1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 The road to productivity: An analysis of commuters’ punctuality and energy levels at work or school Charis Loong School of Urban Planning McGill University* E-mail: charis.loong@mail.mcgill.ca Dea van Lierop School of Urban Planning McGill University* E-mail: dea.vanlierop@mail.mcgill.ca Ahmed El-Geneidy, Corresponding Author School of Urban Planning McGill University* E-mail: ahmed.elgeneidy@mcgill.ca * McGill University Suite 400, 815 Sherbrooke St. W. Montreal, Quebec, H3A 2K6 Canada Tel.: 514-398-8741 Fax: 514-398-8376 Word count: 5495 words in text + 6 tables/figures = 6995 July 2015 Paper accepted for presentation at Transportation Research Board 95th Annual Meeting (January 10–14, 2016) in Washington, D.C. For citation please use: Loong, C., van Lierop, D., & El-Geneidy A. (2016). The road to productivity: An analysis of commuters’ punctuality and energy levels at work or school. Paper to be presented at the 95th Annual Meeting of the Transportation Research Board, Washington D.C., USA. 2 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ABSTRACT The strain of daily commute can negatively impact work productivity. This study differentiates how various modes of commuting influence productivity at work and school. The data for this study come from the 2013 McGill Commuter Survey, a university-wide survey in which students, staff and faculty described their typical commuting experience to McGill University, located in Montreal, Canada. Ten logistic regressions are used to determine the factors that impact 1) a commuter’s feeling of being energized when he or she arrives at work or school and 2) his or her punctuality. Our results show that weather conditions and mode of transportation have the greatest influence on energy at work and punctuality of individuals. The models indicate that driving is less conducive to productivity than taking the metro, and that bus riders experience the most energydraining commute. On the other hand, cyclists have the greatest likelihood of feeling energized and are the least likely to experience negative impacts on their punctuality. With these findings in mind, transit agencies need to find ways to improve the experience for those using the bus, and schools and employers need to promote the habit of commuting by bicycle in order to enhance academic performance and work productivity. Keywords: Commute, Productivity, Energy, Punctuality, Transit, Bicycles 3 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 INTRODUCTION Commuting is without a doubt a necessary part of many people’s daily routine. However, the strain associated with commuting can have a negative impact on work productivity and school performance. Long travel distances, in particular, contribute to an individual’s level of stress and lack of energy (1-3), which lead to further consequences of unproductivity at work and school (46). A new Canadian study has shown that 40% of employees have fallen asleep at work, and 74% of young adults (between the ages of 18-24) have fallen asleep during a class (7). Furthermore, the same study found that 26% of the survey respondents had taken a sick day only to catch up on sleep. This is estimated to be the equivalent of 3.9 million sick days or 750 million dollars in lost productivity. Therefore, it is critical to understand the relationship between commuting and productivity. The objective of this paper is to investigate how an individual’s commute affects his or her potential productivity at work or at school by assessing what influences an individual’s 1) feeling of being energized and 2) punctuality. The study uses data from a university-wide travel behavior survey conducted during the spring of 2013 in which students, staff and faculty described their typical commuting experiences to McGill University, located in downtown Montreal, Canada. We hypothesize that individuals who commute using active transportation feel the most energized when they arrive at their destination, while public transit users are the least energized and the least punctual due to the nature of the commuting mode. The paper begins with a review of the existing literature about the impact of commuting on an individual’s feeling of being energized and punctuality. It then presents the data used for the study, and describes the results of a series of logistics regression analyses to determine the factors of a commute that affect productivity. Finally, the paper concludes with a discussion of the results and proposes suggestions for future transportation studies and policy recommendations. LITERATURE REVIEW The Impact of Commuting on an Individual’s Feeling of Being Energized Commuting can be a tiring experience (8-10). Transportation researchers have typically related fatigue as an effect of commuting stress, where higher stress levels are correlated with exhaustion (2; 11). Few researchers, however, have specifically examined the factors that influence how energized a person feels after a commute. Those who have studied the relationship between commuting and fatigue suggest that attitudes toward congestion and time use, as well as individuals’ lifestyle preferences and socio-demographics influence a person’s energy level (2; 12-15). Trip characteristics such as travel mode, duration and distance are also known to be important determinants of how tired an individual feels after a commute (1; 14; 16-18). Kluger (1) reported that longer car commutes were positively correlated with fatigue, and in their analysis of the 2007-2008 French National Travel Survey, Mokhtarian et al. (2) found that people with longer commutes felt more tired than others. It is important to note that some researchers differentiate between mental energy and physical energy. For example, Gatersleben and Uzzell (19) found that bicycle trips are the most mentally stimulating, while walking trips are the most relaxing for commuters. On the other hand, Mokhtarian et al. (2) suggested that those who travel by active transportation would be more physically tired, and those who commute by public transportation or by driving would be the most mentally tired. Understanding how each mode affects the physical and mental energy of commuters is important to analyze the productive capacity of employees and students. For 4 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 example, if an employee works in a labor-intensive job, he or she may consider using transportation modes that are less physically draining. The Impact of Commuting on an Individual’s Punctuality It has been well established and understood that commuting, due to its potential unpredictability, has an impact on punctuality (1; 20). More specifically, the greater the commuting distance the more likely that an individual could be late (21). In addition, a previous study has shown that weather plays a role in determining when a person arrives at work (22). Surprisingly, however, to the authors’ knowledge, no previous studies have examined how the impact of commuting influences punctuality across different modes. METHODOLOGY Survey The data used for this study are derived from the 2013 McGill Commuter Survey, an online commuter survey conducted during March and April 2013, targeting McGill University students, staff and faculty. In total, 20,851 survey invitations were sent to randomly selected members of the McGill University community. Respondents had a window of thirty-five days to complete the online survey, and prizes were offered as incentives for participation. The survey had a response rate of 31.7%, which is comparable to previous studies such as those by Redmond and Mokhtarian (23) with a 25% response rate, and Whalen, Páez and Carrasco (24) with a 22% response rate. After cleaning the database by removing incomplete and unreasonable responses, 5,599 records were retained. The survey recorded the respondents’ typical commute from their home location to their destination within the two McGill University campuses for a cold and snowy day, and likewise for a warm and dry day. The respondents answered detailed questions regarding each aspect of their daily commute, including duration, satisfaction with service quality, and mode. The survey also collected information about the respondents’ socio-demographic information, travel preferences, and personal attitudes toward the commute (25). Study Samples This study focuses on individuals who travelled to McGill University’s downtown campus by walking, cycling, driving or transit (bus and metro). The decision to concentrate only on commuters travelling to McGill University’s downtown campus is based on the fact that there are stark differences between the experiences of travelling to McGill University’s suburban Macdonald campus compared to McGill University’s downtown campus, which is located in the middle of the city. Using a 5-point Likert scale, where “1” = strongly disagree and “5” = strongly agree, survey respondents reported their level of accordance with the statements: 1) “I feel energized when I arrive at McGill” and 2) “My commute to McGill negatively impacts my punctuality / attendance / working hours”. Since these were based on self-reported answers, it is debatable as to whether one person’s strong agreement with “I feel energized when I arrive at McGill” can be interpreted as that he or she truly feels more energized than another person who only somewhat agreed with the same statement. Hence, for each respective statement, ordinal responses were transformed into binary variables by recoding “1” and “2” as “no”, “4” and “5” as “yes”, while all “3” were discarded. Because the responses were recoded as binaries, two sample sizes were calculated. As a result, the final sample size for the energy question is 1,579 observations, and for the punctuality question, 2,030 observations. 5 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Respondents are classified by their main mode of transportation; for example, those who used public transit for at least one leg of their trip are identified as transit users. The study does not include commuters who carpooled as car passengers, made use of the commuter train or rode the private university shuttle bus, which offers transportation service between the two McGill University campuses due to the small number of observations. Car passengers were not grouped with drivers as universal car users because the commute of a car passenger requires a different level of energy exertion than that of a driver. Additionally, those who claimed to drive, but did not possess a driver’s license were also eliminated from this study. The survey asked respondents to rate their satisfaction of various aspects of their trip using a 5-point Likert scale. To ensure that the impact of mode-specific attributes can be evaluated accordingly, records of respondents who did not provide an answer or stated that they had no opinion were removed. Finally, it is important to note that each respondent reported details of their typical commute for two weather conditions: 1) a cold, snowy day and 2) a warm and dry day. Consequently, one of the two weather conditions was randomly assigned so that each respondent was associated only with a single weather condition in the data. Tables 1 and 2 summarize the population samples by mode of transportation and present the independent variables that will be tested across all modes for energy and punctuality respectively in the following section of the paper. Despite the difference in sample sizes for the two models (energy and punctuality), the population samples are similar as can be seen when comparing Tables 1 and 2. Both samples are composed of approximately 50% students, 30% staff and 20% faculty, while the mode split is approximately 20% driving, 40% public transit, 25% walking and 15% cycling. A brief assessment of the two population samples reveals that drivers tend to be older (mean age of approximately 45 years) and budget the most additional time for their commute. Transit users have the longest commute (mean of approximately 41 minutes) while pedestrians have the shortest commute (mean of less than 20 minutes). Pedestrians have the highest proportion of students and hence are younger (mean age of 31 years). Together with the cyclists, they place the highest importance of living in proximity to the university and not having to drive. Cyclists also have a slightly higher life satisfaction (7.9/10) than users of other modes (7.5/10). 6 Loong, van Lierop, El-Geneidy 1 TABLE 1 Summary Statistics for Energy at Work– Mean of Variables GENERAL 1579 DRIVE 262 TRANSIT 683 WALK 395 CYCLE 239 I feel energized when I arrive at McGill. 0.54 0.44 0.41 0.59 0.94 Warm, dry day 0.51 0.48 0.40 0.49 0.92 Duration (minutes) Additional budgeted time (minutes) I use my commute time productively (1-5) PERSONAL ATTRIBUTES Age Male Student Staff Faculty Life satisfaction (1-10) HOME SELECTION Importance of my residence being in… ...proximity to McGill (1-5) ... proximity to public transit (1-5) ... a location where I wouldn't have to drive (1-5) MODE USED Driving Transit Walking Cycling 32.41 11.70 3.32 35.85 18.70 3.14 41.34 14.85 3.42 19.93 5.60 3.20 23.75 5.13 3.44 36.35 0.39 0.47 0.31 0.22 7.55 44.66 0.40 0.21 0.36 0.43 7.70 36.04 0.34 0.47 0.39 0.14 7.37 31.97 0.40 0.65 0.15 0.20 7.55 35.33 0.53 0.50 0.25 0.25 7.89 3.50 4.07 3.79 2.90 3.39 2.60 3.09 4.37 3.71 4.40 3.86 4.36 3.81 4.33 4.34 0.17 0.43 0.25 0.15 na na na na na na na na na na na na na na na na Sample size ENERGY WEATHER TIME 2 na “not applicable” 7 Loong, van Lierop, El-Geneidy 1 TABLE 2 Summary Statistics for Punctuality – Mean of Variables Sample size PUNCTUALITY My commute to McGill negatively impacts my punctuality / attendance / working hours. WEATHER Warm, dry day TIME Duration (minutes) Additional budgeted time (minutes) I use my commute time productively (1-5) PERSONAL ATTRIBUTES Age Male Student Staff Faculty Life satisfaction (1-10) HOME SELECTION Importance of my residence being in… ...proximity to McGill (1-5) ... proximity to public transit (1-5) ... a location where I wouldn't have to drive (1-5) MODE USED Driving Transit Walking Cycling 2 GENERAL 2030 DRIVE 380 TRANSIT 896 WALK 501 CYCLE 253 0.32 0.34 0.42 0.24 0.08 0.51 0.47 0.40 0.51 0.92 32.08 11.31 3.30 33.99 17.26 3.13 41.04 13.94 3.42 19.36 5.46 3.18 22.65 4.66 3.35 37.29 0.42 0.45 0.31 0.24 7.58 46.48 0.47 0.18 0.34 0.48 7.80 37.07 0.36 0.44 0.39 0.17 7.44 31.72 0.41 0.65 0.16 0.19 7.50 35.30 0.53 0.50 0.24 0.26 7.86 3.51 4.06 3.73 3.01 3.45 2.57 3.12 4.37 3.73 4.41 3.86 4.31 3.84 4.32 4.34 0.19 0.44 0.25 0.12 na na na na na na na na na na na na na na na na na “not applicable” 8 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 Figures 1 and 2 respectively show the proportion of respondents who feel energized when they arrive to McGill and those whose commute negatively impacts their punctuality. In general, travel mode and weather conditions are significant determinants of a commuter’s productivity, and this observation is confirmed by a series of t-tests and chi-square tests. More specifically, users of active transportation have higher rates of feeling energized at McGill and are less likely to be late for work. For instance, on a typical warm and dry day, 95% of cyclists reported that they feel energized when they arrived at McGill, and only 5% experienced problems with punctuality. This is in contrast to transit users, of which only 59% felt energized when they arrived on a typical warm and dry day, and 26% reported that they arrived late. FIGURE 1 Proportion of commuters who feel energized when they arrive McGill classified by mode and weather. 9 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 FIGURE 2 Proportion of commuters whose commute negatively impacts their punctuality, attendance or working hours classified by mode and by weather. Method Using binary logistic regressions, this study sets out to determine which factors influence a commuter’s potential productivity, as expressed through his or her energy level at work as well as punctuality. Based on an evaluation of the independent variables discussed in literature, the models retain variables that proved to be theoretically relevant and consistently statistically significant in the final results. The models also include controlling factors, such as home selection variables, to enable appropriate interpretation of the results. The decision on whether to keep or drop a variable that was not significant in a model was based on its effect on the model’s Log-likelihood and the changes that occurred in the other variables. General models and mode-specific models were developed to improve the understanding of how each mode and specific aspects of different modes influence a commuter’s productivity. The general models consist of universal variables as well as dummy variables to indicate the mode used. The universal variables presented in the general models are also found in the mode-specific models, which contain additional variables specific to a particular mode that is tested. These modespecific variables are generally related to the satisfaction of the different aspects of the modes used. For example, a respondent who typically rides a bus and takes the metro during his or her commute would answer questions regarding his or her satisfaction for both the bus and the metro. Yet, someone who only takes the bus would only rate his or her satisfaction with the bus. In order to analyze transit users as one group regardless of how many transit modes they use, the average satisfaction for the different modes was generated as a proxy for their overall satisfaction with public transit. In addition, variables were developed to indicate the number of buses and metro lines used by each commuter. 10 Loong, van Lierop, El-Geneidy 1 2 3 4 5 RESULTS AND DISCUSSION Tables 3 and 4 present the results of the binary logistic regression analyses using odds ratio (OR). The log-likelihood values are reported for each model, and the Likelihood Ratio test reveal that the mode-specific models have better data fit than the general models. 11 Loong, van Lierop, El-Geneidy 1 TABLE 3 Logistic Regression Results for Energy at Work “I feel energized when I arrive at McGill.” GENERAL WEATHER A typical warm, dry day TIME Additional budgeted time (minutes) Additional budgeted time (x2) I use my commute time productively (1-5) DRIVE TRANSIT WALK CYCLE OR P > z OR P > z OR P > z OR P > z OR P > z 3.911 0.00 3.144 0.00 2.985 0.00 3.657 0.00 24.586 0.00 0.928 1.001 0.00 0.00 0.950 1.001 0.01 0.01 0.917 1.001 0.00 0.00 0.884 1.005 0.02 0.04 – – – – 1.352 0.00 1.097 0.50 1.224 0.01 1.632 0.00 – – 0.602 1.141 0.00 0.00 1.300 1.248 0.51 0.06 0.715 1.146 0.14 0.01 0.316 0.965 0.00 0.70 – – – – – – – – 1.082 – 0.49 – – – – – 1.093 – 0.46 – – 0.270 – 0.02 1.129 0.01 na na 1.170 0.03 – – – – na na 2.132 0.00 – – – – – – na na 1.526 0.01 na na – – – – na na na na 1.304 0.02 na na na na na na na na 1.296 0.01 na na na na na na – – – – 1.751 0.00 1.688 0.07 PERSONAL ATTRIBUTES Student Life satisfaction (1-10) HOME SELECTION Importance of my residence being in… ...proximity to McGill (1-5) ... proximity to public transit (1-5) ... a location where I wouldn't have to drive (1-5) SATISFACTION WITH MODE I am satisfied with the length of travel time (1-5) I feel safe from traffic when I travel (15) I am satisfied with how long it takes me to reach my bus stop or metro (1-5) The waiting time for the bus or metro is reasonable (1-5) I feel comfortable when I travel (1-5) 12 Loong, van Lierop, El-Geneidy GENERAL BIXI I have had a Bixi membership in the past year DRIVE TRANSIT WALK CYCLE OR P > z OR P > z OR P > z OR P > z OR P > z na na na na na na na na 0.275 0.04 0.250 0.266 na 0.390 na 0.325 v 0.00 0.00 na 0.00 na 0.00 na na na na na na na na na na na na na na na na na 0.541 na 0.701 na na na na 0.00 na 0.13 na na na na na na na na na na na na na na na na na na na na na na na na na na na na na na MODE(S) USED Driving Bus Number of bus routes Metro Number of metro lines Walking Compared to: Cycling n Pseudo R2 LL AIC BIC 1 2 1579 0.26 -806.11 1636.21 1700.59 262 0.30 -124.79 269.58 305.27 683 0.23 -356.34 736.68 791.00 – “found to be insignificant and removed”, na “not applicable”, v “comparison variable” 395 0.27 -195.68 409.37 445.18 239 0.21 -44.37 98.75 116.13 13 Loong, van Lierop, El-Geneidy 1 TABLE 4 Logistic Regression Results for Punctuality “My commute to McGill negatively impacts my punctuality / attendance / working hours.” GENERAL WEATHER DRIVE TRANSIT WALK CYCLE OR P > z OR P > z OR P > z OR P > z OR P > z A typical warm, dry day 0.312 0.00 0.494 0.01 0.370 0.00 0.216 0.00 0.040 0.00 TIME Additional budgeted time (minutes) Additional budgeted time (x2) 1.085 0.999 0.00 0.00 1.055 0.999 0.00 0.01 1.088 0.999 0.00 0.00 1.101 0.998 0.00 0.02 1.059 – 0.01 – Student 2.358 0.00 2.355 0.00 2.139 0.00 3.090 0.00 – – HOME SELECTION Importance of my residence being in… ...proximity to McGill (1-5) ... proximity to public transit (1-5) ... a location where I wouldn't have to drive (1-5) 1.124 – 0.01 – 1.149 – 0.13 – 1.192 – 0.00 – – – – – – 1.958 – 0.10 – – – – – – 1.161 0.08 – – na na 0.559 0.00 – – 0.745 0.07 – – na na na na 0.841 0.05 na na na na na na na na 0.685 0.00 na na na na na na – – – – 0.661 0.01 – – PERSONAL ATTRIBUTES SATISFACTION WITH MODE I am satisfied with the length of travel time (1-5) I am satisfied with how long it takes me to reach my bus stop or metro (1-5) The waiting time for the bus or metro is reasonable (1-5) I feel comfortable when you travel (1-5) 14 Loong, van Lierop, El-Geneidy GENERAL MODE(S) USED Driving Bus Number of bus routes Metro Number of metro lines Walking Compared to: Cycling n Pseudo R2 LL AIC BIC 1 DRIVE TRANSIT CYCLE OR P > z OR P > z OR P > z OR P > z OR P > z 2.178 3.148 na 1.496 na 1.388 v 0.00 0.00 na 0.01 na 0.10 na na na na na na na na na na na na na na na na na 2.116 na 1.198 na na na na 0.00 na 0.37 na na na na na na na na na na na na na na na na na na na na na na na na na na na na na na 2030 0.20 -1015.09 2050.17 2106.33 380 0.18 -200.01 414.03 441.03 896 0.21 -480.24 980.48 1028.46 – “found to be insignificant and removed”, na “not applicable”, v “comparison variable” WALK 501 0.24 -208.62 433.23 466.96 253 0.22 -54.47 116.94 131.07 15 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 General Model Since the factors affecting both energy at work and punctuality are very similar, this section discusses the results of both models simultaneously while highlighting the relevant differences. First, dummy variables are included for each mode to determine how mode choice influences productivity. Although in this study individual modes are generally mutually exclusive since respondents are classified according to their main mode of transportation, it is possible for a transit rider to use both the bus and metro. In previous studies, bus and metro riders have been grouped under the single category of public transit users (2; 26). However, the findings from this study suggest that the commuting experience of individuals using different transit modes impact their productivity in different ways. Hence, there is a distinction between bus and metro users in the general models. According to the results, cyclists are predicted to be considerably more productive than users of other modes. Compared to cyclists, other commuters are 2.56 to 4.00 times less likely to feel energized, and 1.39 to 3.15 times more likely to be late due to the commute. In descending order of probability of productivity are cyclists, pedestrians, metro users, drivers, and then bus users. However, given the possibility of using both bus and metro during a commute, transit users who use both the bus and metro are predicted to fare the worst in terms of productivity. Secondly, other aspects that influence a commuter’s energy level include additional budgeted time that is allotted for the trip and how productive an individual felt during the commute. Additional budgeted time, however, is the only time-related factor that significantly influences a commuter’s punctuality. Travel duration and distance were tested, but not incorporated in the final models. The reasons are as follow: firstly, travel distance correlates strongly with travel duration (Pearson’s correlation coefficient = 0.70), and as a result, only one of the two variables should be included. Travel duration, however, serves as a better representation of the actual commute than distance because it accounts for different travel speeds, as well as any stops that the commuter makes or delays that occur along the way (2; 27-29). Additionally, the travel time for each commute was reported by the individual and should therefore be interpreted as their perception of travel duration. Nevertheless, this variable is not statistically significant at α = 0.90 and is therefore removed from the models despite having a heavy presence in the literature. On the other hand, additional budgeted time is shown to have a non-linear relationship with productivity. Since planning extra time for a commute is usually a response to unpredictability in the length of travel time, it is probable that it is not travel duration itself that affects productivity, but rather its inconsistency. However, it is important to note that variability in travel time usually increases for longer durations. Moreover, when planning for additional travel time, the more extra time allocated up to a certain point, the lower the predicted productivity. After surpassing a certain point, the reverse holds true. This nonlinear relationship may indicate that some people are not allocating enough additional time for their commute, and thereby negatively affecting their productivity. Thirdly, weather plays a significant role in affecting an individual’s productivity. More precisely, an individual is predicted to be 3.91 times more likely to feel energized at work on a warm and dry day than on a cold and snowy day. Likewise, commuting on a warm and dry day translates into 8.22 times less chance of being late for work. These effects of the weather can be interpreted as an indirect result of higher stability of the transportation systems and consequently, lower energy exertion required on the part of the individual. Next, the models account for residential self-selection variables by testing various home selection criteria, including being in proximity to McGill and being in a location where driving is not necessary. According to the results, those who valued the importance of not having to drive are 1.13 times more likely to be energized after the commute, while those who considered it 16 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 important to live in proximity to the university are 1.12 times more prone to being late. Although the latter part of the previous statement may not seem intuitive, it is possible that those who live closer to the university campuses may be underestimating their commute time. Nonetheless, it is also probable that those who are aware of their tendencies to be late consider it more important to be living near the university. Finally, with regard to personal attributes, students, when compared to staff and faculty, are estimated to be 1.66 times less likely to feel energized and 2.36 times more prone to being late because of their commute. This finding agrees with the theory that there are significant differences between the behavior of students and those of workers (30; 31). These dissimilarities can be attributed to differences in attitudes, lifestyle, responsibilities and stages of life. Yet, these same findings contrasts that of Mokhtarian et al. (2), in which students found their commute to be less tiring than those of workers. A possible reason for the difference in findings is that this study focuses on students at the university level, whereas Mokhtarian et al. (2) examined students as young as six years of age. Other socio-demographic variables were tested in the models, of which age, income and the number of children living in the same household were significant, while gender was not significant. However, it was found that the indication of student status explained more variation in data than the other socio-demographic variables combined. When placed in the same model, only the student status variable was statistically significant. This is because being a student often also implies other personal characteristics such as younger age, having fewer children and having a lower income. For instance, the average age of a student is 25 years compared to the average age of a non-student, which is 45 years. Moreover, the median income category for students is from $0 to $19,999, while that of faculty and staff is $40,000 to $59,999. Lastly, individuals with higher life satisfaction are predicted to be 1.14 times more likely of being energized at work, but life satisfaction does not have a significant impact on an individual’s punctuality. Mode-Specific Models When interpreting the results of the mode-specific models, it is critical to understand that even though the same variables may appear across different mode-specific models, there are important distinctions. For instance, a comfortable experience for a driver is different than a comfortable experience for a public transit user. More specifically, a driver may be concerned with the congestion he or she is facing while a public transit user may desire more room and seating. Drivers The models for those who commute by driving predict that increased satisfaction of travel duration leads to 2.13 times higher likelihood of drivers feeling energized at work and 1.79 times less likely of being late to work. Lower satisfaction of travel time may be indicative of increased commuting stress (27) as well as higher levels of inconsistency in travel duration. In addition, a greater sense of safety from traffic is estimated to increase an individual’s probability of feeling energized at work by 1.53 times. Driving on a warm and dry day, instead of a cold and snowy day improves the chance of the commuter feeling energized at work by 3.14 times and being punctual by 2.02 times. This is expected, as road conditions in Montreal during winter can become quite challenging due to the presence of snow and ice. There are, however, few personal attributes, that significantly determine the potential productivity of someone who drives to work. This is likely due to a lower variation in socio-demographics of those who drive. 17 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Transit Users According to the transit models, commuting by bus is less conducive to a person’s productivity than commuting by metro. For every bus route an individual takes, he or she is predicted to be 1.85 times less likely to be energized, and 2.12 times more likely to be late. Meanwhile, for every metro line a person takes, he or she is estimated to be 1.43 times less likely to feel energized at work, and 1.20 more times to be late. Transit users are more sensitive to the time it takes to reach their bus stop or metro station and how much time they have to wait for their transit service, than the actual trip duration. Increased satisfaction of the time it takes to reach the bus stop or metro station is expected to result in 1.30 times greater probability of an individual feeling energized and 1.19 times greater probability of being punctual to work. Furthermore, greater satisfaction of waiting time is predicted to enhance an individual’s likelihood of feeling energized at work by 1.30 times and his or her prospect of being on time by 1.46 times. In other words, waiting time and the time it takes an individual to reach his or her desired transit service have similar levels of influence on an individual’s energy. However, waiting time is a greater determinant of a transit user’s likeliness to be punctual. Transit agencies, therefore, should make an effort to improve both service accessibility and reliability as such improvements are expected to have an impact on the productivity of the individuals they are serving. Pedestrians For pedestrians, both their comfort level during the commute as well as their satisfaction of travel time affect whether they feel energized at work. Comfort, although it has greater influence than satisfaction of travel time on an individual’s energy, does not significantly impact punctuality. Greater comfort during the commute is expected to result in 1.75 times increased chance of being energized at work, but increased satisfaction of travel time by only 1.34 times. Nevertheless, greater satisfaction of travel duration is predicted to improve the prospect of an individual arriving on time by 1.51 times. Cyclists The results for cyclists contrast those found for the other modes. First, weather conditions have an extreme impact on a cyclist’s productivity; cycling on a warm and dry time instead of a cold and snowy day increases the probability of feeling energized at work by 24.59 times and punctuality by 25.00 times. This may be due to a bias in the survey, as it was conducted at the end of a long winter. In Montreal, winter conditions extend until at least the end of March. Since the survey inquired about commuting on a typical warm, dry day, cyclists who were eager for nicer weather may have over-predicted their energy levels and underestimated how much the commute impacts their punctuality. Second, cyclists who deemed it important to live near public transit are less likely to be productive at work or at school; greater importance of living in proximity to public transit is predicted to decrease the likelihood of feeling energized at work by 3.70 times and decrease the prospect of being punctual by 1.96 times. A possible explanation is that those who live near public transit may live in areas with busy traffic. Hence, more energy and effort would be required to navigate the roads safely by bicycle. Third, holders of a Bixi membership are predicted to be 3.64 times less energized than other cyclists. There could be a couple of possible reasons for this finding: Bixi is a Montreal bicycle-sharing system, which operates annually from April to November. Cyclists who subscribe to a Bixi membership are most likely to be seasonal cyclists. Additionally, Bixi bicycles are also heavier and more difficult to maneuver than most personal bicycles, requiring more energy exertion. Based on a comparison of the various models, cyclists 18 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 seem to be a unique population of commuters. Hence, a deeper investigation into the factors that affect the potential productivity of cyclists should be conducted. CONCLUSION The findings of this study suggest that similar factors are significant in influencing whether a commuter feels energized upon arrival at his or her destination, and whether a commute negatively impacts his or her punctuality. The results support the hypotheses and demonstrate that the impact of commuting on productivity is significantly determined by the mode used and the weather conditions. Furthermore, this study indicates that driving is less conducive to productivity than taking the metro, and that bus users experience the most energy-draining commutes. Time is an important factor, especially for pedestrians, drivers and transit users. However, it is not trip duration, but rather the additional time that a commuter budgets which is a significant determinant of an individual’s probability of feeling energized at work and being punctual. Greater satisfaction with travel time leads to a higher probability of punctuality for pedestrians and drivers alike, but also increases the chance of being energized at work for drivers. In addition, the productivity of transit users is sensitive to the time it takes them to reach their bus stop or metro station, as well as their waiting time, indicating that transit agencies should prioritize the improvement of service accessibility and reliability to provide a better commuting experience for their customers. Results from this study indicate that being productive while commuting increases the likelihood of being energized at work. However, this relationship is not well understood and further investigation is required. Results also suggest that cyclists have the greatest likelihood of feeling energized and are the least likely to experience negative impacts on their punctuality. However, more research needs to be conducted to better understand how the conditions of cyclists’ commutes affect their productivity when they arrive at their work or school destination. Future studies should also distinguish between physical and mental fatigue, as it will lead to a better understanding of how each mode affects the physical and mental energy of commuters and ultimately, the productive capacity of employees and students. Finally, businesses and policy makers should encourage and promote the use of a bicycle for work and school commutes as the results of this study confirm that compared to other transportation modes, cycling leads to the highest probability of being productive. ACKNOWLEDGEMENTS We would like to thank Kathleen Ng and Brian Karasick, from the McGill Office of Sustainability, for their feedback and guidance at various stages of this project. We would also like to thank Daniel Schwartz from IT Customer Services for his assistance in developing the online survey and managing the distribution of the survey to the McGill community. Thanks for Geneviève Boisjoly for her constructive feedback on the manuscript. Finally, we gratefully acknowledge the financial support received from the Social Sciences and Humanities Research Council (SSHRC) and the Natural Sciences and Engineering Research Council of Canada (NSERC). 19 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 REFERENCES [1] Kluger, A. Commute variability and strain. Journal of Organizational Behaviour, Vol. 19, No. 2, 1998, pp. 147–165. [2] Mokhtarian, P., F. Papon, M. Goulard, and M. Diana. What makes travel pleasant and/or tiring? An investigation based on the French National Travel Survey. Transportation, 2014, pp. 1-32. [3] Waddell, J. Are long commutes killing your employee productivity? http://www.business2community.com/human-resources/long-commutes-killing-employeeproductivity-0848737. Accessed June 10, 2015. [4] Taris, T., and W. Schaufeli. Individual well-being and performance at work.In Well-being and performance at work: The role of context, Psychology Press, New York, 2014. pp. 15-34. [5] Gnoth, J., A. Zins, R. Lengmueller, and C. Boshoff. Emotions, mood, flow and motivations to travel. Journal of Travel & Tourism Marketing, Vol. 9, No. 3, 2000, pp. 23-34. [6] Adecco Canada. How a grueling commute can quash productivity. http://leadblog.ca/2013/02/08/commute/. Accessed June 10, 2015. [7] Mediabrands and Reprise. Sleep centre study: Summary report, IPG Mediabrands, 2015. [8] Evans, G., R. Wener, and D. Phillips. The morning rush hour: Predictability and commuter stress. Environment and Behavior, Vol. 34, No. 4, 2002, pp. 521-530. [9] Koslowsky, M., A. Kluger, and M. Reich. Commuting stress: Causes, effects, and methods of coping. Plenum Press, New York, 1995. [10] Stutzer, A., and B. S. Frey. Stress that doesn't pay: The commuting paradox. Scandinavian Journal of Economics, Vol. 110, No. 2, 2008, pp. 339-366. [11] Barden, K., and J. Lucas. Perceived commute strain, negative physical symptoms, and exhaustion in employees who commute. Psi Chi Journal of Undergraduate Research, Vol. 8, No. 4, 2003, pp. 149-152. [12] Bagley, M., and P. Mokhtarian. The impact of residential neighborhood type on travel behavior: A structural equations modeling approach. The Annals of Regional Science, Vol. 36, No. 2, 2002, pp. 279-297. [13] Koslowsky, M., A. Aizer, and M. Krausz. Stressor and personal variables in the commuting experience. International Journal of Manpower, Vol. 17, No. 3, 1996, pp. 4-14. [14] Hilbrecht, M. Time needed to travel to work. In Encyclopedia of quality of life and wellbeing research, Springer, The Netherlands, 2014. pp. 6647-6649. [15] Archer, M., R. Paleti, K. Konduri, R. Pendyala, and C. Bhat. Modeling the connection between activity-travel patterns and subjecive well-being. Transportation Research Record, No. 2382, 2013, pp. 102-111. [16] Humphreys, D., A. Goodman, and D. Ogilvie. Associations between active commuting and physical and mental wellbeing. Preventive Medicine, Vol. 57, No. 2, 2013, pp. 135-139. [17] Lyons, G., and K. Chatterjee. A human perspective on the daily commute: Costs, benefits and trade‐offs. Transport Reviews, Vol. 28, No. 2, 2008, pp. 181-198. [18] Hansson, E., K. Mattisson, J. Bjork, P. Ostergren, and K. Jakobsson. Relationship between commuting and health outcomes in a cross-sectional population survey in southern Sweden. BioMed Central Public Health, Vol. 11, No. 1, 2011, p. 834. [19] Gatersleben, B., and D. Uzzell. Affective appraisals of the daily commute: Comparing perceptions of drivers, cyclists, walkers, and users of public transport. Environment and Behavior, Vol. 39, No. 3, 2007, pp. 416-431. 20 Loong, van Lierop, El-Geneidy 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 [20] Nicholson, N., and P. Goodge. The influence of social, organizational and biographical factors on female absence. Journal of Management Studies, Vol. 13, No. 3, 1984, pp. 234-254. [21] Leigh, J., and J. Lust. Determinations of employee tardiness. Work and Occupations, Vol. 15, No. 1, 1988, pp. 78-95. [22] Muesser, R. The weather and other factors influencing employee punctuality. Journal of Applied Psychology, Vol. 37, No. 3, 1953, pp. 329-338. [23] Redmond, L., and P. Mokhtarian. The positive utility of the commute: Modeling ideal commute time and relative desired commute amount. Transportation, Vol. 28, No. 2, 2001, pp. 179-205. [24] Whalen, K., Z. Páez, and J. Carrasco. Mode choice of university students commuting to schooland the role of active travel. Journal of Transport Geography, Vol. 31, 2013, pp. 132-142. [25] Shaw, E., E. St-Louis, K. Manaugh, D. van Lierop, C. Stewart, and A. El-Geneidy. Findings from the 2013 McGill Commuter Survey, McGill University, Montreal, Quebec, 2013. [26] Grimsrud, M., and A. El-Geneidy. Transit to eternal youth: Lifecycle and generational trends in Greater Montreal public transport mode share. Transportation, Vol. 41, No. 1, 2014, pp. 1-19. [27] Legrain, A., N. Eluru, and A. El-Geneidy. Am stressed, must travel: The relationship between mode choice and commuting stress Presented at 14th Meeting of the International Association of Travel Behavior Research (IATBR), London, UK, 2015. [28] St-Louis, E., K. Manaugh, D. van Lierop, and A. El-Geneidy. The happy commuter: A comparison of commuter satisfaction across modes. Transportation Research Part F: Traffic Psychology and Behaviour, Vol. 26, 2014, pp. 160-170. [29] Gordon, P., A. Kumar, and H. Richardson. Gender differences in metropolitan travel behaviour. Regional Studies, Vol. 23, No. 6, 1988, pp. 499-510. [30] Barr, S., and M. Hitt. A comparison of selection decision models in manager versus student samples. Personnel Psychology, Vol. 39, No. 3, 1986, pp. 599-617. [31] Carpenter, J., S. Burks, and E. Verhoogen. Comparing students to workers: The effects of social framing on behavior in distribution games. In Field Experiments in Economics, No. 10, Emerald Group Publishing Limited, 2005.