Role of maternal non-transport pro-environmental behaviors in adolescents’ travel-to-school mode choices

Abstract This study investigates in UK context the relationship between adolescents’ choice of sustainable transport modes (e.g. active transport like walking or cycling and public transport like buses or subways) for their journey to school and maternal non-transport pro-environmental behaviors, such as energy conservation and environmentally friendly purchases, as well as its temporal changes. Data from waves 4 and 10 of the UK Understanding Society survey were separately analyzed using multinomial logistic regression to explore the relationship between frequency of mothers’ non-transport pro-environmental behaviors and adolescents’ sustainable transport to school. Additionally, to understand changes in the strength of this relationship over time, a regression analysis was conducted examining the interaction of mothers’ non-transport pro-environmental behaviors with the survey year. Findings indicate substantial correlations between an array of variables including adolescents’ age, ethnicity, mothers’ occupational and transport behaviors, the number of cars owned by the household, and the nature of residence (urban vs rural), with the adolescents’ active or public transport choice to school, consistently across both waves. As the primary focus of the study, a positive relationship between mothers’ non-transport pro-environmental behaviors and adolescents’ public transport to school is found, although the strength of this relationship declined over time. Importantly, more easily observable mothers’ non-transport pro-environmental behaviors holds a stronger strength of correlation with adolescents’ use of public transport to school, compared to maternal psychological factors like pro-environmental attitudes. Hence, encouraging a range of sustainable behaviors among mothers is crucial to promote adolescents’ public transport to school.


Introduction
Adolescents' transport choices for their daily commute to school can significantly affect their health and the environment.For example, the benefits of the adoption of active transport modes (e.g.walking, bicycling, or using scooters/ skates) for commuting to school on adolescents' physical and mental well-being are well documented (Jussila et al., 2023;Xu et al., 2013).In terms of environment, the transition from private vehicle use to active or public modes of transport for school commutes can reduce traffic-related air pollution (Bearman & Singleton, 2014;Sukor et al., 2017).Kaiser et al. (2007) suggested that using sustainable transport modes to school (i.e.active and public transport) can be considered a pro-environmental behavior among adolescents, encompassing actions that minimize negative environmental impacts or even produce positive effects.
Despite these benefits, the past few decades have witnessed a decline in adolescents' use of sustainable transportation to school, along with a surge in private car use in several countries, including the Czech Republic (Haug et al., 2021), the United States (McDonald et al., 2011), Brazil (S� a et al., 2015), England (Department for Transport, 2022), andScotland (Hands Up Scotland Survey, 2021).More specifically, in England, for example, between 1995 and 2019, the percentage of adolescents commuting to school by car rose by nearly 4% (Department for Transport, 2022).Identifying the factors that correlated with adolescents' transport choice to school can inform interventions to promote sustainable transport choices (Steg & Vlek, 2009).
Previous research has underscored the crucial role of parental factors, including sociodemographic aspects such as parental employment status and education level, as well as psychological components such as parental positive attitudes toward the natural environment and environmental protection (pro-environmental attitudes), in influencing adolescents' transport choices for school commutes (Aranda-Balboa et al., 2020;Assi et al., 2018;Black et al., 2001;Mehdizadeh et al., 2019).Moreover, the parental transportrelated pro-environmental behaviors (e.g.sustainable transport behaviors) are also among the important factors related to the adolescents' sustainable transport choice (Beck et al., 2023).Siiba (2020) has documented a positive correlation between parents' regular engagement in active commuting, such as walking or cycling at least twice a week, and their children's inclination toward adopting these active modes for their school commutes.
Although prior research has highlighted the correlation of parents' transport-related pro-environmental behaviors with adolescents' sustainable transport choice to school, it remains unclear whether other parental pro-environmental behaviors such as green purchasing and energy conservation (non-transport pro-environmental behaviors) are also correlated with adolescents' transport choice to school.Existing research found that parental these non-transport pro-environmental behaviors have correlation with adolescents' formation of pro-environmental psychological constructs such as pro-environmental attitudes and values (Collado et al., 2017;Gong et al., 2022;Grønhøj & Thøgersen, 2017;Jia & Yu, 2021;Wallis & Kl€ ockner, 2018).Notably, once formed, adolescents' these pro-environmental psychological constructs may catalyze a range of pro-environmental behaviors that extend beyond the confines of the parental non-transport pro-environmental activities that initially influenced the constructs.It is therefore possible that parental non-transport pro-environmental behaviors can correlated with a greater inclination toward sustainable transport to school among adolescents.However, there is no research directly examining this correlation.Exploring the role of parents' these non-transport pro-environmental behaviors is essential, as they exhibit greater susceptibility to intervention measures and higher participation rates than pro-environmental transport behaviors (Lynn, 2014).This creates improved opportunities for interventions to promote sustainable commuting among adolescents.
Besides, Previous research has identified inconsistencies between parental pro-environmental attitudes and behaviors (Juvan & Dolnicar, 2014), necessitating an in-depth exploration of the different strength of correlation of parental pro-environmental attitudes and actions with adolescent behaviors.Grønhøj and Thøgersen (2012) argued that compared to parents' pro-environmental attitudes, their behaviors, including green shopping, recycling, and energy conservation, have a stronger strength of correlation with their children's corresponding behaviors.However, no study has examined the different correlation strength of parental non-transportation pro-environmental actions and proenvironmental attitudes with adolescents' sustainable transport to school.Lastly, there has been a significant increase in green shopping behaviors and energy-saving behaviors among adults (parents) over time in UK (Elmarasi, 2017;Huebner et al., 2023).It remains unclear whether this increase impact the strength of correlation between parental non-transport pro-environmental behaviors and their children's sustainable transport to school.
To address these research gaps, this study was conducted in the United Kingdom (UK), focusing on adolescents aged 10-15 years and their mothers.Mothers were selected as the primary focus of this study due to their predominant role in their children's education and their responsibility for escorting them to school (Matthies et al., 2012;Motte-Baumvol et al., 2017).Additionally, Collado et al. (2019) indicated that mothers' non-transport pro-environmental behaviors have a more substantial impact on adolescents' behaviors than fathers' actions.Beck et al. (2023) underscored the important relevance of maternal travel patterns in shaping adolescents' choices of transport, particularly active travel modes, a connection not as evident in paternal travel behaviors.
Specifically, the first aim of this study is to investigate the correlation of mothers' non-transport pro-environmental actions with their adolescent children's sustainable transport choice for commuting to school.Moreover, this study also aims to differentiate the strength of correlation of mothers' non-transport pro-environmental actions and pro-environmental attitudes with adolescents' transport choice to school.Lastly, we aimed to explore whether the increasing prevalence of non-transport pro-environmental behaviors over time in UK amplifies the strength of correlation of mothers' non-transport pro-environmental behaviors with their children's sustainable transport to school.Based on research aim and existing literature, we propose the following hypotheses: Hypothesis 1: The non-transport pro-environmental behaviors of mothers have correlation with sustainable transport to school for adolescents.
Hypothesis 2: Maternal non-transport pro-environmental behaviors have a stronger strength correlation with adolescents' sustainable transport to school than maternal pro-environmental attitudes.
Hypothesis 3: The strength of the association between maternal non-transport pro-environmental behaviors and adolescent's sustainable transport to school increases as maternal nontransport pro-environmental behaviors become more prevalent over time.

Study design
Testing the first two hypotheses is feasible through the analysis of single-year data, whereas verification of the third hypotheses necessitates the integration of data spanning two distinct years (Refer to 3.4.2Multinomial logistic model for methodological specifics).As a result, the study hinged on survey data from two distinct years, specifically from the 2012/14 (Wave 4) and 2017/19 (Wave 9) data of the UK Household Longitudinal Study (UKHLS) (University of Essex, 2019).UKHLS has been used in studies to analyze transport choices (Roberts et al., 2017).
The rationale behind choosing data from Waves 4 and 9 of the UKHLS was the five-year gap between the time points of these two waves of data, which was deemed adequate for observing the desired trends in the growth of non-transport pro-environmental behaviors among mothers.In addition, Wave 4 and 9 data provide comprehensive coverage of variables crucial for this research such as transport choice to school.Another important reason for utilizing Wave 4 and wave 9 data lies in its temporal positioning prior to the commencement of the COVID-19 pandemic in the UK.This holds considerable significance because the pandemic unquestionably had a profound impact on transport behaviors, pro-environmental behaviors and even pro-environmental attitudes (Downey et al., 2021;Ramkissoon, 2020;Shulman et al., 2022).The Wave 4 and 9 data was collected prior to the onset of the UK pandemic, ensuring that it remained untainted by disruptions associated with the pandemic.

Sample characteristics
The Wave 4 of UKHLS dataset included 47,071 individuals aged above 15 years and 4,045 adolescents aged 10-15, who completed the respective questionnaires.In Wave 9, 36,055 individuals aged above 15 years and 2,821 adolescents aged 10-15 participated.To establish mother-child dyads, adolescents who completed the youth questionnaire were linked to their mothers who completed the adult questionnaire.Consequently, 3,846 mother-child pairs were generated in Wave 4 and 2,301 in Wave 9.It is worth noting that to compensate for the missing variables related to pro-environmental attitudes and non-transport pro-environmental behaviors in Wave 9, data from 2018/19 (wave 10 of UKHLS) were included.After supplementing the missing variables in Wave 9 with Wave 10 data and excluding dyads with incomplete data, as well as repeated entries of adolescents or mothers from Waves 4 and 9/10, the study retained 3,287 and 1,985 mother-adolescent pairs for Wave 4 and Waves 9/10, respectively.In addition, adolescents who used multiple transport modes to school were excluded due to the survey's lack of specifications regarding combined modes.The final sample comprised 3,022 mother-adolescent pairs in Wave 4 and 1,890 in Waves 9/10.

Dependent variable
In this study, the dependent variable was defined as adolescents' mode of transportation to school, including three distinct categories: active transport, public transport, and car.This categorization is based on the sustainability of the various modes of transport.Car is regarded as the unsustainable mode of transport due to its significant contribution to pollution.Public transport is one of sustainable transport mode with the medium level of environmental sustainability.While it is motorized and does contribute to environmental pollution, it offers the advantage of accommodating a large number of passengers, which helps reduce per capita emissions (Sukor et al., 2017).Active transport such as walking or cycling is another type of sustainable transport which is at the highest level of sustainability (Bearman & Singleton, 2014).This categorization has been used in several existing literature on sustainable transport to school (M€ uller et al., 2008;Nordfjaern et al., 2016).
These three categories of dependent variable were aggregated based on the most frequently used mode of transportation to school, as answered by adolescents in the two waves (wave 4 and wave 9/10) survey.Adolescents were asked if they usually walked, bicycled, went by car, went by train, or went by bus or subway to school.Specifically, selections indicating "walk all the way to school" or "bike all the way to school" were amalgamated under the "active transport" category of the dependent variable.The "public transport" category of the dependent variable, on the other hand, subsumed those opting for "Ride the bus or subway" or "Ride the train."The unaltered survey response of "car" continued to denote its distinct category ("Car" category of the dependent variable).

Independent variables
Mother's pro-environmental attitudes: To measure mothers' pro-environmental attitudes, six questions like the extent of agreement or disagreement with the notion that 'the changes made must be in line with my current lifestyle' that measured mothers' pro-environmental attitudes toward climate change were included (refer to Table 1 for the other five questions), with responses rated on a 5-point Likert scale (1 ¼ strongly agree; 5 ¼ strongly disagree).In this context, a higher score indicated a stronger pro-environmental attitude or more support for environmental conservation.It is worth noting that the mothers completed these questions in both Wave 4 and Wave 9/10.These questions originated from the New Ecological Paradigm Scale (Dunlap et al., 2000), an instrument developed and continually improved by drawing from decades of environmentalist research commencing in the 1970s.The aim of these questions was to represent a broad understanding of the environment while also capturing fundamental viewpoints and attitudes regarding the interplay between humanity and the natural environment.It is important to highlight that the study treated mothers' pro-environmental attitudes as a latent variable, which was constructed based on each mother's responses to the six questions.More information can be found in Section 2.4.1.
Mother's non-transport pro-environmental behaviors: To measure mothers' non-transport pro-environmental behaviors, we incorporated daily family activities pertinent to the environment and frequently engaged in by mothers.Simultaneously, it was crucial to include behaviors that varied in terms of specific motivations, opportunities, and capabilities to participate.This approach helps avoid the generation of observations that are only valid within the context of specific behaviors (McKenzie-Mohr et al., 1995).It is particularly important to consider which maternal environmental behaviors are more conspicuous in adolescents.This is because the visibility of maternal behaviors plays a significant role in the transmission of pro-environmental behaviors between mothers and children (Matthies et al., 2012).Consequently, based on these criteria, we selected maternal green purchasing behaviors and energysaving behaviors, which have been demonstrated to impact the same behaviors in children (Grønhøj & Thøgersen, 2017).Accordingly, six survey questions were implemented to assess the degree of maternal engagement in energy conservation and eco-friendly purchasing practices within households.The questions included such as inquiries about the frequency with which mothers turn off lights in unoccupied rooms and their regular use of their own shopping bags when shopping (refer to Table 1 for the other four questions).Responses were measured on a scale from 1 (never) to 5 (always), with higher value indicating more frequent behaviors.Consistent questions in both surveys (survey 4 and survey 9/10).
In this study, in both waves, the overall non-transport pro-environmental behaviors score for each mother was calculated by averaging her responses to the six questions, with higher scores indicating greater engagement of non-transport pro-environmental behaviors.The distribution of scores for non-transport pro-environmental behaviors of mothers in waves 4 and 9/10 is shown in Table A1 in the Appendix A. The method to determining scores for mothers' nontransport pro-environmental behaviors was informed by research based on UKHLS data of Hand (2020).The rationale behind this method stems from the recognition that proenvironmental behaviors do not necessarily exhibit internal consistency.For instance, a mother might be actively engaged in certain non-transport pro-environmental activities but not in others.As a result, treating non-transport pro-environmental behaviors as a latent variable, which assumes internal consistency among the behaviors, is not appropriate.Instead, these behaviors are more accurately conceptualized as a formative construct by calculating the mean (Diamantopoulos et al., 2008).In a formative construct of mother's non-transport pro-environmental behaviors, mother's each behavior contributes to the overall construct of pro-environmental behaviors, but they do not have to correlate strongly with one another, acknowledges the multifaceted and heterogeneous nature of pro-environmental actions (Coltman et al., 2008).
To enhance the reliability and robustness of the study, based on the relevant literature (Irawan et al., 2022;Marzi et al., 2023;Yeung et al., 2008), our research incorporated a variety of control variables.These included demographic characteristics of the adolescents such as age, gender, and ethnicity; household factors like income, whether they live in an urban or rural area, and car ownership; and maternal socio-demographic elements like employment status and education level.We also considered the travel behaviors of mothers, encompassing the frequency of cycling and using public transportation.This consideration is important as the travel behaviors of mothers can partially reflect the quality of sustainable transport infrastructure and services available in their respective regions.The details of these variables are presented in Table 1.

Confirmatory factor analysis
The initial phase of this research applied confirmatory factor analysis (CFA), an established analytical tool belonging to structural equation modeling techniques that investigate relationships between latent variables and observed items in Taking along a shopping bag for shopping Mother's Pro-Environmental Attitudes (1-strongly agree, 5-strongly disagree) EA1 Any changes I make to help the environment must fit with my lifestyle.

EA2
The so-called 'environmental crisis' has been greatly exaggerated.

EA3
Climate change is beyond our control; it is too late to do anything.

EA4
The effects of climate change will occur too far in the future to truly worry me.

EA5
It is not worth me doing things to help the environment if others do not.

EA6
It is not worth the UK combatting climate change because other countries will cancel out what it does.

Dependent Variable A_school transport choice
Transport choices for adolescents commuting to school: Cars, public transport, and active transport.
a-priori-specified, theory-derived models.First, a hypothesis is formulated about the model structure as a collection of elements underpinning a set of phenomena; next, the degree of covariance between the items the hypothesized factor structure can capture is determined (Mueller & Hancock, 2001).In this research, six questions related to pro-environmental attitudes were employed as observational item in both wave 4 and waves 9/10, which were utilized to measure the latent construct of mothers' pro-environmental attitudes.The CFA model was run in R (R Core Team, 2021), using the weighted least squares mean and variance adjusted (WLSMV) estimator, as all modeled observed variables were ordered as categorical variables with non-normal distributions (Beaujean, 2014).
An essential stage of CFA involves evaluating the goodness of fit of the proposed model and measuring how well the model matches the observed data (Sahoo, 2019).Accordingly, the association between observed and theoretical data was assessed using various model-fit indices measuring this relationship.Model fit indices can be combined with thresholds or hypothesis testing to determine whether the proposed model should be rejected or maintained (Mueller & Hancock, 2001).The chi-square statistic, goodness-of-fit index (GFI), adjusted GFI, root mean square error of approximation (RMSEA), and comparative fit index (CFI) are the five leading model fit indices used in the CFA (Alavi et al., 2020).However, the chi-square statistic is nonparametric and is impacted by sample size (Schermelleh-Engel et al., 2003).Therefore, an overemphasis on the model's chi-square could lead to bias for smaller samples, with the null hypothesis not rejected, resulting in the acceptance of poor models and imprecise parameter estimations (Alavi et al., 2020).Therefore, chi-squared does not contribute to the understanding of the fit of models with large sample sizes (such as the present model), but it is reported for completeness.We calculated the GFI, modified GFI, and RMSEA.According to previous studies, appropriate models will have a CFI, GFI, and AGFI > 0.90, and a RMSEA < 0.08 (Mueller & Hancock, 2001).Furthermore, we tested the reliability and validity of the observational items.
After conducting the CFA, factor scores for pro-environmental attitudes were computed for each mother in Wave 4 and Wave 9/10.These scores were then used as independent in subsequent analyses.Such a methodology has been previously adopted in transportation research, exemplified by the research of Song et al. (2021).The distribution of factor scores for pro-environmental attitudes of mothers in waves 4 and 9/10 is shown in Table A1 in the Appendix A.

Multinomial logistic model
The existing research of studying determinants of adolescents' transport choice to school mainly establish the multinomial logistic model, which is suitable for analyzing the three categorical dependent variables, such as active transport (walking and cycling), public transport, and cars (Li et al., 2022).Therefore, this research uses the multinomial logistic model to examine the association between sustainable transport modes (active and public transport) employed by adolescents for commuting to school and the pro-environmental attitudes and non-transportation behaviors of their mothers.This research employed cross-sectional analyses on samples from waves 4 and 9/10 respectively.It is worth noting that in order to compare the differences in the strength of the association of mother's pro-environmental attitudes and her non-transport pro-environmental behaviors with adolescents' sustainable transport to school, these two variables were sequentially included in the model.Thus, for each of these samples, two multinomial logistic models were formulated.In the Wave 4 sample, Model 1 was constructed considering only the mothers' pro-environmental attitudes, while Model 2 additionally incorporated the mothers' non-transport pro-environmental behaviors.Similarly, for the Wave 9/10 sample, Model 3 was formulated accounting solely for the mothers' pro-environmental attitudes, whereas Model 4 additionally encompassed their mother's non-transport pro-environmental behaviors.Additionally, each model integrated 10 control variables.In all the models, the dependent variable was classified into three distinct transport modes employed by adolescents for commuting to school: active transport, public transport, and car.Among these, the car was designated as the reference category, hence its parameters were fixed at zero.To aid in the interpretation of the models, odds ratios were computed.
To explore the temporal change in the strength of correlation of mothers' non-transport pro-environmental behaviors with adolescents' sustainable transport to school between Waves 4 and 9/10, we combined the samples of wave 4 and wave 9/10, and evaluated a two-way interaction term involving mothers' non-transport pro-environmental behaviors and survey year (mothers' non-transport proenvironmental behaviors � year) using a multinomial logit model (model 5), where the car was the reference category.Survey year was applied as the categorical variable in the model.By examining the regression outcomes of the model (outcomes related to interaction term), we determined whether temporal change in the strength of correlation between mothers' non-transport pro-environmental behaviors and adolescents' sustainable transport to school were statistically significant.This method has been applied in multiple studies to analyze temporal shifts in the strength of correlation between independent variables and dependent variables (Olsen et al., 2017;Scholes & Bann, 2018).

Results
Table 2 provides descriptive statistics on questions related to mothers' non-transport pro-environmental behaviors and pro-environmental attitudes in our sample.Between Wave 4 and Wave 9/10, there was a noticeable increase in the frequency of the practice of shopping with a shopping bag, turning off lights in an empty room, switching off the tap while brushing teeth, and refraining from purchasing items due to excessive packaging.The most substantial increase was observed in the behaviors of mothers shopping with their own bags during Wave 4 to 9/10.This could potentially be attributed to the implementation of UK regulations that imposed limitations on the utilization of disposable plastic bags, which were enforced in July 2015 (Kish, 2018).Regarding to pro-environmental attitudes, the results suggest there was a statistically significant uptick in mother's responses to all questions related to pro-environmental attitudes.
Table 3 displays the standardized factor loadings for latent variable (mothers' pro-environmental attitudes) in CFA of Wave 4 and Wave 9/10.All standardized factor loadings are statistically significant and are higher than the 0.4 suggested by Stevens (2012).Table 3 also illustrates our model fit indices of the CFA, in which GFI, AGFI, and CFI exceeded 0.9, whereas RMSEA was less than 0.05, indicating an acceptable CFA fit.
Moreover, Table 3 provides insight into the reliability and validity indicators for the latent variable construct, "mother's pro-environmental attitudes," in the CFA model.It is evident that in both waves, both Cronbach's ? and the composite reliability exceeds the recommended threshold of 0.7 (Cheung et al., 2023), establishing good reliability for the latent variable construct.Additionally, the average variance extracted for the latent variable construct exceeded 0.5 in both waves, demonstrating that the research scale employed exhibited acceptable convergent validity (Fornell & Larcker, 1981).Since there was only one latent variable in the CFA for this study, no discriminant validity test was conducted.
Table 4 presents the descriptive statistics for Wave 4 and Wave 9/10.The proportion of male and female adolescents, and their age distribution were similar between the groups.Certain demographic shifts were observed between Wave 4 and Wave 9/10 (according to result of Chi-Square Tests).There was a decrease in the proportion of adolescents from white ethnic groups, while the proportion of mothers with a degree or higher degrees, as well as the proportion of working mothers, saw an increase.A downward trend was noticed in the proportion of mothers traveling by bicycle at least once per week.Pertaining to household factors, an increase was noted in the proportion of households owning more than one car and those with a net income exceeding £2,000 per month.However, the distribution of households in urban areas remained consistent across both waves.Changes in the variables identified within the groups align with the findings of population survey in UK (Office for National Statistics, 2022).
Regarding the dependent variable, the selection distribution of diverse transport modes to school was uniformly maintained across both surveys.The highest proportion of adolescents engaging in active transport methods was consistently observed in both sample groups, while the smallest proportion was noted in the use of public transport.
Table 5 presents the results of the multinomial logistic regression based on samples from waves 4 and 9/10 respectively.Models 1 and 2 used a sample of 3,022 adolescents and their mothers from Wave 4, whereas Models 3 and 4 drew from a sample of 1,890 adolescents and their mothers from Wave 9/10.Adolescents' demographic characteristics (age, gender, ethnicity), mothers' socio-demographic characteristics and transport behaviors (mother's job, educational qualifications, frequency of bus and bicycle trips), and household factors (car ownership, living in the rural area, household income) were included as control variables in each model.It is worth noting that in order to compare the differences in the strength of the association of mother's pro-environmental attitudes and her non-transport proenvironmental behaviors with adolescents' sustainable  transport to school, these two variables were sequentially included in the model.Thus, the first models in each sample set, that is, Models 1 and 3, focused on control variables, and mothers' pro-environmental attitudes.Models 2 and 4 further included mothers' non-transport pro-environmental behaviors.Model 5 combined the data from Waves 4 and 9/ 10, incorporating 4,912 adolescents and their mothers.Model 5 incorporated an interaction term for the survey year with mothers' non-transport pro-environmental behaviors to explore the temporal change in the strength of correlation between mothers' non-transport pro-environmental behaviors and adolescents' sustainable transport to school during Waves 4 to 9/10.Each model utilizes the same dependent variable categories: active transport (A), public transport (P), and cars (C).The car use is defined as reference category.Several control variables, including adolescent age, ethnicity, mother's job status, bicycle and bus use frequency, car ownership, and living area type, showed statistically significant associations with adolescents' choices of active or public transport choice to school across all models.The directions of these associations align with previous research findings (Assi et al., 2018;Carver et al., 2013;Siiba, 2020;Simons et al., 2017).In terms of independent variables, in Models 1 and 3, a positive relationship was detected between a mother's pro-environmental attitudes and her child's use of public transport to school (OR¼ 1.37 and 1.66, p < 0.05).However, in Models 2 and 4, this correlation became statistically insignificant, after the mothers' non-transport pro-environmental behaviors was included in the model.Moreover, there was a statistically significant positive correlation between mothers' non-transport pro-environmental behaviors and adolescents' choice of public transport to school in Models 2 and 4 (OR¼ 1.27 and 1.63, p < 0.05).Notably, both of mothers' pro-environmental attitudes and non-transport pro-environmental behaviors had no correlation with adolescents' choice of active transport to school.Model 5 examined changes in the strength of relationship between mothers' non-transportation pro-environmental behaviors and adolescents' sustainable transport choices to school from Wave 4 to Wave 9/10.A notable finding was the diminishing strength of correlation of parental non-traffic-related pro-environmental behaviors with adolescents' preferences for public transport modes to school over time (OR¼ 0.79, p < 0.05).

Discussion
In the context of the UK, this study explored the relationship between adolescents' selection of sustainable transport modes for their journey to school and maternal non-transport pro-environmental behaviors (e.g.energy conservation and environmentally friendly purchases), as well as its temporal variations.Furthermore, the study differentiated the strength of correlation of mothers' non-transport pro-environmental actions and pro-environmental attitudes with adolescents' transport choice to school.The primary findings indicate a positive correlation between maternal non-transport pro-environmental behaviors and adolescents' adoption of sustainable transport to school.Nonetheless, this correlation displays a decreasing strength over time.Notably, this correlation consistently demonstrates a stronger strength of association compared to the relationship between children's travel-to-school mode choices and mothers' pro-environmental attitudes.Additionally, the study identified correlations between various control variables and adolescents' mode of travel to school.

Results interpretation
Concerning control variables, our study found that a clear relationship between working mothers and an increased likelihood of adolescents choosing active and public transport to school.This is reasonable, as employment could limit the time available for mother to drive their children to school (He & Giuliano, 2017).Furthermore, adolescents whose mother cycled infrequently (less than once weekly) demonstrated a lower propensity for selecting active transport methods to school, compared to those whose mothers engaged in cycling at least once per week.This observation aligns with prior study (Siiba, 2020), which documented a correlation between parental active commuting behaviors and their offspring's active transport to school.It is worth noting that this study found that mothers who used buses less frequently each week were less likely to have children who chose either active or public transport to school than those who used buses at least once per week.This discovery expands upon the research finding of Siiba (2020) by demonstrating that not just parental active transport behaviors are linked to adolescents' active commuting to school, but the parental public transport behaviors is also connected to adolescents' active transport to school.Our analysis also showed that residing in rural areas enhanced the likelihood of using public transport to school, possibly due to increased subsidies from rural authorities for school bus programs in UK (Van Ristell et al., 2015).In contrast, living in a rural area diminished the chances of active commuting to school, possibly due to longer distances between home and school in rural area (Mandic et al., 2017).Moreover, an inverse correlation was observed between adolescents whose mothers had attained a degree or higher and their inclination to engage in active transport to school.Similarly, there was a negative relationship between the net household income and the preference of adolescents from such households to choose active and public transport options.Ermagun and Samimi (2015) provide a possible interpretation, positing that educated parents may have a heightened risk awareness that deters them from allowing their children to walk or use public transport.Noonan (2021), examining the British scenario, introduces a different explanation by arguing that families with higher incomes tend to exercise choice over school-selection.This frequently corresponds to increased commuting distances, thereby fostering a predilection for car-based commuting among adolescents (Fyhri et al., 2011).This propensity is also evident among families where Odds ratios: mjn for an explanatory variable is the odds ratio for a unit increase in the explanatory variable of mode m 2 (A, P) versus mode n 2 (C), where A denotes active transport (walking/biking), C denotes car, and P denotes public transport (train, subway or bus).
The model 1 and 2 were estimated using 3,022 adolescents and their mothers based on Wave 4 samples.The model 3 and 4 were estimated using 1,890 adolescents and their mothers based on Wave 9/10 samples.
parents have higher educational qualifications (Burgess et al., 2011).In terms of independent variables, our findings indicate that non-transport pro-environmental behaviors exhibited by mothers have a crucial correlation with adolescents' preferences for public transport to school, which confirms the previously postulated Hypothesis 1.One possible interpretation of this correlation is that mothers can invariably convey the importance of environmental protection to adolescents through their daily non-transport pro-environmental behaviors such as energy conservation and environmentally conscious shopping, thus fostering adolescents' psychological constructs related to pro-environmental behaviors such as pro-environmental attitudes and awareness of responsibility for environmental protection.It is worth noting that once formed, these adolescents' psychological constructs may catalyze a range of behaviors such as public transport to school beyond the scope of parental non-transport pro-environmental activities that initially influenced the form of psychological constructs.This finding enriches the existing literature on the intergenerational transmission of pro-environmental behaviors, indicating that this transmission should not be limited solely to similar behaviors between parents and children.
Additionally, this research discovered a positive connection between the pro-environmental attitudes of mothers and the use of public transport by adolescents when traveling to school, solely when considering controlled variables and mothers' pro-environmental attitudes.Prior studies have posited that the parental pro-environmental attitudes indirectly correlated with their children's choices in using sustainable transport methods to school, primarily through the parents' own sustainable transport practices (Black et al., 2001;Mehdizadeh et al., 2019).However, in this study, when controlling for mothers' sustainable transport behaviors, both mothers' pro-environmental attitudes and sustainable transport behaviors remained associated with adolescents' public transport to school.It is worth noting that this association was no longer statistically significant when mothers' non-transport pro-environmental behaviors, sustainable transport behaviors, and pro-environmental attitudes were also considered.Instead, the statistical significance shifted to the positive correlation between mothers' non-transport pro-environmental behaviors and adolescents' use of public transport on their way to school.Moreover, the model encompassing controlled variables, mothers' proenvironmental attitudes and non-transport pro-environmental behaviors offers a superior fit for the data compared to the model that solely incorporates controlled variables and mothers' pro-environmental attitudes, as indicated by the Akaike Information Criterion.Therefore, this result underscores the intricate link between maternal pro-environmental attitudes and the adolescents' transport choice to school.The correlation between mothers' pro-environmental attitudes and adolescents' use of public transport to school may be mediated by mothers' non-transport pro-environmental behaviors.Moreover, this result also demonstrated that mothers' non-transport pro-environmental behaviors have the stronger strength of correlation with adolescents' public transport choice of commuting to school than mothers' proenvironmental attitudes.This conclusion aligns with our second hypothesis.
Moreover, in terms of the change in mother's non-transport pro-environmental behaviors, in our study, the frequency of mother's most of non-transport proenvironmental behaviors was either already at a high level or had experienced a noticeable increase over time (from wave 4 to wave 10).Moreover, a notable finding was the diminishing strength of correlation of mother's non-trafficrelated pro-environmental behaviors with adolescents' preferences for public transport modes to school over time.This result contradicted our third hypothesis.These findings suggest that as pro-environmental behaviors become more commonplace among mothers, adolescents may perceive them as ordinary rather than exceptional, leading to reduced attention.Consequently, the ability of these mother's behaviors to shape adolescents' psychological constructs related to proenvironmental behaviors such as pro-environmental environmental awareness and attitudes may weaken, ultimately leading to the decrease in the strength of correlation between mother's non-transport pro-environmental behaviors and adolescents' preferences for public transport modes to school.

Implications
Our findings have several implications.In terms of the result about correlation between mother's non-transport pro-environmental behaviors and adolescents' preferences for public transport to school, this result suggests that interventions aimed at promoting sustainable transport behaviors among adolescents should consider involving and targeting not only adolescents but also their parents.This result informs future intervention development to increase the use of sustainable transportation to schools by improving a broader range of mother's pro-environmental behaviors.Moreover, the result about the diminishing strength of correlation of mother's non-traffic-related pro-environmental behaviors with adolescents' preferences for public transport modes to school over time emphasizes the importance of continuously and deliberately communicating the significance and positive environmental implications of these behaviors to adolescents.This ensures that, despite mother's these behaviors increasing normalization, the environmental importance and rationale of mother's these behaviors are clearly understood by adolescents, which is essential for promoting adolescents' sustainable transport to school through their mothers' a broader range of non-transport pro-environmental behaviors.In addition, the result about correlation between mother's non-transport pro-environmental behaviors and adolescents' public transport to school has important implications for the evaluation of measures aimed at promoting sustainable behaviors in parents and adolescents.That is, it may be important to consider changes in a range of maternal and child sustainable behaviors when evaluating interventions designed to promote maternal sustainable behaviors.For example, when assessing the impact of an intervention on maternal energy-saving behaviors, it may be useful to consider changes in the sustainable travel behaviors of adolescents, in addition to changes in maternal and adolescent energy-saving behaviors.

Limitations
Certain limitations were noted in this study.The exclusion of adolescents who use multiple transport modes could bias the results for certain sustainable transport choices, given that multimodal journeys likely involve at least one ecofriendly method, such as walking, biking, or public transport.Despite this limitation, the "multiple modes" category constituted a small fraction of the total responses (8% from wave 4 and 4% from wave 9/10), thus its exclusion is unlikely to significantly skew our overall analysis and findings.Additionally, our study solely focused on walking and cycling as active transport modes to school, overlooking other active transport forms such as scooters.However, according to Department for Transport (2022), in England, the utilization of these other active transport forms for adolescents commuting to school amounted to less than 1% in both 2014 and 2019.Therefore, the lack of other active transport modes is unlikely to seriously affect our overall analysis and results.Moreover, given the secondary data used in this study, certain pertinent variables, such as maternal safety concerns, transport-related attitudes, adolescents' pro-environmental attitudes, and distance to school were not accounted for, as they were not included in the original dataset.Despite this limitation, their absence did not fundamentally diminish the credibility of our research for four reasons.First, the impact of maternal safety concerns on adolescents' public and active transport choices to school is expected to decrease with child age.Our target population, adolescents aged 10-15 years, is a group in which these concerns are less influential (van den Berg et al., 2020).Second, while the study did not thoroughly investigate maternal transport-related attitudes, potentially introducing an omitted variable bias, it importantly considered maternal transport behaviors.Such behaviors inherently mirror transport attitudes, serving as their practical manifestations (Abrahamse et al., 2009).Therefore, by accounting for these maternal transport behaviors, the study indirectly encapsulates the corresponding attitudes, mitigating potential biases.Third, adolescents' pro-environmental attitudes, while of interest, are potential intermediary variable as they could be shaped by parental non-transport pro-environmental behaviors and may correlated with adolescents' sustainable transport choice to school.As our study primarily examined whether there is a relationship between mother's non-transport pro-environmental behaviors and adolescents' sustainable transport to school, the absence of data on adolescents' attitudes is unlikely to skew our results significantly.Finally, distance to school, although a critical factor in active transport to school, does not detract from the result of our study.We contend that the correlation of mother's non-transport pro-environmental behaviors and adolescents' public transport to school is independent of distance.For instance, mother's non-transport pro-environmental behaviors may form their adolescent's psychological constructs related to pro-environmental behaviors, encouraging adolescents to choose public transport to school over car, even over long distances.

Conclusions
This research investigated the relationship between mothers' non-transport pro-environmental actions, such as energysaving practices and environmentally friendly purchasing, and adolescents' choices of sustainable transport to school in UK.The study utilized data from Waves 4 and 9/10 of the UKHLS.The contribution of this study is the identification of the correlation between mothers' non-transport proenvironmental behaviors and the use of public transport by adolescents to get to school.Moreover, the strength of this correlation may weaken as mothers increasingly engage in non-transport pro-environmental behaviors over time.This study enhances our comprehension of the factors influencing adolescents' decisions regarding transport choice to school and offers practical insights that can be valuable for policymakers, sustainable development advocates, and public health professionals.Future research should further investigate whether adolescents' pro-environmental attitudes or other psychological factors associated with pro-environmental behaviors mediate the relationship between mothers' non-transport pro-environmental actions and adolescents' use of sustainable transportation to school.Additionally, it would be valuable for future research to explore potential connections between fathers' and other family members' non-transport pro-environmental behaviors and adolescents' transportation choices to school.The scores for mothers' non-transport pro-environmental behavior were computed by taking the average of the responses (ranging from 1, indicating "never," to 5, indicating "always") across all behavioral questions.Mothers' environmental attitudes, on the other hand, were derived from the factor scores obtained through Confirmatory Factor Analysis (CFA).

Table 1 .
Definition of variables.

Sociodemographic Traits and Travel Behaviors
M_Job Mother's employment status at the time of survey: working or not working M_Qualification Mother's education level: degree or higher, or no degree M_bikefeq Mother's frequency of traveling by bike: At least one time a week, or less than one time a week M_busfeq Mother's frequency of traveling by ordinary bus: At least once a week, or less than once a week Household Factors Income Net household income, where "net" refers to the net of taxes on earnings and national insurance contributions: < £1,000, £1,000-£2,000, or > £2,000 per month.Urban or rural Binary indicator of whether an address is in an urban or rural area Car Number of cars owned by the household: none, one, or more than one.Mother

Table 3 .
CFA factor loadings and goodness of fit.

Table 5 .
Continued.Confidence IntervalThe dependent variable is the transport to school mode choice: active transport, public transport, and car use.Car use is designated as the reference category.

Table A1 .
The distribution of scores for non-transport pro-environmental behaviors and attitudes of mothers in waves 4 and 9/10.