Travel intention: Relative value of transport alternatives

Theory of Planned Behavior

Building on the Theory of Reasoned Action (Fishbein & Ajzen, 1975), the Theory of Planned Behavior (Ajzen, 1991) proposes that behavioral intentions are the immediate antecedents to behavior. Like the Theory of Reasoned Action (Fishbein & Ajzen, 1975), the Theory of Planned Behavior (Ajzen, 1991) proposes that behavioral intentions are a function of the person’s attitude toward performing the behavior and the subjective norm about performing the behavior.

Attitude

According to Fishbein and Ajzen (1975), attitudes toward behaviors are made up of beliefs regarding positively or negatively valued outcomes or attributes of the behavior. This has been called the value expectancy model of attitudes, as it is suggested that the attitude towards an object (e.g., travelling by public transport) is a result of the valued attributes of the object (e.g., comfort) and the likelihood that the object will provide the expected value of the attributes.

Subjective norm: Prescriptive behavioral norm

According to Fishbein and Ajzen (1975), subjective norms about behaviors are made up of beliefs regarding the extent to which important others would approve or disapprove of the person performing the behavior. Beliefs about what others think you should do can be called prescriptive or injunctive norms. Other researchers have suggested extending the concept of norms to also include mere descriptive norms, i.e., beliefs about what others do, in addition to beliefs about what others think you should do (e.g., Rivis & Sheeran, 2003). Fishbein and Ajzen (2010) suggest measuring both injunctive and descriptive norms, but also say it is not yet clear how descriptive norms should best be incorporated into their model of behavioral prediction. We decided to concentrate on the established model in the Theory of Planned Behavior and the objective of comparing the predictive strength of relative and absolute measures. Thus, we adopted the original definition of subjective norms in this study.

Perceived behavioral control: Controllability and self-efficacy

The Theory of Planned Behavior differs from the Theory of Reasoned Action in that it adds perceived behavioral control as an antecedent to behavioral intention (Ajzen, 1991). Both actual control and perception of control may have an impact on an outcome. Factors outside your control (i.e., actual control) may block you from carrying out your intentions (e.g., you can’t go by bus because there is no bus service due to a labor strike). Beliefs about whether or not you will be successful (i.e., perception of control) may influence your motivation to try to carry out the behavior (e.g., there is a risk there will be a bus strike tomorrow, so I will probably not be able to go by bus and should plan to take the car instead).

Ajzen (1991) originally defined perceived behavioral control as people’s perception of the ease or difficulty of performing a behavior. He argued that perceived behavioral control was similar to Bandura’s (1977) concept of self-efficacy and different from Rotters’s (1966) concept of the perceived locus of control, the latter referring to the belief that an event (e.g., a behavior) or outcome may be influenced by the person (i.e., internal factors) and/or external factors. Ajzen (2002) later argued that perceived behavioral control consists of two components, self-efficacy and controllability. According to Ajzen (2002), controllability refers to the extent to which performance of the behavior is up to the actor. Ajzen (2002) continued to argue that controllability was not the same as Rotter’s (1966) concept of the perceived locus of control. However, if externally controlled factors have a different (e.g., my spouse might take our car) impact on behavior compared to internally controlled factors (e.g., I know how to drive the car to my destination), it seems that the external locus of control may function as a barrier to controllability. We included controllability and self-efficacy as two separate factors in this study, in line with Ajzen’s (2002) conceptualization of perceived behavioral control.

Past behavior

According to Ajzen (2011), past behavior has been found to contribute independently to the prediction of intentions, over and above attitudes, subjective norms, and perceived behavioral control. Ajzen (2011) did not accept the possibility that past behavior itself may be a direct cause for behavioral intentions, but argued instead that the predictive power of past behavior on behavioral intentions represent the influence of some undiscovered or unconfirmed factors (e.g., identity and/or anticipated affect). Other researchers have argued that past behavior may contribute to conscious decisions about behavior because past behavior represents decision making in the past (Aarts, Verplanken, & van Knippenberg, 1998), because we may use past behavior as a substitute for elaborate reasoning (Bem, 1972), or because we are motivated to be consistent over time (Festinger, 1957).

Participants and procedure

We recruited 498 participants (265 men and 233 women) from a commercial panel of participants in Norway. The panel aims to be representative for the general adult population in Norway. For this survey, we only recruited participants from the four largest cities (population between 100,000 and 600,000 inhabitants) in Norway (247 from Oslo, 89 from Bergen, 78 from Trondheim, and 84 from Stavanger) because we wanted to limit the study to people living in areas with a relatively good public transport service. The participants were paid the standard rate at the time for their participation. The mean age of the participants was 46.8 years (SD = 17.7). Compared to the adult population in the four cities, our sample was slightly older (mean age 46.8 years vs. 45.2 years) and consisted of slightly more men (53% vs. 50%). The participants received a recruitment message by e-mail, linking them to the web pages of the commercial survey company where they completed the survey online. The data was collected in winter (February) in Norway, when people generally travel less by bicycle and more by public transport than during the rest of the year. All statistical relationships were investigated using regression analysis.

Apparatus

All measures were repeated for four modes of transport: car, public transport, bicycle, and walking. Behavioral intention was measured by three items, e.g., “How likely are you to travel by car next week?” on a scale from 1 (“very unlikely”) to 7 (“very likely”). The Cronbach’s alpha for Behavioral intention was between .96 (bicycle) and .97 (car). Attitude toward the behavior was measured by seven items, e.g., “Would travelling by car be a good solution to your transport needs next week?” from 1 (“very bad”) to 7 (“very good”). The Cronbach’s alpha for Attitude toward the behavior was between .76 (bicycle) and .79 (public transport). The Subjective (prescriptive behavioral) norm was measured by three items, e.g., “How do you think your friends think you should travel next week?” from 1 (“definitely not by car”) to 7 (“definitely by car”). The Cronbach’s alpha for the Subjective norm was between .90 (bicycle) and .94 (walking). Controllability was measured by three items, e.g., “How likely is it that travelling by car would be an available alternative for your transport needs next week?” from 1 (“definitely not available”) to 7 (“definitely available”). The Cronbach’s alpha for Controllability was between .75 (public transport) and .90 (car). Self-efficacy was measured by three items, e.g., “If you decided to, how confident are you that you could travel by car next week?” from 1 (“very unconfident”) to 7 (“very confident”). The Cronbach’s alpha for Self-efficacy was between .74 (bicycle) and .84 (walking). All the measures were internally consistent according to Nunnally’s (1978) threshold for Cronbach’s alpha > .70. Past behavior was measured as the number of journeys made by the specific mode of transport (e.g., car) last week. The relative measures were computed by dividing the absolute measure of the specific transport mode (e.g., car) by the mean of the absolute measures of all four transport modes, e.g., Equation 1 for the relative attitude toward travelling by car.

Theory of Planned Behavior predicting travel intentions

Past behavior, attitude, subjective norm, and controllability were significant predictors for intention to travel by car, using absolute measures. Past behavior, attitude, subjective norm, and self-efficacy were significant predictors for intention to travel by public transport, bicycle, and on foot, using absolute measures. Past behavior, attitude, subjective norm, controllability, and self-efficacy were significant predictors for intention to travel by car, using relative measures. Past behavior, attitude, subjective norm, and self-efficacy were significant predictors for intention to travel by public transport, bicycle, and on foot, using relative measures. Overall, the results support the Theory of Planned Behavior.

Since past behavior also contributed to the prediction of travel intentions, the results suggest that attitudes, subjective norm, and perceived behavioral control are not a sufficient set of factors to explain travel intentions. Past behavior predicts travel intentions, either as a causal factor or as a proxy for other causal factors related to both past behavior and behavioral intentions. This study does not provide an answer to the question regarding the specific nature of the relationship between past behavior and behavioral intentions.

The finding that the relationship between the two components of perceived behavioral control and travel intentions differed among the four transport modes supports the decision to include both controllability and self-efficacy as distinct components of perceived behavioral control. Whereas both controllability and self-efficacy predicted intentions to travel by car, only self-efficacy predicted intentions to travel by public transport, bicycle, and on foot. Public and non-motorized modes of transport may be perceived as generally available modes of transport in the city, but modes that vary in the individual effort needed to make use of them. A car may not be available at all, or its availability may be uncertain. The descriptive statistics show that controllability was lower for car than for public transport and walking, but higher for car than for cycling. The variation in controllability was larger for car than for the other modes of transport.

Relative measures to predict behavioral intentions in choice situations

By using relative rather than absolute measures the coefficient of determination increased for all four modes of transport. Relative measures predicted travel intentions better than absolute measures. For each of the predictors, the relative measure added to the variance accounted for beyond what could be explained by the absolute measure. This supports Ajzen and Fishbein’s proposition regarding prediction of behavioral intentions in choice situations.

Implications

The theoretical implications of these results are that the results strengthen the Theory of Planned Behavior and the proposition that relative measures should be used in choice situations. The results not only support the Theory of Planned Behavior, but also suggest that previous studies that did not use relative measures in choice situations may have underestimated the predictive strength of the Theory of Planned Behavior. The practical implication is that past behavior, attitude, subjective norm, controllability, and self-efficacy can be used to predict behavioral intentions. Assuming the statistical relationships represent causal relationships, attitude, subjective norm, controllability, and self-efficacy can also be targeted in interventions aiming to promote behavioral change, e.g., from car use to public or non-motorized transportation.

Limitations

Drawing any conclusions on questions regarding causality is heavily restricted by the cross-sectional nature of the survey. The sample was drawn from urban areas in Norway, and may not be representative of rural areas where public transport may be a less relevant alternative mode of transport. The sample was recruited from a commercial panel of participants and may be more representative of people with more motivation to participate in surveys in general or people motivated primarily by the economic compensation. The sample was slightly older and contained slightly more men than found in the adult population in the four cities, but not to the extent that it poses a serious threat to the validity of the results.

The data was collected in the winter, when people generally travel less by bicycle and more by public transport. As all questions were asked specifically in relation to the next week, this should not be a problem when testing the relationship between the factors in the model. The fact that people travel less by bicycle during the winter may be explained by people having less favorable attitudes towards riding a bike during the winter. However, the variance in the data may be limited due to the data being collected during only one season in the year.

Future research

Future research should test the proposition regarding relative measures in choice situations including other factors promising to predict behavioral intention (e.g., descriptive norms, identity, and anticipated affect). Future research should also test the proposition regarding relative measures in choice situations including the relationship between behavioral intentions and future behavior (i.e., longitudinal studies). Future research should use designs that are better for tests of causal relationships, e.g., intervention studies aimed at improving attitudes, norms, or perception of behavioral control. Future research should also test the hypothesized relationships in other samples, e.g., children and rural populations, to improve the generalizability of the conclusions regarding the Theory of Planned Behavior and travel mode choice combined across studies.