Introduction

Interest in alternative modes of transportation has increased with the growing concern about various environmental impacts of the transportation system. It is estimated that 29% of the greenhouse gas emissions (GHG) in the United States (U.S.) come from the transportation sector1. This significant contribution from transportation emissions presents an urgent need to reduce overall GHG emissions in the U.S. by adopting environmentally friendly modes of transportation. Over the past decade, travelers have witnessed a growing number of such transportation modes with electric options surging the market (electric vehicles, hybrid vehicles, e-scooters, e-bikes, and others). The availability of these new modes can alter travel choices and GHG emissions as a result. A critical, yet often overlooked, factor in people adopting different modes of transportation is their awareness of their carbon footprint. Improved understanding of one’s travel-related carbon footprint may have the potential to accelerate the transition to decarbonized travel. Carbon calculators may be useful tools for increasing awareness of the relative CO2 contributions of different transportation modes, and consequently altering human travel behavior to reduce CO2 emissions. Accordingly, in this study, we aimed to evaluate the efficacy of carbon calculators to (i) improve users’ understanding of carbon emissions associated with different travel modes, and (ii) change the travel behavior of users.

There have been many studies that examined factors that influence travel behavior2,3,4,5,6,7,8, and separately, the efficacy of carbon calculators to change consumption behavior beyond travel9. Travel behavior studies have focused on the influence of sociodemographic characteristics, values, and attitudes of travelers. Of note, there is a lack of evidence that sociodemographic variables strongly correlate with environmental values10. Other travel behavior studies have focused on the impact of travel-activity companionship2, happiness5, weather8, and Covid-196. It has been found that income, vehicle ownership, safety, and comfort were the dominant influential travel mode choice factors3. While, Donald et al.4 found that trip cost, services provided, journey duration, and vehicle availability were the most influential factors. In general, environmental awareness was not considered as a travel mode choice factor. Separately, Dreijerink and Paradies9 provided a comprehensive literature review of studies that focused on carbon calculator use (including, although not limited to, travel calculators for travel), and found that carbon calculators are often used by an environmentally motivated demographic and that environmental awareness and behavior are not necessarily linked.

While a carbon calculator has the potential to serve as an intervention to change travel behavior, much of the related research varies in methodological approach. In this instance, Dreijerink and Paradies (2020) considered study quality based on the number and diversity of participants, how data is reported (e.g., self-reported), and incorporation of a control group4,9. Donald et al.4 observed that pre- and post-intervention studies are few. In our study, a transportation mode carbon calculator was coupled with surveys to examine the effectiveness of the carbon calculator in changing the travel behavior of users. Moreover, the survey component of our study served to compare pre- and post-intervention travel choices, environmental awareness, attitudes, and behaviors. We further introduce a CO2 equivalence (CO2e) in the form of a cheeseburger index to our carbon calculator. This was intended to communicate emissions quantities in a more user-friendly and accessible manner. Acknowledging that people in general have a hard time conceptualizing 1 kg of CO211, we further introduce a CO2 equivalence (CO2e) in the form of a cheeseburger index to our carbon calculator. For Madison, Wisconsin (study location), the cheeseburger was a culturally appropriate surrogate, and was intended to communicate emissions quantities in a more user-friendly and accessible manner. Note, CO2e includes CO2 as well as other GHGs. Historically, carbon calculator users have tended to be a more environmentally motivated demographic9, thus communicating emission values in intuitive ways can increase calculator-use participation.

The remainder of the paper is organized as follows: The section “Methods” describes the calculator design, data collection, and modeling methods. Section “Results”, provides an analysis of the results. Consequently, the section “Discussion and conclusions” discusses the study limitations and concludes with final remarks and suggested direction for future work.

Results

Survey data analysis

A total of 49 participants completed the initial survey, and 37 participants completed the final survey. Participants’ demographics included age, gender, race, education level, income, and occupation. Participants were also asked about their travel habits, including mode accessibility. In addition, participants were asked both before and after the calculator use portion of the study to self-assess their attitudes and beliefs about the environment and transportation.

A generally mixed demographic of participants completed the surveys, with the exception of race (the majority of participants were White) (Table 1). There was a 2:1 ratio of female to male participation, and ages ranged from 18 to 66. Income ranged from $14,999 to $200,000, with a mode income of $50,000 to $74,999. While education levels spanned from not completing high school to having obtained graduate or professional degrees, the majority of participants had graduate or professional degrees. This overall high level of educational attainment is consistent with U.S. census data for Madison, Wisconsin (U.S. Census, 2023). Participants represented a wide array of professions, with science, education, and engineering as the dominant fields.

Table 1 Demographic distribution

Participants were polled about their access to transportation, specifically how many modes and types were accessible, and their level of access to public transportation available for all desired destinations. The majority of participants (n = 27, or 55% of participants) reported that they had access to three transportation modes in addition to walking. Only one participant reported access to four modes of transportation. Notably, nine participants (18% of participants) reported access to only one transportation mode. The majority of users strongly agreed that public transportation was only sometimes available.

Participants were asked both before and after the calculator use portions of the study to describe their (i) beliefs about environmental impacts associated with different modes of travel, (ii) knowledge about carbon dioxide equivalent, and (iii) self-assessed level of environmental consciousness (Fig. 1). Among the 37 participants who completed both the initial and final surveys, the majority of people had consistent responses with respect to their thoughts on the relationship between travel mode choice and climate change (most participants somewhat agreed or strongly agreed that travel mode choices have a significant impact on climate change), as well as their self-identification as environmentally conscious (most participants somewhat agreed or strongly agreed). However, after completing the calculator-use portion of the study, two participants reported increased agreement that travel mode choices have a significant impact on climate change, while six participants reported decreased agreement with the statement. Conversely, seven participants reported increased self-identification as environmentally conscious, and two participants reported decreased self-identification as environmentally conscious after completing the calculator-use portion of the study.

Fig. 1: Participants’ attitudes and beliefs before and after using the emissions calculator (self-reported).
figure 1

The first table reports the attitude per participant on the impact of travel mode choice on climate change. The second table shows how the understanding of the measurement unit carbon dioxide equivalent changed. The third table reports the level of environmental consciousness per participant.

Notably, participants’ self-assessment of having a clear understanding of the measurement unit carbon dioxide equivalent (CO2e) changed the most from pre- and post-calculator use. Approximately 46% (16 out of 35) of participants reported an increased understanding of CO2e in the final survey.

In line with understanding CO2e as a unit of measurement and environmental consciousness, participants were asked to rank the relative CO2e emissions contributions of various transportation modes from greatest to least, both before and after completing the calculator-use portion of the study. Of note, 47 participants completed this item for the initial survey, while 35 participants completed both the initial and final surveys. In the initial survey, only 13% (6 out of 47) of participants correctly identified the relatively decreasing CO2e emissions contribution order of car (gasoline), car (hybrid), bus, and car (electric). This is nearly consistent with the percentage of participants (14%, 5 out of 35) that initially correctly identified the relative CO2e transportation mode emissions and completed both the initial and final surveys. Among the participants who completed both surveys and were unable to identify relative transportation mode CO2e contributions, 28% (10 out of 35) of participants improved their responses to the correct sequence in the final survey.

At the end of the study, participants were asked to provide feedback regarding their experiences during the use phase of the transportation mode emissions calculator as well as any lasting impact going forward. Exit questions presented to study participants are presented in Table 2. Responses were categorized by recurring themes. The most surprising finding reported by calculator users was related to CO2e emissions associated with bus use. For instance, users’ comments (paraphrased) about what they found surprising included:

  • The electric car has lower emissions than the bus.

  • Driving a car and taking the bus has less of a difference in emissions than was expected.

  • I thought buses were more energy efficient.

Table 2 Reflection responses

Participants’ repeated comments of surprise regarding bus emissions revealed the lack of awareness about the environmental footprint associated with bus transportation, particularly given the overall highly educated, environmentally conscious demographic. One possible explanation for this awareness gap may be that in the U.S. the bus is often thought of as a mass transit mode, and mass transit is often perceived as a more environmentally friendly option.

Similarly, many participants expressed surprise about the use of cheeseburgers as an environmental CO2e metric. Related statements included:

  • I liked the cheeseburger equivalences as a conversion factor. The cheeseburgers made it easier to understand the CO2 emissions per trip.

  • I had never thought about carbon emissions as cheeseburgers. I think that cheeseburgers make more sense to the average person’s understanding.

  • A high cheeseburger equivalent is shown for even short car rides. I am surprised that transportation generates so much CO2! For reference, I think I am a particularly knowledgeable person on climate change.

Based on the above feedback, it may be plausible that a cheeseburger factor will resonate more than traditional CO2e unit data with populations that are less environmentally conscious. In essence, a cheeseburger-like factor could prove to be uniquely helpful in engaging less environmentally-motivated populations.

Individuals also reported surprise about their environmental footprints related to their transportation mode choices and enlightenment about the relative emissions contributions of different transportation modes, although this did not translate to behavioral changes for all participants. Interestingly, when asked what behavioral changes resulted from calculator use, the most common response (50%) was “none”. However, those responses were generally accompanied by supporting text that indicated that the participant had a—self-assessed—high level of environmental awareness of transportation mode emissions and was already making the most environmentally informed transportation mode choices prior to study participation. Conversely, 50% of responses noted some form of behavioral change, subcategorized as increased transportation mode selection based on reduced CO2e, increased carpooling, route changes, and increased self-reflection. Post-calculator use, participants overwhelmingly reported that they will have an increased level of environmental consciousness going forward, while many participants asserted that they will maintain their already high level of environmental consciousness.

Finally, study participants were asked to rank factors that influence their transportation mode choice. The factors to choose from were environmental impact, comfort, cost, travel time, convenience, exercise, and health and safety (i.e., COVID precautions). Thirty-five participants completed this item in both the initial and final surveys. Initially, the majority of participants responded that the most important factor that influenced mode choice was convenience (37%). Travel time and cost were ranked as the second and third most important transportation mode choice factors by 29% and 26% of participants, respectively. In the final survey, participants ranked travel time as the most important (37%) travel mode choice factor. Results from the final survey also showed that travel time and convenience were tied as second most important to survey participants (tied at 26%).

The dominant mode choice factors of importance identified in this local, city-level study are consistent with findings from other parts of the world with the exception that other studies do not typically include environmental impact among top influential factors3,4,12. However, these studies did not employ a pre- and post-measurement of attitudes and behaviors.

Characteristics of modal shifts

Using our model discussed in section “Modeling transportation modal shifts”, we analyze the influential factors in modal shifts. Notably, we find that trip distance, age, mode of transportation used, environmental awareness, and income were the most important features. Here, feature importance is a technical term related to Decision Trees, whereby the most important features are those that result in the highest predictive power to explain modal output (i.e., whether model shift occurred or not). Figure 2 shows the relevant feature importance.

Fig. 2: Feature importance.
figure 2

The red bars show the mean SHAP value for major features considered in the decision tree model.

However, feature importance is not enough to explain how each feature can impact the decision criteria by the user to change their mode of transportation. For this, we adopt the concepts of SHapley Additive exPlanations (SHAP), to explain how each feature influences modal shift. SHAP value is based on game theory technique and works to assign a value of how the model prediction (i.e., modal shift) changes as features change. The results are shown in Fig. 3.

Fig. 3: SHAP value analysis.
figure 3

Red dots represent a high feature value, while blue dots represent low feature value. The location of these dots along the axis shows the impact of a feature on the model output. A positive (negative) SHAP value corresponds to a positive (negative) contribution towards probability of modal shift.

Interestingly, our analysis reveals that modal shifts can occur at any trip distance. Based on the results, it appears that when trip distance increases, it becomes less likely that one changes the mode. This may be expected, as some modes of travel (e.g., walking, biking, etc.) may not be tenable for longer trips. However, we wonder if more mode choice changes would be observed if more tenable long-distance modality options were available (e.g., railway, which is not an option in Madison, Wisconsin). We note that more complexities to this phenomenon can occur in cases of mode chaining (i.e., using multiple modes for a single trip) and emerging technologies (e.g., electric bicycles), whereby travelers will have a different mode choice behavior as a function of trip length. The authors have prior work on analyzing this for different emerging transportation technologies13,14. As may be reasonably anticipated, we also saw what mode of transportation one is using prior to using the calculator is also influential. Specifically, using the car (gasoline) and bus was found to contribute positively to someone’s modal shift decision. On the contrary, when users were walking to their destination they did not change their original mode. This also follows from the idea that walking has no attributed CO2e emissions and thus does not influence users to change their mode, unlike car (gasoline) or bus. Further, users who already valued environmental emissions as a factor in choosing their desired mode of transportation (rank_EI in Fig. 3) were more likely to change their mode after usage of the calculator. A similar behavior was observed for those who considered trip cost as their primary decision factor in using a mode of transportation. Finally, it was less likely for older users to change their mode, while those with lower income were more likely to change.

Environmental implications of modal shifts

The types and number of transportation modal shifts as a result of participants’ calculator use were documented throughout the study (Fig. 4a). Nearly half (48%) of participants had at least one modal shift as a result of calculator use. Across all trips documented throughout the study, 16% of trips (155 out of 978) had modal shifts. The top three modal shifts observed during the study were: (1) from gasoline-fueled car to bus (n = 37 trips), (2) from bus to walking (n = 34 trips), and (3) from gasoline-fueled car to walking (n = 22 trips).

Fig. 4: Modal shifts and their environmental implications as observed from calculator usage data.
figure 4

a Count of modal shifts observed. b Change in emissions (CO2e kg). Positive values signify a reduction in emissions.

The corresponding CO2e emissions gains or losses due to reported mode changes were recorded as well (Fig. 4b). In addition, calculator use data revealed when users did not elect to change their travel mode choice for a given trip. Typically, documented mode changes resulted in reduced CO2e emissions. Overall, the greatest reductions were observed in mode changes from gasoline-fueled cars to less carbon-intensive transportation modes. This is not surprising, given that gasoline-fueled cars were the greatest CO2e emitters among the modes listed. Although, it is surprising that some observed mode shifts resulted in increased emissions. Specifically, mode changes from bus to gasoline-fueled car (n = 4 trips) and bike to bus (n = 2 trips) were observed. A possible explanation for the shifts to more carbon-intensive modes may be the associated recorded short trip distances, which result in smaller emissions differences among mode choices than for longer trip distances. Consequently, a mode change for short distances may not be perceived as having a significant emissions impact on the individual trip basis. We note that given the nature of surveys, there is always a chance that these reported mode changes may be calculator-user errors. However, survey feedback from participants obtained at the end of the study indicated that some participants reconsidered using the bus since bus emissions were higher than the participants expected, thus supporting the validity of these results.

The cumulative emissions savings from transportation mode changes across all recorded trips was 247.01 kg CO2e (equivalent to 130.07 cheeseburger units per the emissions calculator). At an aggregate level, we estimate a 4.5% reduction in CO2e emissions across all observed trips (956) done by the users). This indicates that the carbon calculator could influence a level of reduction in environmental footprint. Moreover, the emissions savings coupled with increased environmental awareness (section “Survey data analysis”) show a relationship between increased environmental awareness and a change in at least some participants’ travel behavior. Of note, it remains unclear if the participants’ travel mode behavioral changes will continue into the future in the absence of calculator use.

Methods

Study design

This study was conducted in four phases (Fig. 5). First, a pre-calculator use survey was administered to study participants. Information collected included socioeconomic data, travel mode accessibility, travel habits, and self-perception of environmental awareness. Second, study participants used a web-based emissions calculator to determine the emissions generated via different travel mode choices for their individual trips. During this stage, participants input their trip distance and review the resulting emissions data. For each calculator use, users would document their original travel mode choice and their travel mode choice after reviewing the emissions data. Third, study participants completed a post-calculator use survey. Information collected included a qualitative reflection on the utility of the calculator. Lastly, the data obtained from the first, second, and third study phases were processed and evaluated as a whole. Sections that follow provide a detailed account of the emissions calculator design, survey data collection, choice modeling, and limitations and boundaries.

Fig. 5: Study design.
figure 5

First, participants completed a pre-calculator use survey. Second, participants used the travel mode emissions calculator 20+ times. Third, participants completed the post-calculator use survey. Fourth, data analysis.

Emission calculator design

The web-based emissions calculator design included computed transportation mode emission factors, a CO2e conversion to the number of cheeseburgers, and a log for the transportation mode choice before and after emissions calculations.

A use-phase environmental analysis was used to determine local transportation mode emissions factors for Madison, Wisconsin. Specifically, the Greenhouse Gases Regulated Emissions and Energy Use in Technologies (GREET) model15 was used to obtain local-level estimates of the well-to-wheel (WTW) life cycle analysis (LCA) of CO2e emissions for different transportation modes. The GREET model is recognized as the industry standard for transportation-related LCAs. Note that the energy and emissions required to produce vehicle fuel and vehicle use-phase emissions are accounted for in WTW LCA. Six modes of transportation are considered for this study: (1) combustion engine car; (2) hybrid car; (3) electric car; (4) conventional bicycle; (5) electric bicycle; (6) bus; and (7) walking—these are generally the modes of transportation typically used by travelers in Madison. GREET-derived emissions factors were calculated for each transportation mode, with the exception of conventional bicycle and walking (which only require human energy and have zero CO2e emissions). In line with typical Madison transportation energy profiles, combustion engine vehicles were modeled as running on a mixture of 90% gasoline and 10% ethanol by volume (abundant and easily accessible in Madison) and assumed to transport one person per trip. The emission factor for buses was based on a typical Madison bus design and an average number of riders (compression ignition direct injection vehicles that operate on low sulfur diesel, and 13 people, respectively). The electric bike emission factor is based on electricity sourced via the standard Wisconsin state energy mix and an average of 10 WH/mile.

Along with CO2e emissions, the transportation mode emissions calculator gave a CO2e equivalence in the number of cheeseburgers for each trip and transportation mode option. The cheeseburger results were included to provide an additional unit to communicate the extent of environmental impact. The cheeseburger was selected as the unit of equivalency for this study because (i) cheeseburgers are generally familiar to the population within the study location (Madison, Wisconsin), and (ii) burgers have been used to communicate/translate other important concepts (i.e., Big Mac Index to convey exchange-rate theory). In our study, more cheeseburgers are meant to be perceived as undesirable as one would not want to consume too many cheeseburgers in one sitting. For context, an estimated 67 burgers/year are consumed per capita in Wisconsin16. The CO2e conversion to cheeseburgers was based on a prior CO2e-to-cheeseburger conversion by Babakhani et al.17, where one cheeseburger equates to 1.9 kg of CO2e emissions. An example of the calculator’s user interface is shown in Fig. 6. The emissions calculator was built in the summer and fall of 2022.

Fig. 6: Web-based calculator used in the study.
figure 6

Participants insert they trip distance, and the calculator populates mode based emissions and cheeseburger equivalence.

Our calculator finds inspiration from the U.S. EPA’s GHG equivalency calculator; however, there are key differences. Our calculator provides the respective emissions associated with multiple modes of transportation and takes into account trip distance so that a user can make an informed choice about transportation mode selection. Conversely, the U.S. EPA’s GHG equivalency calculator does not include trip distance (so a user would need to use a separate tool to calculate emissions from trip distance) and does not include emissions for all local modes of transportation choices. Furthermore, our survey built into our calculator enables us to document users’ changes in transportation mode choices, whereas the U.S. EPA’s GHG equivalency calculator does not document such changes.

Survey design

Recruitment process

A recruitment process was launched in January 2023, whereby users would use this calculator for their daily trips. Transportation mode study participants were recruited through an email announcement distributed via listserv, an online Craigslist posting, and a paper advertisement posted to two coffee shop message boards. Study participation criteria required that participants live in the greater Madison, Wisconsin, area, be age 18 or over, and have internet access. Participation in the study was confidential. Interested parties were notified that a total of 50 dollars could be earned for completing the surveys. The first 50 respondents to the recruitment postings were selected to participate in the study. This study received Institutional Review Board (IRB) approval of protocol (submission number 2022-1561).

Survey questions

As described in the section “Study Design”, users were required to complete a pre-calculator use survey before gaining access to the travel mode emissions calculator. A complete list of pre-calculator use survey questions can be found in the supplementary information (SI).

The calculator-use portion of the study required participants to record what mode of transportation they intended to use before looking at their carbon footprint and cheeseburger equivalence, and what transportation mode they selected after reviewing the calculator results. Users were required to use the travel mode calculator at least 20 times over a 3-month time period to receive payment for this segment of the study.

To finish the study, users were also required to complete a post-calculator use survey that consisted of 11 questions. Questions included:

  • “What surprised you the most regarding the results of the calculator”

  • “How did your behavior change as a result of using the calculator?”

  • “What is the likelihood of using and recommending the use of CO2e cheeseburger equivalence calculator if it was available in an app?”

Participants were also asked to rank the relative contribution of different transportation with respect to least to most environmental impact per person-miles traveled, as they were asked in the pre-calculator use survey. Similarly, participants were asked to disclose their self-perception of environmental awareness. Thirty-eight participants submitted the final survey. One survey was filtered out as there were no responses included, therefore 37 participants fully completed this portion of the study. A full list of post-calculator use survey questions can be found in the SI.

Data collection process

Data was collected online throughout the duration of the study. Qualtrics, which provides online tools to administer surveys and document survey data, was used to administer pre- and post-calculator use surveys and document survey responses for this study. Study participants were provided with a Qualtrics link to access the pre- and post-calculator use surveys. Data collected throughout the calculator-use portion of the study was obtained through the combined use of Google Sites and Google Forms. Google Sites was used to house the web-based travel mode emissions calculator (link to calculator). Participants were assigned unique identification numbers, which they would enter each time they used the calculator. Participants’ travel mode choices and trip distances were documented for each calculator use. Data obtained through calculator use was saved and accessed via Google Forms.

Modeling transportation modal shifts

Given the collected data on the user’s transportation mode choices—before and after—using the emission calculator, a natural question arises here on what can influence such modal shift. Specifically, we seek to understand what characteristics of the user (e.g., sociodemographic, travel behavior, etc.) and the trip (e.g., distance, mode used, etc.) are influential on the decision-making of the user’s modal shift. To do so, we use the well-known Decision Tree algorithms to build a hierarchical structure of factors influencing the modal shift. Decision Tree is a machine learning model that makes decisions by recursively splitting the data based on features, forming a tree-like structure. It can be used for both classification and regression tasks, albeit here our problem is of a classification nature. A Decision Tree predicts an outcome by traversing the tree from the root to a leaf node. Our model here is one based on a binary classification task, as the overall goal is to be able to predict whether someone changed their mode of transportation or not—after using our calculator. We further use the popular XGBoost algorithm to optimize our tree search. XGBoost is a well-documented and popular approach in such a supervised learning approach. We refer readers to the original paper for more details18. We note that Python was used to train our model, and we supplement this work with the respective code and the trained model.

The model was trained on a total of 956 trips completed by the users. We validate our model based on the cross-validation technique of 80% training data and 20% testing data. The overall accuracy of the model on the testing dataset was found to be 90%.

We then use the SHAP (SHapley Additive exPlanations) analysis to gain further insights from the Decision Tree predictions. SHAP is a method in machine learning interpretability that assigns a value to each feature’s contribution in predicting a specific outcome. It provides insights into the impact of each feature on a model’s predictions, helping to understand and interpret complex models like decision trees.

Discussion and conclusions

The aim of this study was to investigate if a transportation mode carbon calculator could (i) increase users’ carbon footprint awareness and (ii) change users’ travel behavior. The results of this study support the position that carbon calculator use could both increase environmental awareness and promote travel behavior that results in reduced consumption-based travel emissions. Key findings from this study include:

  • Approximately 46% of participants reported an increased understanding of CO2e after using the emissions calculator over a 3-month period.

  • Travel mode changes influenced by calculator use usually, but not always, resulted in decreased environmental footprint.

  • In the rare instances when travel mode changes resulted in increased emissions, the trips were relatively short distances. It’s possible that emissions differences between mode choices were relatively small for short distances, so travelers did not feel like they were faced with an impactful decision.

  • Trip distance, environmental awareness, age, income, and mode of transportation used were the most influential features in predicting modal shifts. These factors are in line with findings from other studies.

  • Many study participants were surprised by the emissions associated with the bus and were under the impression that bus emissions were much lower. We speculate that people generally perceive mass transit (including buses) as having significantly lower emissions than gasoline-fueled vehicles, which may explain the misconception about bus emissions. This finding supports that more community education about the relative emissions of different transportation modes, particularly buses, is needed.

  • The cheeseburger equivalence as an emissions factor appeared to resonate with many participants. We suggest that the inclusion of the cheeseburger factor in emissions calculators may serve as a useful means to engage and educate populations that have historically been less likely to use emissions calculators (i.e., less environmentally motivated demographics).

  • There may be more intuitive equivalency units than a cheeseburger (e.g., apples.). Furthermore, different equivalency units may resonate more with different populations, and be more intuitive as a function of temporal and geographic scale. To address this, future renditions of the emissions calculator could include a variety of choices for functional units, possibly made available through a dropdown menu. Also, a future survey can identify what choices would be perceived as effective. Furthermore, clarification about the negative or positive connotation associated with emissions equivalency units should be included in future calculators (e.g., clarify that more cheeseburgers have a negative connotation).

  • There is demonstrated interest in emissions calculators that use creative emissions equivalency factors. For instance, Google Flights shows airplane emissions on a tree-equivalent basis. The U.S. EPA’s GHG equivalency calculator also shows emissions on a trees-equivalent basis (i.e., tree seedlings grown for 10 years, acres of U.S. forests preserved from conversion to cropland in one year, and acres of U.S. forests in one year). We interpret this as an indication that the interest in these types of calculators represents an underlying belief or assumption that providing emission information can effectively induce positive changes in human travel behavior. However, the efficacy of these calculators in persuading change in human travel behavior has not been evaluated. Our study aims to address this knowledge gap, with a focus on travel behavior in Madison, Wisconsin.

  • While the number of participants in the study is fairly small (49 participants), we can still glean insight from the data. Participants were asked to use the emissions calculator and answer a series of follow-up questions at least 20 times. Therefore, we have collected data for at least 20 trips for each user that completed the study, which provides a compelling dataset with a minimal amount of longitudinal data. A larger study involving more participants and geographic areas is desired in the future for generalizability.

There are different limitations in this study that we detail here. While this study provided insight into the emissions calculator users’ attitudes, beliefs, behaviors, and changes over the duration of the study, the results remain limited and in need of further supplementation. The sample size was relatively small, and the study was limited by location and timeframe. Similar study efforts from different geographical regions and cities, and with more participants, may provide a more diverse set of results. Also, extending the study period for a longer timeframe than three months may capture important longitudinal data that was not accounted for in this study, particularly with respect to weather. As for the calculator design, we did not consider trip chaining or mixed modes (i.e., trips with multiple segmented modes used). Updates on the calculator design could consider the addition of such capabilities to account for more travel scenarios.

While study participants provided feedback that the cheeseburger equivalency unit was useful, there is a potential ambiguity about the negative or positive connotation associated with cheeseburger consumption. Therefore, calculator users may have had different interpretations of the calculator output even when working with the same information (units of cheeseburgers). Future calculators should clarify in text the positive or negative connotation associated with greater emissions equivalency values, no matter the unit.

Additional opportunities to build on this research could explore how goal setting incorporated into emissions calculators may impact users’ travel behavior. For instance, is travel behavior impacted when a weekly emissions goal is set by a calculator user? Similarly, is the travel behavior of calculator users affected when the emissions results are shared amongst a calculator user community versus individual users? A community-based calculator may offer a line of accountability or peer support and encouragement that influences emissions calculator users’ travel behavior. Finally, future studies may include how the results can be used to inform transport policy measures (e.g., improved public transport and subsidies of renewable fuel).