Including social influence in choice models for electric vehicle purchase preferences: comparison of different model formulations
Bibliographic record
Abstract
People influence each other in everyday life, affecting thoughts, intentions and behaviours. Social influence can occur in human interactions, either face-to-face contacts (direct influence) or in cognitive processes which are unconsciously stimulated by listening and watching other people or other sources of information (indirect influence) (Forgas and Williams 2001). For this reason, social influence and social interactions are becoming increasingly important in the study of transport demand and travel behaviour. Classic discrete choice models assume that the rational individual maximises his utility by making an independent choice, which takes into account his budget constraints. Although these models are commonly used in transportation research, it appears reductive to investigate traveller’s behaviours without considering the inherent social network influence (e.g. family and friends) when looking for a better understanding of the dynamics behind the daily choices (Cornes and Sandler 1996; Brock and Durlauf 2001; Paez et al. 2008; Walker 2011; He et al. 2014). However, the inclusion of social influence is very complex. It is linked to several psychological aspects so that it is not yet clear how to model it precisely and this may lead to a specification of transport models that does not reproduce the investigated phenomenon in an appropriate manner. Every analysed context could require different solutions. In this work, we have estimated model specifications for electric vehicles purchase preferences; these specifications include descriptors of social influence such as the number of individuals in the social network with pro-environmental attitudes. These attitudes can be contagious among peers and can contribute to explore heterogeneity in electric vehicle purchase preferences. Just few studies in the literature of transportation have presented a utility function with the inclusion of social influence. In particular, following Brock and Durlauf (2001), who originally proposed a model where the utility of an individual of a determined social group is directly related to what people of that group choose, some research have included social influence as an exogenous variable. This variable usually takes into account other people’s choices on the decision making process of an individual (Walker 2011; Paez et al. 2008; Kim et al. 2014). Another technique to treat social influence has been recently developed by Kamargianni et al. (2014) who have incorporated the social influence in a hybrid choice model as an component of latent variable. This component focuses on the unobserved effect of the social environment, i.e. the household, and they have specifically modelled the influence on children generated by parents’ attitudes toward walking behaviours. The data used in this study was collected between 2010 and 2011 in a workplace of 500 employees. 57 of them had previously participated to a study called 'Battery Electric Vehicles (BEV) project'. Later, 191 participated to a screening survey with four main parts: transportation patterns, information on the previous experience with BEVs, relations with co-workers and demographic information. Then, among 124 selected, 105 employees completed a semi-structured interview in order to create a heterogenic sample in terms of socio-economic characteristics, attitudes towards green technologies and lifestyle. The measurement of attitudes required a 5-point Likert scale. The survey included a state preference design with nine different exercises choosing between conventional vehicles (CV) and electric vehicles (EV) (Axsen et al., 2013). With this sample, Axsen et al. (2013) deeply analysed the interpersonal influence from a qualitative perspective but the data can also be used to estimate more complex discrete choice models to include the influence in a workplace social network. In this work, we explore various hybrid choice model structures that incorporate social influence. Initially, we investigate the attitudes of the residents using a factor analysis of the attitudinal items from the survey. Among the individual latent constructs, we identify an 'environmentally friendly and physically active' nature, which is included in the hybrid choice models. Next, with a cluster analysis of the attitudinal items and the relationship matrix among co-workers, we identify the environmentally friendly and physically active contacts (we refer to these as the 'green and active' contacts) in each individual’s social network. In order to investigate whether individuals with that specific attitude can influence peers in their social network, we explore different hybrid choice model specifications. The first formulation includes in the choice model component an exogenous variable accounting for the number of 'green and active' contacts of each individual. A second formulation accounts for this variable as part of the structural equation for the latent attitude. A third and more challenging formulation takes into account the 'green and active' contacts as an additional indicator of the measurement model. Finally, we compare the results of the three models. This study aims to develop insights on how best to quantify social influences within travel demand models.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.019 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.009 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".