it is necessary to have a consistency check on the data obtained from field surveys and use only the data/responses, which are consistent. Several other questions are included in the questionnaire to check the consistency of the responses. Consistency checks were made to remove the impure data based on the rank information given by the respondents. Removal of such data resulted in 76 valid samples for the purpose of analysis. Though socioeconomic parameters influence the values of attributes, those effects are not studied in the present paper. Dummy variables are used to effects code (-1, +1) the attribute levels. In order to avoid singularity, the third level of each attribute is omitted during estimation. The coefficient of third level is calculated later using the property of effects coding that sum of coefficients of all levels is equal to zero (Louviere et al. 2000). The model used for estimation of coefficients of attributes’ levels is Discrete Choice Multinomial Logit Model of LIMDEP 8.0 & NLOGIT 3.0 (Greene 2002). All attributes are effects coded excluding price attribute so that the coefficient of fare can be utilized in estimating the values associated with the changes in the levels of other attributes. The results are shown in Table 2.


The stated choice approach used in this case study provided information on attributes and their respective levels. Model’s goodness of fit is indicated by the pseudo-R2 (ρ2). ρ2 value between 0.2 and 0.4 indicates acceptable model fits. Also, the t-value, 1.96 or above indicates that the mean/coefficient is statistically significantly different from zero at 95% or greater confidence level. However, values as low as 1.6 are also accepted by the research community while estimating coefficients of attributes. While considering the part-worths, even lower values are accepted (Hunt 2001). The coefficient estimates given in Table 2 reflect the relative effects of attributes on the choice decision based on their levels present in the alternative. The signs of the coefficients are as expected. The pseudo-R2 value 0.243 indicates that the model fit is good. The t-statistics also show that most of the attributes’ levels considered in the study are statistically significant.

It can be inferred from the analysis that trip makers are very sensitive to the comfort level. For the attribute comfort, the utility decreases rapidly at a rate of 9.5 paise/km travel when the comfort level changes from seating to standing comfortably than when it changes from standing comfortably to standing in crowd where the utility decreases at a rate of 4.6 paise/km travel. Attribute headway appears to have almost linear variation with in its part worth values. However, the utility change due to change in the headway level from 30 min to 45 min is 1.5 times that of the change due to change in the headway level from 45 min to 60 min Similarly for the attribute Travel time, utility decreases rapidly when the travel time level changes from 10% reduction to 5% reduction rather than when it changes from 15% reduction to
10% reduction. The reductions in utilities are 38 paise/min and 22 paise/min respectively for reductions in travel time from 10% to 5% and 15% to 10%. This shows that people are more sensitive to the travel time change between 10% reduction and 5% reduction.


In the present paper, the application of Conjoint analysis for transportation services in rural regions has been investigated. The stated choice responses obtained from users of rural transportation system are found to be consistent and encouraging for carrying out conjoint analsysis.

Effects coding is also explored to capture the nonlinearity associated with levels of attributes. For the study route the perceived values associated with comfort, headway and travel time are found to vary non-linearly with change in attribute levels. The reduction in trip maker’s utility for a change in comfort level from seating to standing comfortably is more than change in comfort level from standing comfortably to crowded. A reduction in travel time from 5% to 10% results into increase in trip maker’s utility, which is more than that caused by a reduction in travel time from 10% to 15%.

In the present work, socioeconomic attributes are collected from respondents in order to understand the domain of the database used. However, the effect of socioeconomic parameters on perceived values associated with various attributes of travel could not be studied due to low sample size. Further works may be carried out with an enhanced database to study the effect of socioeconomic attributes on perceived values associated with travel attributes.

The methodology explained in the present work can also be used for studying the end users’ perceptions to other modes of transport or roadway facilities.


The work presented in this article is based on a research project sponsored by the Department of Science and Technology (DST), Ministry of Science and Technology, Government of India.


1. Adamowicz, W., Louviere, J., and Swait .J. 1998. Introduction to Attribute-Based Stated Choice Methods, Final Report, Submitted to National Oceanic and Atmospheric Administration, US Department of Commerce.

2. Bates, J. 1982. Stated preference technique for the analysis of transportation behavior. In Proceedings of World Conference of Transportation Research, Hamburg, W. Germany, pp. 252–265.

3. Ben-Akiva, M., and Lerman, S. R. 1985. Discrete Choice Analysis: Theory and Applications to Travel Demand,
MIT Press, Cambridge.

4. Hensher, D.A. 1994. Stated preference analysis of travel choices: the state of practice. Transportation, 21(2), pp. 107-133.

5. Hunt, J. D. 2001. A stated preference analysis of sensitivities to elements of transportationand urban form. Paper presented at the 80th Annual Meeting of the Transportation Research Board.

6. Jose Holguin-Veras, P. E. 2002. Revealed preference analysis of commercial vehicle choice process. Journal of Transportation Engineering 128 (4): 236–346.

7. Kroes, E. P., and R. J. Sheldon. 1988. Stated preference methods: An introduction. Journal of Transport Economics and Policy 22: 11–25.

8. Louviere, J. J. 1988. Conjoint analysis modeling of stated preferences: A review of theory, methods, recent developments and external validity. Journal of Transport Economics and Policy 22(1): 93–119.

9. Louviere, J. J., Hensher, D. A., and Joffre, D. S. 2000. Stated Choice Methods. Analysis and Applications. Cambridge University Press.

10. Lai, K., and Wong, W. 2000. SP Approach toward Driver Comprehension of Message Formats on VMS, Journal of Transportation Engineering, 126(3), pp. 221-227.

11. LIMDEP 8.0 Reference Guide. 2002. Econometrics Software Inc.

12. McFadden, D. 1974. Conditional Logit Analysis of Qualitative Choice Behavior. In P. Zarembka (ed.), Frontiers in econometrics, New York: Academic Press, 105-142.

13. Opaluch, J.J., Swallow, S., Weaver, T., Wessels, C., and Wichlens, D. 1993. Evaluating impacts from noxious waste facilities: including public preferences in current siting mechanisms. Journal of Environmental Economics and Management. 24:41-59.

14. Praveen, K. S., and Rao, K. V. K. 2002. Estimation of Passenger Demand on a Proposed Passenger Water Transport System. Proceedings of Transportation Planning and Methodologies for Developing Countries, IITBombay, Mumbai.

15. Thurstone, L. L. 1927. A Law of Comparative Judgment. Psychological Review. 4. 273-286.

16. Transportation Economics & Management Systems (TEMS) Inc. 2001. Cleveland-Columbus-Cincinnati High Speed Rail Study. Final Report. Prepared for Ohio Rail Development Commission.