International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Do attitudes cause behavior or vice versa? An alternative conceptualization of the attitude-behavior relationship in choice modeling
Maarten Kroesen

Last modified: 28 March 2017

Abstract


Attitudes, broadly defined as affective evaluations (favorable or unfavorable) with regard to particular objects or behaviors (Ajzen and Fishbein, 1977), are generally thought to play an important role in determining people’s travel mode choices (Ben-Akiva et al., 2002; Gärling et al., 1998). Initial studies investigating the attitude-behavior relationship in a transportation context date back to the late 1970s (Dobson et al., 1978; Tardiff, 1977; Tischer and Phillips, 1979). Since then, more elaborate theoretical frameworks to study the effects of attitudes on behavior have been developed, the most prominent and influential being the Theory of Planned Behavior (Ajzen, 1991). This model has also been extensively applied in the transport domain (Bamberg, 2006; Bamberg et al., 2003; de Groot and Steg, 2007; Heath and Gifford, 2002; Thøgersen, 2006). Over the past decade, it has also become popular to include attitudes in discrete choice models, leading to a class of so-called hybrid choice models (Ben-Akiva et al., 2002).

While conceptual and related empirical models generally assume that attitudes influence behavior, researchers typically acknowledge that a reverse relationship, i.e. from behavior to attitudes, may also exist (Ajzen, 2015). Clearly, the question of “which causes which” has important implications for behavioral research, (choice) model development, and policy design (Chorus and Kroesen, 2014). Concerning the latter: if attitudes (primarily) influence behavior, it makes sense to try to influence behavior (in desired directions) via promotional and information campaigns targeted at people’s attitudes. If, on the other hand, behavior (primarily) influences attitudes, policy should rather focus on changing people’s behavior directly, e.g. via regulations, or focus on the ‘hard’ determinants of behavior, like travel costs, e.g. via pricing policies. Given that knowledge about the direction of causation is of crucial importance from a research and policy perspective, the scarcity of research addressing this subject is surprising and problematic.

Our aim is to reinvigorate interest in this topic by (1) assessing the direction of causation between attitudes and behavior using panel data and (2) by presenting a new framework to study attitude-behavior (in)consistency over time. This latter framework identifies the processes through which consistency and inconsistency occur over time and thereby actually transcends the original question of whether attitudes cause behavior or vice versa.

The first aim is achieved by specifying a two-wave cross-lagged panel model (CLPM) (Finkel, 1995). For the second aim a two-wave latent transition model (LTM) is specified (Collins and Lanza, 2013). To estimate the models a mobility survey is developed, which is administered twice (in 2013 and 2014) among a (random) subset of the Longitudinal Internet Studies for the Social (LISS) sciences panel members. This panel is based on a true probability sample of Dutch households. The present analyses are based on those respondents who completed both surveys (N=1,376).

To assess people’s travel behavior, the following open question was formulated: ‘how many kilometers do you travel (approximately) in a regular week, using the following modes of transport?’ Five modes were measured, but for the present analysis only the three most relevant modes in terms of distance travelled are considered: car as driver, bicycle and public transport (including bus, tram, metro or train). Given the skewness in the distributions, the variables were recoded into 5-point ordinal scales.  To measure respondents’ attitudes towards the three modes, we followed Fishbein and Ajzen’s (1977) recommendation to use the attitude towards the behavior, since this is the best a predictor of behavior. To increase the reliability of the attitude measurements multi-item composite scales were used. Again, these scales were recoded to 5-point ordinal variables. Finally, for the LTMs, measures of the degrees of dissonance were needed. For each mode, these measures were obtained by computing the absolute difference between the 5-point behavioral and 5-point attitude scale.

For each mode, a separate CLPM was estimated using Mplus 7.2 (Muthén and Muthén, 2005). Since the behavioral variables represent ordinal outcomes, they were modelled using ordered probit regressions. Similar to the CLPMs, three mode-specific LTMs were estimated. Latent Gold was used to this end (Vermunt and Magidson, 2013). In these models, ordinal logit models were used to estimate the relationships from the latent class variables to the indicators (at each point in time). In addition, since the latent class variables themselves are nominal, the over-time relationship between the latent class variables was captured by a multinomial logit model. In both sets of models a set of five exogenous control variables were included (gender, age, primary occupation, education level and personal net income).

The results of the CLPMs indicate that, for each mode, the use of the mode and the attitude towards using the mode mutually influence each other over time. Contrary to conventional wisdom, however, the effects of behaviors (mode use) on attitudes are much larger than vice versa. The results of the LTMs indicate that people who have consonant (i.e., aligned) attitude-behavior patterns are more stable than those who have dissonant patterns. In addition, in line with the results of the CLPMs, dissonant travelers are more likely to adjust their attitudes than their behavior. Finally, we find relatively large classes of dissonant individuals with favorable bicycle and public transport attitudes, but low actual use of these modes.

Overall, the results have important implications for choice modelers who ascribe an important role to attitudes in the explanation of (variation in) behavior (see for example studies cited in Chorus and Kroesen (2014)). If empirical models allow for effects of attitudes on behavior, and not (also) the other way around, our results suggests that it is likely that such models (strongly) overestimate the effects of attitudes. Our recommendation to researchers studying the role of attitudes, and in particular to those interested in modeling attitudes in the form of latent variables in so-called Hybrid Choice Models, is to carefully consider the bi-directional relation between attitudes and choices – even if this requires the use of more involved econometric techniques for model identification.

 

References

 

Ajzen, I., Fishbein, M. (1977) Attitude-behavior relations: A theoretical analysis and review of empirical research. Psychological bulletin 84, 888.

Ajzen, I. (2015) The theory of planned behaviour is alive and well, and not ready to retire: a commentary on Sniehotta, Presseau, and Araujo-Soares. Health Psychol Rev 9, 131-137.

Bamberg, S. (2006) Is a residential relocation a good opportunity to change people's travel behavior? Results from a theory-driven intervention study. Environ Behav 38, 820-840.

Bamberg, S., Ajzen, I., Schmidt, P. (2003) Choice of travel mode in the theory of planned behavior: The roles of past behavior, habit, and reasoned action. Basic Appl Soc Psych 25, 175-187.

Ben-Akiva, M., Mcfadden, D., Train, K., Walker, J., Bhat, C., Bierlaire, M., Bolduc, D., Boersch-Supan, A., Brownstone, D., Bunch, D.S., Daly, A., De Palma, A., Gopinath, D., Karlstrom, A., Munizaga, M.A. (2002) Hybrid choice models: Progress and challenges. Market Lett 13, 163-175.

Chorus, C.G., Kroesen, M. (2014) On the (im-)possibility of deriving transport policy implications from hybrid choice models. Trans Pol 36, 217-222.

Collins, L.M., Lanza, S.T. (2013) Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. John Wiley & Sons.

De Groot, J., Steg, L. (2007) General beliefs and the theory of planned behavior: The role of environmental concerns in the TPB. J Appl Soc Psychol 37, 1817-1836.

Dobson, R., Dunbar, F., Smith, C.J., Reibstein, D., Lovelock, C. (1978) Structural models for the analysis of traveler attitude-behavior relationships. Transportation 7, 351-363.

Finkel, S.E. (1995) Causal analysis with panel data. Sage.

Fishbein, M., Ajzen, I. (1977) Belief, attitude, intention, and behavior: An introduction to theory and research.

Gärling, T., Gillholm, R., Gärling, A. (1998) Reintroducing attitude theory in travel behavior research: The validity of an interactive interview procedure to predict car use. Transportation 25, 129-146.

Heath, Y., Gifford, R. (2002) Extending the theory of planned behavior: Predicting the use of public transportation. J Appl Soc Psychol 32, 2154-2189.

Muthén, L.K., Muthén, B.O. (2005) Mplus: Statistical analysis with latent variables: User's guide. Muthén & Muthén Los Angeles.

Tardiff, T.J. (1977) Causal inferences involving transportation attitudes and behavior. Transport Res 11, 397-404.

Thøgersen, J. (2006) Understanding repetitive travel mode choices in a stable context: A panel study approach. Transport Res a-Pol 40, 621-638.

Tischer, M.L., Phillips, R.V. (1979) The relationship between transportation perceptions and behavior over time. Transportation 8, 21-36.

Vermunt, J.K., Magidson, J. (2013) Technical guide for Latent GOLD 5.0: Basic, advanced, and syntax. Belmont, MA: Statistical Innovations Inc.

 


Conference registration is required in order to view papers.