Content uploaded by Akshay Vij
Author content
All content in this area was uploaded by Akshay Vij on Jun 27, 2016
Content may be subject to copyright.
A preview of the PDF is not available
Hybrid choice models: The identification problem
Figures
Figures - uploaded by Akshay Vij
Author content
All figure content in this area was uploaded by Akshay Vij
Content may be subject to copyright.
... 96-105). A well-known example are Hybrid Choice Models (HCM), which have gained popularity in recent years (see Vij and Walker, 2014). 18 By using the same sociodemographic characteristics to explain both the LV and the utility parameters, it is possible to test for their effects beyond the LV. ...
... With advances in computational power and data availability, researchers tend to estimate more complex models to get new insights into consumer behavior. However, while implementing a structural equation model (SEM) component instead of building the LV beforehand could have led to increases in accuracy in one part of our model, implementing a SEM would also have required more data or less accurate choice model components, as otherwise complexity is likely to become a serious issue (see, e.g., Bouscasse, 2018;Vij and Walker, 2014). Making this trade-off is up to researchers (see also Dekker et al., 2016). ...
... The questionnaire also included several attitudinal statements on frequency of use of video calls before, during and after (anticipated behaviour) the pandemic; safety concerns when travelling (including attitude towards NPIs); and the 'Big Five' personality traits (Rammstedt and Joh,n 2007). Given the nature of available data, we employed the Hybrid Choice Modelling (HCM) approach (Walker, 2001, Vij andWalker, 2014) to model the choices and elicitation of preferences while incorporating psychometric and other unobservable measures alongside directly measured attributes, such as travel cost and duration. ...
... The different variables affecting the decision-making process of the individual were included in a hybrid choice model (HCM). The HCM is an integrated discrete choice and latent variable model that enables us to account for psychometric and other unobservable measures within a mixed model formulation (Vij and Walker, 2014) by incorporating structural relationships between observable and latent variables and correcting for measurement errors to reduce the variance of the estimates (Vij and Walker, 2016). As with simpler discrete choice models, the HCM is based on the random utility maximisation (RUM) framework (Domencich andMcFadden, 1975, McFadden, 1981), however, its formulation has three different components (Walker, 2001, Ben-Akiva et al., 2002. ...
The COVID-19 pandemic and the consequent travel restrictions have had an unprecedented impact on the air travel market. However, a rigorous analysis of the potential role of safety perceptions and attitudes towards COVID-19 interventions on future air passenger choices has been lacking to date. To investigate this matter, 1469 individuals were interviewed between April and September 2020 in four multi-airport cities (London, New York City, Sao Paulo, Shanghai). The core analysis draws upon data from a set of stated preference (SP) experiments in which respondents were asked to reflect on a hypothetical air travel journey taking place when travel restrictions are lifted but there is still a risk of infection. The hybrid choice model results show that alongside traditional attributes, such as fare, duration and transfer, attitudinal and safety perception factors matter to air passengers when making future air travel choices. The cross-national analysis points towards differences in responses across the cities to stem from culturally-driven attitudes towards interpersonal distance and personal space. We also report the willingness to pay for travel attributes under the expected future conditions and discuss post-pandemic implications for the air travel sector, including video-conferencing as a substitute for air travel.
... In this structural model, are the parameters to be estimated and inr are normally distributed disturbances with mean zero and variance-covariance matrix Ψ. A common practice in HCMs is to use a unitary fixed value at Ψ to ensure the identification of the parameters; nevertheless, extensive discussions of other options for identification are offered by Raveau et al. (2012), Daly et al. (2012), Vij and Walker (2014), and among others. ...
For many years, the economic literature has recognized the role of attitudes, beliefs, and perceptions in estimating the value of a statistical life (VSL). However, few applications have attempted to include them. This article incorporates the perceived controllability and concern about traffic and cardiorespiratory risks to estimate VSL using a hybrid choice model (HCM). The HCM allows us to include unobserved heterogeneity and improve behavioral realism explicitly. Using data from a choice experiment conducted in Santiago, Chile, we estimate a VSL of US$3.78 million for traffic risks and US$2.06 million for cardiorespiratory risks. We found that higher controllability decreases the likelihood that the respondents would be willing to pay for risk reductions in both risks. On the other hand, concern about these risks decreases the willingness to pay for traffic risk reductions but increases it for cardiorespiratory risk reductions.
... Identification issues in ICLV models is another aspect that needs particular attention. Studies by Abou-Zeid and Ben-Akiva (2011); Vij and Walker (2014) provide guidelines to ensure theoretical identifiability for ICLV models. In addition, variability issues in the SEM part and issues related to sample size and sampling technique (Ma et al. 2015) must be addressed to ensure empirical identification. ...
The paper examines the influences of attitude and perceptions on mode choice decisions while accounting for household interactions. A nested ICLVmodel is estimated, incorporating six types of tour typology at the upper level, and the lower level considers four modes (active, PT, MTW, and car) under each tour type. The formulation of tour typologies, consist of a combination of activity types (maintenance, discretionary, and a mix of both these activities) and ‘with-whom’ travel arrangements (household members and non-household members), is an important contribution of the study. Three latent constructs—green lifestyle, perceived residential neighborhood, and safety consciousness—extracted from 10 measurement items are integrated into the nested framework to account for the taste heterogeneity arising from variations in the perceptions and attitudes of decision-makers. This model is estimated at the household level by considering the activity-travel diary data of Bhubaneswar city, comprising 858 households, 1454 adult individuals, and 2214 joint tours. The tour typologies reveal that intra-household interactions are more pronounced than inter-household interactions involving non-family members. The results from the structural model of the ICLV framework indicate that household socio-demographics and built environment variables significantly influence the attitudes and perceptions of adult household members. With the inclusion of latent factors in the nested ICLV model estimation, a substantial part of the intrinsic taste (or distaste) variability was captured that was previously ascribed to alternative specific constants. Additionally, the study contributed to the literature by deriving distinct values of travel time savings for PT (₹19.978/hr.), MTW (₹30.687/hr.), and car (₹39.635/hr.) for different tour typologies. A sensitivity analysis of selected control variables is carried out, and policy interventions from a broad viewpoint are also discussed.
... The issue is rather complicated and it is even more so because the identification conditions may not be the same across many latent class model configurations mentioned throughout the paper. Plus, the current transportation literature lacks discussions about the issue, aside from the ICLV model specifically (Vij and Walker, 2014a); hence, it is beyond our scope. Although more research is needed to clarify the identification conditions for each latent class model configuration, for interested readers, it is instructive to refer to sources such as Vermunt and Magidson (2004); Huang and Bandeen-Roche (2004); and Gu and Xu (2020). ...
... This study estimated the parameters by maximum simulated likelihood, using 500 draws of Halton sampling type for the generation of draws for random distributions (Halton, 1960;Hess and Palma, 2019a). To deal empirically with the identification issues of the model, this study followed the previous literature (Soto et al., 2018;Vij and Walker, 2014): First, this study attempted to verify that the estimated parameters lie inside the range of reasonable values; Also, the model was estimated several times, trying to employ different starting values for the parameters, confirming that the same solution can be reached in all cases thereby supporting the belief that the solution was a global maximum. ...
Autonomous vehicles (AVs) are expected to have a tremendous impact on travel behavior of people with disabilities because driving skills will be unnecessary. However, there has been little empirical research on their travel behavior, perceptions, and needs regarding AVs. This study explores their potential mode choice when assuming the autonomous vehicle transportation services (AVTS) are available in the future. The main objective is to understand what factors affect their mode-choice decision when considering AVTS compared to conventional modes of transportation, including bus, paratransit, and personal vehicles. To explain not only the impacts of the observed variables such as individual characteristics and mode attributes but also the impacts of their attitudes and perceptions of current mobility issues and AV technology on their mode choice, this study employs a hybrid choice model. This study focuses on individuals with physical disabilities and individuals with visual impairments. The model results show that individuals with disabilities who have a negative attitude toward the current public transit services and neighborhood built environments were more likely to choose AVTS. Furthermore, the presence of an onboard human attendant increased the probability of AVTS being chosen. The findings corroborate that more targeted strategies should be prepared to make AVTS a viable mode of transportation among people with disabilities. For example, it is worth noting that human attendants might be needed to relieve the anxiety of people with disabilities over traveling without anyone who can help them use AVTS. Based on the findings, this paper suggests some policy recommendations for providing viable AVTS for people with disabilities.
... 12 As methods to account for both explained and unexplained preference heterogeneity 13 became more frequently used, fundamental issues were identified, including computational challenges (eg, premature convergence) [14][15][16] and behavioral aspects (eg, accounting for scale heterogeneity along with other forms of preference heterogeneity, choice certainty). [17][18][19][20] Various approaches have been put forward to address these issues, although there is limited guidance on their relative strengths and weaknesses when accounting for preference heterogeneity in health-related DCE data. [21][22][23][24][25][26][27][28][29][30] The number and complexity of available methods to account for heterogeneity and the limited available guidance can lead to difficulties for practitioners and decision makers in making judgments about the suitability of these approaches. ...
Objectives
Discrete choice experiments (DCEs) are increasingly used to elicit preferences for health and healthcare. Although many applications assume preferences are homogenous, there is a growing portfolio of methods to understand both explained (because of observed factors) and unexplained (latent) heterogeneity. Nevertheless, the selection of analytical methods can be challenging and little guidance is available. This study aimed to determine the state of practice in accounting for preference heterogeneity in the analysis of health-related DCEs, including the views and experiences of health preference researchers and an overview of the tools that are commonly used to elicit preferences.
Methods
An online survey was developed and distributed among health preference researchers and nonhealth method experts, and a systematic review of the DCE literature in health was undertaken to explore the analytical methods used and summarize trends.
Results
Most respondents (n = 59 of 70, 84%) agreed that accounting for preference heterogeneity provides a richer understanding of the data. Nevertheless, there was disagreement on how to account for heterogeneity; most (n = 60, 85%) stated that more guidance was needed. Notably, the majority (n = 41, 58%) raised concern about the increasing complexity of analytical methods. Of the 342 studies included in the review, half (n = 175, 51%) used a mixed logit with continuous distributions for the parameters, and a third (n = 110, 32%) used a latent class model.
Conclusions
Although there is agreement about the importance of accounting for preference heterogeneity, there are noticeable disagreements and concerns about best practices, resulting in a clear need for further analytical guidance.
... I 8 is particularly interesting, as this item addresses regional product beliefs and asked respondents about their willingness to contribute monetarily to the purchase of regionally manufactured goods. This could be a first indicator that positive beliefs may not necessarily translate into Vij and Walker, 2014). 13 All parameter estimates can be found in the Supplementary Materials Section III. 14 For the second LV, regional product beliefs, we removed one item to increase reliability (cf. ...
Reflecting a broader trend towards regional products in Germany, the recently established “System for Guarantees of Regional Origin” allows operators of subsidized renewable energy plants to market their generation as regional electricity. However, it remains unclear whether and why consumers are willing to pay a premium for regional electricity generation. While a few studies have examined the willingness to pay (WTP) of households for regional electricity generation, little is known about the underlying factors driving WTP. We fill this gap with a representative survey of 838 German households that includes both a choice experiment and questions capturing individual motivations. Data is used to estimate a comprehensive Hybrid Choice Model (HCM) that allows to determine respondents' WTP, explain their electricity tariff choices in the past, and integrate underlying individual motivations. Our model results show that, on average, German households are willing to pay a small premium of less than 2% for regional electricity generation. However, results also show that WTP differs between respondents. More specifically, we find that respondents with stronger regional product beliefs and green values have a higher WTP. Practitioners, such as energy suppliers, could use this information to explicitly address subgroups of consumers in order to more effectively market regional electricity. However, a substantial part of preference heterogeneity remains unexplained, leading to the conclusion that preferences for regional electricity generation are not solely driven by the aforementioned motivations. Future research could disentangle the factors driving WTP in more detail by incorporating more complex behavioral theories into HCMs.
As there are many types of tourism, related activities, behaviors, and movement patterns, various approaches, and levels of analysis have been developed in academia and tourism practice. While some may not be very sensitive concerning data used and definitions applied, others may be very demanding in this respect.
This paper presents a general methodology and framework for including latent variables—in particular, attitudes and perceptions—in choice models. This is something that has long been deemed necessary by behavioral researchers, but is often either ignored in statistical models, introduced in less than optimal ways (e.g., sequential estimation of a latent variable model then a choice model, which produces inconsistent estimates), or introduced for a narrowly defined model structure. The paper is focused on the use of psychometric data to explicitly model attitudes and perceptions and their influences on choices. The methodology requires the estimation of an integrated multi-equation model consisting of a discrete choice model and the latent variable model's structural and measurement equations. The integrated model is estimated simultaneously using a maximum likelihood estimator, in which the likelihood function includes complex multi-dimensional integrals. The methodology is applicable to any situation in which one is modeling choice behavior (with any type and combination of choice data) where (1) there are important latent variables that are hypothesized to influence the choice and (2) there exist indicators (e.g., responses to survey questions) for the latent variables. Three applications of the methodology provide examples and demonstrate the flexibility of the approach, the resulting gain in explanatory power, and the improved specification of discrete choice models.
Although computer programs may estimate values for unidentified parameters, a parameter must be identified in order for there to exist a unique point estimate of its value. We provide a series of rules that can be applied easily to measurement models of complexity one to demonstrate the identifiability of their parameters. These rules can be applied to models that contain one or more latent variables and that contain observed variables with correlated measurement errors. If the model is not identified, the rules pinpoint the parameters that are not identified and, thus, help researchers formulate a testable model.
Model Notation, Covariances, and Path Analysis. Causality and Causal Models. Structural Equation Models with Observed Variables. The Consequences of Measurement Error. Measurement Models: The Relation Between Latent and Observed Variables. Confirmatory Factor Analysis. The General Model, Part I: Latent Variable and Measurement Models Combined. The General Model, Part II: Extensions. Appendices. Distribution Theory. References. Index.