Knowledge about Alternatives in Stated Response Surveys 21
8th International Conference on Survey Methods in Transport
Annecy, France, May 25-31, 2008
Level of Knowledge and Awareness about Choice Alternatives – A Missing Link of Stated Response Surveys? - A Hypothesis
Gerd Sammer, Institute for Transport Studies, University of Natural Resources and Applied Life Sciences, Vienna, Austria
Christian Gruber, Sammer & Partner Transport Consultancy Ltd., Graz, Austria
Reinhard Hoessinger, Institute for Transport Studies, University of Natural Resources and Applied Life Sciences, Vienna, Austria
Gerald Roeschel, Sammer & Partner Transport Consultancy Ltd., Graz, Austria
abstract
This paper is about information provided unintentionally or deliberately in stated response surveys (SP-surveys). Following from previous research, information is assumed to increase the level of knowledge and awareness of the respondents and to bear a strong influence on the result of such surveys. In most cases of SR-surveys it is however ignored due to the lack of a clear procedure how to be integrated into the survey. After a literature review indicating the current state of research, a methodological concept is presented how to integrate the information provided in SR-surveys in a systematic manner, and how to measure information respectively the level of knowledge and awareness in order to be used for SR-data analysis. Finally, two case studies provide evidence for the assumed importance of information provided in SR-surveys. The first case shows that the existing level of knowledge and awareness of the respondents as well as the information provided about the alternatives of the SR-games strongly affect the choices made by the respondents. The parameters of the utility function of the SR-choice model are estimated more valid, if the level of knowledge and awareness feeds into the model. The second case yields two results. First, the deliberate provision of background information over the attributes of alternatives can strongly affect the SR-choices. It identifies SR-surveys to be an appropriate tool to test and model the behavioural effectiveness of information considering the relevant market penetration. Second, the inevitable description of the alternatives by means of their attributes increases the respondents' level of knowledge and awareness about the attributes under study, enhancing their effectiveness compared to revealed preference surveys. Therefore the lacking consideration of the level of knowledge and awareness can be interpreted as missing link of SR-surveys. This paper touches on the intrinsic problem of information provided unintentionally in SR-surveys and the chances to use it deliberately. But it remains an avenue of further research.
Problem definition
Nowadays, stated response surveys (SR-surveys) are used frequently in many countries to explore influencing variables and their behavioural parameters. Analysis indicates that the quality of the SR-surveys and their results are often unsatisfactory. This can be shown for example by comparing resulting indicators such as the value of time, or the explanatory quality of the derived models: SR-surveys indicate a tremendous range for the value of time which cannot be explained easily. Moreover behavioural variables frequently explain considerably less than 50 percent of the reaction, an unsatisfactory result from the modeller’s point of view. In revealed preference surveys the observed behaviour of the respondents is almost unbiased by the survey itself; the bias depends on the survey methodology. In stated response surveys the respondent is confronted with a hypothetical new situation concerning his choice. This means that he receives more or less additional information, knowledge and awareness about his alternatives which he would normally not have for his everyday choices. This additional information can raise his overall level of knowledge and his awareness of the relevant characteristics addressed in the stated response experiment. This means that the following questions need to be addressed: Firstly, is the respondent’s choice influenced or biased compared to a revealed preference survey and if yes, how strongly; and secondly, which consequences must this have for the stated response survey to avoid biased results.
GOAL OF THE PAPER
The paper in hand has the goal to dicuss the following problems and questions of stated response surveys (SR-surveys):
· To what extend does the provided information about the attributes of the alternatives influence the respondent’s level of knowledge and awareness unintentionally and inevitably in SR-surveys?
· To what extend does the level of knowledge and awareness of the respondent influence and explain the choice of alternatives in SR-surveys?
· Are SR-surveys an appropriate tool in order to estimate the influence of deliberately provided information?
· How can we measure the level of knowledge and awareness?
When addressing these questions the following issues require particular attention:
· Information provided to the respondent in SP-surveys increases the level of knowledge, if it is new to him.
· Information provided to the respondent in SP-surveys can change the respondent’s awareness even if he already knows it.
· The procedure of SR-surveys forces the user to provide information for the respondent about the attributes of the choice alternatives, which he has not available absolutely in real life choices.
· The procedure of SR-surveys gives the opportunity to the user, to provide additional information for the respondent deliberately, which he doesn’t need absolutely for the choice experiment.
· There is no difference from the respondent point of view, whether the information is provided unintentionally or deliberately in the SR-survey. This distinction is only relevant for the user of the SR-survey.
These questions lead to the following hypotheses: The information provided in SR-surveys has a strong influence on both, the respondent’s level of knowledge and awareness. Measuring the level of knowledge and awareness in SR-surveys and taking it into account in SR-analysis increases the quality of results. Under the term “SR-survey” the following survey techniques of the literature are subsumed in this paper: stated preference, stated adaptation, stated tolerance, stated prospect, stated choice, choice based conjoint (Lee-Gosselin 2003 and 1997, Axhausen and Sammer 2001).
Literature Review
As far as the consideration of the level of knowledge and awareness is concerned, an analysis of international literature leads to the following results:
- One of the most recent papers about SP-surveys (Jones und Bradley 2006, p. 348) provides a thorough analysis of the current state of knowledge and its use. It identifies 7 elements which are critical for the quality of SR-surveys about alternative choices: design of the experiment, customization, scenarios, survey administration, type of judgements by the respondents, survey presentation, and analysis. When making their recommendations the authors point out that the provision of background information via the attributes of alternative choices can have an impact upon the attitude of the respondent and thus upon the choice result (Podgorski et al. 2004). As an example some background information about a different pricing policy for the use of various modes of transport is mentioned. The authors state that little is known about the impact of such information provided as part of the alternative choices presented to the respondent; no solutions are suggested.
- Several publications address the importance of information and the awareness of alternative choices in general. In this context, one paper needs to be mentioned which analyses and interprets the available literature (Chorus et.al. 2006, 2007). It states that information and the awareness of alternative choices (mode, route, destination) have an impact upon the decision-making process of the transport user. The reliability of comments regarding the travel time is particularly addressed; additionally the authors point out that many transport users are only aware of those transport options which they themselves use regularly. A number of recent and not so recent publications are mentioned which address this problem (e.g. Ben-Akiva 1998, Bonsall 2004, Koppelmann and Pas 1980, Ramming 2002, Recker and Golob 1976, Molin and Chorus 2004). This analysis of relevant literature is done in connection with the impact of traveller information services upon the travel behaviour. The issue of stated response surveys and analyses are not addressed.
- But this problem is specifically dealt with in a workshop report about quality assurance for SR-surveys (Sammer 2003): „Specific attention regarding the definition of the choice situation in the interview is to be paid to the information and awareness level of the target person. This applies on the one hand to content and intensity of the information provided to the respondent for describing the hypothetical situation of choice. To find out the influence of the level of information on the choice, in certain cases, it may be useful to change the information status when the number of "choice games'' increases. On the other hand, the result is influenced by the level of awareness of the target person of the individual consequences of the choice (experience has, for example shown, that the hypothetical choice is influenced if the respondent is made aware of the long-term monetary consequences of his/her decision for a means of transport). It is recommended that both the information level and that of awareness of the respondents should be collected as contextual data during the interview“. This publication specifically recommends collecting the levels of knowledge and awareness as contextual data, but it does not offer precise solutions.
- Empirical results of an SR-analysis (stated adaptation) can be found in a study (Sammer et al. 1994) about behavioural reactions of transport users to a scenario based on the assumption that cars which emit CO2 are prohibited from entering the city centre. For this scenario the SR-survey used two different assumptions about the provision of information to analyse the corresponding stated preference behaviour. The scenario under consideration is based on the following assumptions: Only vehicles without CO2-emission are allowed into the city centre, i.e. vehicles without combustion engines running on conventional fuels. Residents driving to and from their homes are excepted. In two sub-scenarios the possibility of using an electric car is indicated by providing limited or detailed information. In the first sub-scenario, only limited information about electric cars as a possible alternative was provided, while in the second sub-scenario detailed information about electric vehicles (e.g. purchase price, for which distance the battery capacity would last, etc.) was given. The changes in travel behaviour resulting from the different scenarios were determined with the aid of a stated adaptation method based on in-depth interactive interviews. If a person's trip was affected by the scenarios, that person had the following options:
· Abandon the trip if it can be avoided. This is only possible for trips without a subjective necessity or if the chore can be taken over by someone else. For example, a commuter trip cannot be relinquished;
· Transfer the trip to another mode of transport or type of vehicle, for example to public transport, bicycle or electric car, etc. This reaction of the respondent is only possible if all activities outside his/her home are still possible in spite of the switch to different means of transport;
· Park the car outside the limited zone and change to another means of transport for the rest of the trip (on foot, public transport for example);
· Choice of another destination outside the limited area; this is only possible if there is a destination of equal value outside the limited area.
The findings indicate that the information provided about the options influences the result of SR-surveys and their analysis significantly (fig. 1).
Figure 1: Comparison of the behavioural reaction for the scenario zero CO2-emission if limited or extensive information about electric cars is provided to the respondent (Sammer et.al. 1994).
The results of the analysis of the relevant literature show that the fundamental problem caused by the level of information and awareness within the framework of an SR-survey is known and frequently discussed. But this does not have any consequences regarding the realization of undistorted SR-surveys and the problem is insufficiently considered when using such techniques or interpreting the results of an SR-analysis.
Methodological Considerations
The basic concept of including the quality of information and knowledge about mode attributes in SR-surveys and RP-surveys can be traced back to the development of the so-called situational approach (Brög, 1981, Brög and Erl 1981 and 1996) in the eighties. The situational approach proceeds on the assumption that the individuals are given a certain scope for action (“objective” situations) by their environment. This scope for action is determined by (figure 2)
- the spatial land-use-structure in which the travel behaviour is embedded;
- the available transport infrastructure;
- the constraints and freedom derived from the social demography of the individual and the other members of his household;
- the social values, standards, and attitudes regarding the areas relevant to travel behaviour.
Each individual experiences these “objective” situations in a specific way. Individually different “subjective” situations (consisting of variables from all four areas mentioned) are created. These “subjective” situations differ from the “objective” ones by incomplete, consciously or unconsciously distorted perceptions and knowledge of the real world (alternative of choice). The degree of deviation depends on the individual person and his/her specific experience. It has to be mentioned that the RP-survey can cause a distortion of the original level of knowledge and awareness of the respondent. In these subjective situations, individual decisions are taken. In order to create a valid model of travel behaviour, it is necessary to reconstruct the chain of “objective situation – individual perception – subjective situation – individual choice – realized travel behaviour”. If one wishes to influence travel behaviour, it is possible to intervene at any point in this chain to accomplish this. If someone wants to explain the individual travel behaviour by using all relevant influences, it is necessary to collect the respondent’s level of knowledge and awareness for all relevant and available alternatives for his choice which influence his preferences. Chorus et al. (2006) have developed a framework on pre-decisional information acquisition with fundamental similarities to the situational approach (fig.2) which consists of four steps: objective characteristics, perceptions (of the objective characteristics), trade-off (costs and benefits) and the behavioural decision.