1. Questions
Loyalty in public transit has been studied to curb current trends in loss of public transit ridership. The reasoning is that loyal public transit users are more willing to keep using transit services overtime and more prone to recommend them to others and to attract new riders (Webb 2010). Even so, there is no consensus on the literature on how loyalty to transit should be conceptualized (Carvalho, Romano, and Gadda 2021; van Lierop, Badami, and El-Geneidy 2018). Most scholars examine the construct as a combination of willingness to reuse and to recommend (Minser and Webb 2010). Loyalty has also been addressed as a one-dimensional construct (Sun and Duan 2019) and several other variables have also been considered.
For instance, there is an ongoing debate on the role of satisfaction in loyalty. While some argue that satisfaction is only strongly related to loyalty but not part of the construct itself, others believe that satisfaction should be added as a dimension (Zhao, Webb, and Shah 2014). The rationale is that transit riders would only be willing to reuse and to recommend transit if they are satisfied. Another possible dimension is importance, which refers to the relevance of an object to a person (Zaichkowsky 1985) and is addressed in this paper as the perceived importance of transit to quality of life. It is believed that loyal users are more emotionally involved with the service, which reflects in higher switching costs (Webb 2010). By reviewing the literature, van Lierop et al. (2018) argue that loyalty is defined best when willingness to reuse, willingness to recommend, satisfaction, and importance are included.
We ask the following research questions: (i) is there a single underlying dimension across willingness to reuse, willingness to recommend, satisfaction, and importance to quality of life? (ii) if so, is the 4-variable structure more reliable than the common 2-variable structure (willingness to recommend and willingness to reuse) or a combination of the 2-variable structure and satisfaction and importance to quality of life individually? and (iii) are the factor results consistent across different contexts? We select Montréal, Toronto, and Vancouver as our case studies. All regions have similar rates of transit mode share and older adults’ population as shown in Table 1.
2. Methods
Drawing on data from the 2023 Aging in Place Survey, we conducted an exploratory factor analysis (EFA) for complete cases of senior transit users from the Greater Montréal (n = 577), the Greater Toronto (n = 408), and the Greater Vancouver (n = 273). Willingness to reuse was measured by “I plan to keep using public transit in my region within the next year”, willingness to recommend by “I would recommend public transit in my region to a friend or family member”, satisfaction by “overall, I am satisfied with the public transit services in my region” and importance to quality of life by “public transit positively impacts my quality of life”. All variables were measured on a 4-point Likert-scale from strongly agree to strongly disagree (neutral not included). The factorability of the samples was ensured by addressing inter-correlation (Pearson correlation > 0.30), the Bartlett test of sphericity, and the measure of sampling adequacy (MSA).
The factor analyses for each combination of variables and region were conducted in R using the psych package. We applied common factor analysis as the extraction method, which is better suited for identifying latent constructs (Hair et al. 2014). The number of factors to extract was defined based on latent root (eigenvalue ≥ 1) and percentage of variance (at least 6o%) criterions and varimax was applied as the rotation method. To reduce the influence of non-normality on the results, the correlation matrix was defined using polychoric correlation as it better deals with variables with less than five categories with asymmetrical distributions (Watkins 2018). Reliability was measured by the Cronbach’s alpha coefficient. Results are compared across regions to assess the consistency of the loyalty construct across regions.
3. Findings
Montreal’s seniors are more willing to reuse and to recommend transit, more satisfied, and more likely to perceive a positive impact of transit in their quality of life than seniors from other regions, as demonstrated by pairwise Kruskal-Wallis’s tests (Table 2). Samples from Toronto and Vancouver are not significantly different across all variables.
In the factor analyses (Table 3), for all variable combinations and regions, a one-factor solution was derived indicating the presence of a single construct. Nonetheless, factors including satisfaction for Toronto and Vancouver displayed limited reliability due to either displaying illogical loadings due to negative variance (Heywood case) or by not sufficiently explaining satisfaction (low communality). Consequently, satisfaction would have to be dropped from the analyses in both cases to improve fit. Therefore, the analysis offers limited support for satisfaction as a dimension in the loyalty construct.
The 2-variable and 3-variable combinations with importance to quality-of-life showed adequate fit across all regions. Reliability in terms of Cronbach’s alpha scores improved from the 2-variable to 3-variable structure, while retaining similar levels of internal factor consistency (item-to-score and inter-item correlations) across all regions. Moreover, levels of variance explained also improved in Toronto and Vancouver, providing evidence for including importance to quality of life as a third dimension of loyalty across all regions.
In conclusion, the factors do not provide consistent support for satisfaction as part of the loyalty construct. Nonetheless, importance of public transport to quality-of-life is supported as a third dimension to be included in the loyalty construct across all contexts. Montreal was the only location where all four combinations of variables displayed adequate levels of fit, which might be explained by the differences in public transit structure compared to the two other regions. Our findings contribute to the understanding of the dimensions of loyalty and how to operationalize it in surveys. Future studies can further test our findings by assessing a more diverse range of public transit users and by evaluating discriminant validity by adding more constructs.
Acknowledgment
The authors would like to thank Meredith Alousi-Jones for her help in constructing and administering the survey, Merrina Zhang and Isabella Jimenez from the National Research Council of Canada (NRC) for their feedback on the survey construct. Thanks to Daniel Schwartz from McGill IT services for his support in managing the survey. This research was funded by the Natural Sciences and Engineering Research Council of Canada grant Towards a better understanding of the determinants and satisfaction of travel among different groups in major Canadian Cities (NSERC RGPIN-2023-03852), the National Research Council Canada’s grant Getting Around to Age in Place: Meeting Older Canadians Transport Needs (NRC AiP-023-1), and the Social Sciences and Humanities Research Council’s partnership grant Mobilizing justice: Towards evidence-based transportation equity policy (SSHRC 895-2021-1009).