Javascript required
Skip to content Skip to sidebar Skip to footer
Home / T of the Project Can Be Found in Http://www.ncitec.msstate.edu/wp-content/uploads/2012-25fr.pdf

T of the Project Can Be Found in Http://www.ncitec.msstate.edu/wp-content/uploads/2012-25fr.pdf

1. Introduction

Freeways are crucial elements of transportation systems as they provide travelers with a high level of mobility services. However, freeways in urban areas oft experience severe traffic congestion due to growing mobility demand that exceeds facility capacity, worsening cities' economic competitiveness and quality of life (American Highway Users Brotherhood, 2015). Reducing throughway congestion requires implementing strategies designed to increment facility chapters or reduce transportation demand. To this end, transportation agencies have introduced high occupancy cost (HOT) lanes. HOT lanes typically allow iii-person carpools to apply the lanes for free, which limits the need for the use of the lane. Tolls are and so instituted to let non-carpools to utilize the lane for a fee, which is the notable difference from high occupancy vehicle (HOV) lanes where single occupancy vehicles are restricted from traveling. With proper variable pricing, which adjusts the price to ensure need remains lower than capacity, congestion tin be prevented and travel speeds in HOT lanes can exist maintained (e.g., 45+ mph more than xc% of the time). Hence, traveling on HOT lanes in particular in a form of carpool or transit is an bonny option during elevation hours because of saved time and toll. This aspect suggests that an introduction of HOT lanes can be considered as a measure enhancing the sustainability of transportation systems as they nudge travelers to switch from solo driving to carpooling or transit. The reduced amount of traffic by increasing vehicle occupancy can contribute to energy saving and emissions reduction.

Since California's State Route (SR) 91 Limited Lanes opened in Dec 1995 (the first HOT lane in the US), approximately 20 HOT facilities take been installed (Guensler et al., 2013). In some cases, such as on Atlanta's I-85 corridor, HOT lanes are created from the conversion of HOV lanes. Still, numerous HOT facilities are nether development beyond the county. To meliorate operations of the current facilities and design of future facilities, transportation agencies need to understand the various impacts and travelers' beliefs changes resulting from the introduction of HOT lanes.

In order to enhance the understanding, this study investigates factors affecting drivers' choices on HOT lane use and carpooling in the Atlanta I-85 HOT corridor. Although this research is not the first effort to assess factors that bear on HOT operations in the I-85 corridor, information technology is unique in that information technology utilizes data obtained from a survey administered to commuters who traveled the corridors. Indeed, previous studies utilized marketing (credit study) data for identifying the socio-economic characteristics of drivers traveling on HOT or regular general purpose (GP) lanes (Khoeini & Guensler, 2014a, 2014b; Sheikh, Misra, & Guensler, 2015). Their arroyo is advantageous in obtaining a sizable data set at a depression toll, but one outcome of uncertainty appears to be unavoidable; driver/household information in the marketing data will not always match those of the observed drivers. This study that employs a survey arroyo is expected to avert the upshot as the survey information allow for direct relating drivers' responses to their perception, socio-economical characteristics and trip patterns (eastward.g., commute distance and piece of work showtime time).

Too as the impacts of drivers' socio-demographic and travel characteristics, this study attempts to capture the impacts of two factors: drivers' perception of HOT performance and whether the driver was a carpooler or HOV lane user before HOT implementation. Their impacts accept rarely been examined in the previous studies. It should be also noted that carpool behavior after the installation of a conversion of an HOV lane to a HOT lane has non yet been handled in the previous studies. The findings from this study will add knowledge to the literature on drivers' behavior concerning HOT lane and carpool choices in HOT corridors.

2. Literature review

The existing studies concerning HOT lanes have usually focused on their operational impacts. One of the potential impacts identified is a alter in transit ridership. Previous studies demonstrated that not a modest portion of new bus riders in HOT lane corridors (for example, 23% in Minneapolis and 53% in Miami) were influenced to have transit by HOT lanes, resulting in an increase in transit ridership (Pessaro, Turnbull, & Zimmerman, 2013). However, the bus ridership increment seems not to be always guaranteed by the introduction of HOT lanes. Castrillon, Roell, Khoeini, and Guensler (2014) reported that person throughput for commuter buses remained stable even with an xviii% increase in express bus throughput in the Atlanta I-85 HOT corridor. By comparing the numbers of carpools earlier and after HOT lane implementations, Burris, Alemazkoor, Benz, and Wood (2014) and Goel and Burris (2012) revealed that HOT lanes tended to decrease carpooling.

It should be noted that the impacts can somewhat vary past the characteristics of the HOT facility (e.g., HOT lane capacity) and exogenous factors such as gas prices, as pointed out by Burris et al. (2014). Transportation demand management (TDM) activities (east.g., carpooling, vanpooling, and transit use) and toll exemption policies can also influence travelers' choices, which in turn impact traffic conditions on both managed lanes and regular GP lanes. Pessaro and Buddenbrock (2015) illustrated how such impacts are expected for the I-95 Express Lanes using scenario-based traffic simulation approaches. This aspect requires ongoing inquiry efforts on HOT operations, so that transportation agencies can gain a comprehensive agreement that helps in establishing appropriate HOT design and operation policies that can enhance the sustainability of the facility.

The review reveals that the decision to travel in HOT lanes can be afflicted by drivers' socio-economic characteristics, such as gender, historic period, income, household size, vehicle ownership, and educational activity level (Khoeini & Guensler, 2014a; Sheikh et al., 2015). Khoeini and Guensler (2014b) also explored this issue using vehicle value every bit a proxy for income. They showed that the value of vehicles using HOT lanes is approximately 23% higher than that of vehicles using regular GP lanes in the instance of the Atlanta I-85 HOT corridor. This issue implies that loftier-income people are more probable to use HOT lanes, every bit was also found in the study of the SR 91 Limited Lanes (Li, 2001). The SR 91 written report concluded that household income, vehicle occupancy, commute trip, and age are important predictors of HOT lane use. Meanwhile, the report identified that such factors as gender, trip length, trip frequency and household size and type (family with or without children) are not significant. Edifice on these studies, further enquiry is encouraged to firmly empathise drivers' behavior on HOT lane choices.

three. Study corridor: the Atlanta I-85 HOT lanes

The spatial scope of this study is the Atlanta I-85 HOT lanes. The HOT lanes were installed by converting existing HOV lanes over a xv.5-mile length. The HOV-to-HOT conversion is the beginning to simultaneously introduce tolling while increasing the occupancy requirement (from HOV2+ to HOV3+), but did non add additional lanes (Guensler et al., 2013). The HOV2+ lane still exists just to the south of the HOT lane corridor, extending into Downtown Atlanta. Since the HOT lane opened on October 1, 2011, dynamic tolling varies toll price in response to congestion. Price exempt vehicles include vehicles carrying three or more persons (HOT3+), transit vehicles, emergency vehicles, motorcycles, and alternative fuel vehicles with proper license plates (hybrid vehicles do not qualify). All vehicles must register for a Peach Pass toll tag, even if they are price-exempt, so that corridor activity can exist monitored.

The field survey data collected over the corridor showed that the number of vehicles traveling every bit HOV2+ in the HOT lanes decreased afterward the conversion, from 3,966 to 613 average weekday travelers during the morning pinnacle catamenia and from 3,941 to 697 travelers during the evening peak period. Meanwhile, the number of HOV2+ carpoolers in the regular GP lanes almost doubled (Burris et al., 2014; Guensler et al., 2013). That is, carpools shifted out of the managed lane into the GP lane. Apropos this state of affairs, Guensler, Xu, Sheikh, Li, and Khoeini (2015) identified that the 2-person carpoolers considered the toll cost is too high, perceiving that the amount of saved travel time is not worth the cost. In add-on, some carpoolers complained that information technology is too difficult to go into and out of the express lanes. Overall, morning commute carpooling on the corridor after HOV-to-HOT conversion decreased by more than 30% (Guensler et al., 2013). Identifying those who are likely to carpool in the HOT lanes, volition assist inform hereafter managed lane operational strategies.

4. Survey data

A questionnaire-based survey was designed to explore the behavioral changes of the travelers along the I-85 HOT lane corridor. The first hurdle of the survey was to identify a sample pool, given that HOT lane users and carpoolers tend to constitute only a pocket-size portion of the overall traveling public. Fortunately, the enquiry team had collected about one.5 million license plates of the vehicles traveling the I-85 corridor 1 yr before and one year subsequently the HOV-to-HOT conversion. The collected information allowed the identification of households that retain frequent HOV/HOT users of the corridor (Khoeini & Guensler, 2014b). Based on the database, ten,000 survey targets were selected and questionnaires were mailed out in the grade of an viii-panel folded sheet with a prepaid return envelope. It was later on discovered that an event with the user database resulted in wrong names printed in the survey address, which might have affected the response rate. Numerous surveys that were completed included notes indicating that the name on the form was wrong. Given the potential problem, a second stage involved sending out two,000 additional surveys to households that were not previously targeted in the initial deployment. This second stage survey was designed to cheque if the previous problem (wrong proper noun on the survey class) afflicted the response rate. The research team conducted the mail-out/mail-back survey in November and Dec 2014 and obtained 642 responses amongst the 12,000 target households (a retrieval charge per unit of 5.4%). The response rates were roughly equal during the both stages, indicating that the name errors in the start stage probably did non significantly influence the response rates.

iv.one. Survey questionnaire

The survey questionnaire was composed of 4 general sections, asking:

  1. Master routes and modes for morn commute earlier and after the HOT lane implementation,

  2. Perception of the HOT lane effectiveness on their commute traffic weather condition,

  3. Reasons why the respondent chose to apply or not employ the HOT lanes, or to carpool, subsequently the HOT lane implementation, and

  4. Individual and household socioeconomic/demographic characteristics.

The questions about the lane choices include: use of HOV lanes or regular GP lanes before HOT implementation, and use of HOT lanes or regular GP lanes later on HOT implementation. The perception was measured past a question ("Have the HOT lanes improved your own commute atmospheric condition?") and respondents were requested to answer based on a 5-point rating scale from "definitely yes" to "definitely no." The socioeconomic and demographic questions include age, gender, household income, number of children, number of workers, car ownership, education, work commencement fourth dimension and job locations described by zip code. Based on the chore location, home to work distances were estimated by utilizing the fourth dimension-based shortest path between the cipher codes of their home and work place with the aid from the office of the Google Maps API. As a unique chemical element of this study, the identified information on whether the respondent is a erstwhile HOV user and/or sometime carpooler, the perception and the socioeconomic characteristics were employed to explain commuters' HOT lane and carpool choices.

4.2. Data selection

Self-administered mail-out/mail service-back surveys are susceptible to missing values and inconsistency of answers, requiring a careful information selection procedure. During an initial bank check, it was found that 17 respondents (2.6% of the retrieved 642 replies) did not properly provide their mode and route/lane choices. After excluding the 17 respondents, a table illustrating changes in mode and lane choices (because all the combinations of carpool, drive lone, HOT and regular GP lanes) later on the HOT lane installation was adult equally shown in Table 1. The table, unfortunately, revealed that the travel patterns of 73 respondents are irrelevant to this study equally they did not drive on the freeway either before or subsequently HOT implementation.

Table ane. Respondents' style and route choices before and after HOT installation (before information screening).

Farther data screening procedures took into business relationship whether the choices are multiple (for example, cases in which respondents marked both HOT and regular GP lanes for their usual travel lanes) and whether respondents answered all the questions related to the explanatory variables discussed in the previous section. The multiple choices can be regarded as normal for some drivers, but it is highly circuitous to include such answers in a statistical model. The screening process removed cases with multiple choices or missing values, resulting in a meaning loss of data from 625 to 271 (Tables ane and 2). Concerning the missing values, approximately 150 respondents did not provide any personal data. In full, 58% (371 out of 642) of the retrieved surveys were not usable for choice-based analysis, where all explanatory variables associated with choice demand to be entered into the statistical models.

Tabular array two. Respondents' mode and route choices before and after HOT installation (after data screening).

An try to develop a multinomial logistic regression model to predict post-opening travel for one-time carpoolers considered four choices (drive alone in the regular GP lanes, drive solitary in the HOT lane, carpool in the regular GP lanes, and carpool in the HOT lane), but the results were unsatisfactory. The main reason for the unsatisfactory effect was most likely the pocket-sized sample size. Multinomial regression using a maximum likelihood estimation method usually requires an even larger sample size than ordinal or binary logistic regression (Agresti, 1996). Given this situation, the authors adult binary selection models separately by route (HOT or regular GP lanes) and fashion option (carpool or drive alone) fabricated subsequently the HOT lane began operating. The road and mode choices before the HOT installation were also utilized every bit contained variables.

To minimize the loss of information, the authors conducted divide data selection procedures for HOT lane and carpool pick models. This is considering more samples are likely to be screened out when HOT lane and carpool choices are simultaneously considered. Every bit previously demonstrated, the procedure screened out cases with multiple choices for both HOT and regular GP lanes (also carpool and drive alone) and missing values for the explanatory variables, resulting in 313 and 332 valid cases for HOT lane and carpool pick models, respectively.

Table iii summarizes the selected sample characteristics based on 11 factors including socio-demographic characteristics, commute option, and perception on HOT lane, showing that the two data sets are very similar. This is non surprising, given that the two data sets share 285 identical respondents (91% and 86% of the information sets for the HOT lane and carpool choices respectively). The sample is equanimous of slightly more males than females. More than one-half of the respondents are older than 50 years (about 52%). More than than 60% of the respondents belong to a high-income group, above USD $100,000 per twelvemonth. The income distribution seems to exist reasonable since the sample contains a grouping of HOT lane users who are likely to accept a higher value of time (Khoeini & Guensler, 2014b). More half of the households have children. Single-worker households comprise less than 30% of the sample, with two-worker households being dominant. Almost 60% of households ain multiple cars for commuting. Virtually 75% of the respondents have a Bachelors' degree or higher. Compared to the general population of Metro Atlanta, the sample has higher proportions of males, older people, loftier-income households and educated people. The sample likewise tends to have more children and workers in the household. Interestingly, the number of vehicles for commute is smaller compared to that of the full general population.

Tabular array 3. Demographic and opinion responses in the sample.

It appears that virtually respondents (virtually lxxx%) first their work between 7 and 9 a.thousand. and virtually half of the respondents commute more than 30 miles. In the HOT lane option sample, 26.2% of the respondents replied that they ordinarily used HOV lanes before the HOT lanes opened. In the carpool selection sample, quondam carpoolers occupy 24.4%. With respect to respondent opinion well-nigh whether the HOT lanes take improved their commutes, 58% were negative (definitely no and probably no), nearly 37% were positive (definitely yes, and probably yes), and less than 5% were not sure. Note, however, that these percentages do non control for whether the respondents are or are not regular HOT users. In fact, information technology turns out that users are mostly positive and not-users are generally negative.

The selected respondents' behavioral changes in HOT lane and carpool choices are summarized in Table 4. The table shows that 34% (107 out of 313) of the respondents ordinarily use the HOT lanes, while the remaining 66% of respondents are regular GP lane users. In addition, it indicates that 65% (53 out of 82) of the former HOV lane users switched to the regular GP lanes. Concerning the carpool pick, 19% (63 out of 332) of the respondents usually carpool while the remaining 81% drive alone for commuting. Changes in carpooling behavior are too observed. Responses point that carpool break-ups outpaced carpool formation. Carpool formation was but 2.viii% (seven out of 251) while 31% (25 out of 81) of sometime carpoolers left their carpools. The information indicate that 89% (56 out of 63) of the remaining carpoolers were old carpoolers.

Table 4. Respondents' behavioral changes in HOT lane and carpool choices.

5. Analytical approaches

Ii approaches, classification trees and logistic regressions, were used for developing statistical models designed to explicate drivers' behavioral responses in their commute travel. The approaches can be used to estimate the form membership of a chiselled dependent variable (Camdeviren, Yazici, Akkus, Bugdayci, & Sungur, 2007). Indeed, this study uses binary dependent variables past assigning an indicator value of i for cases where respondents choose HOT lanes (or carpool lanes), and zero otherwise.

5.ane. Nomenclature trees

To obtain a better agreement of commuter characteristics, a multi-dimensional analysis considering interactions betwixt factors was conducted using the tree-based regression and classification technique. This approach is attractive because the resultant copse provide a symbolic representation that lends itself to easy human interpretation (Camdeviren et al., 2007). In detail, this study applies classification trees to discrete dependent choices of HOT lane use (or carpool lane apply), with the selected contained variables. The technique splits the data through a binary partition, thus generating two resultant regions. As the partitioning process continues, the tree tends to grow, resulting in over-fitted and complicated models. Meanwhile, a tree that is too small might non capture the of import structure of the information. Thus, an optimal tree size should be adaptively chosen from the data.

This study utilizes the cross-validation technique in finding an optimal tree. In the approach, the cost of the tree by tree size is computed based on the ten-fold cross-validation method (Breiman, Friedman, Stone, & Olshen, 1984; Hastie, Tibshirani, & Freidman, 2001). The cost is the sum over all terminal nodes of the estimated probability of that node times the sum of the misclassification errors of the observations in that node. The best tree size, or the number of terminal nodes, is the one that produces the smallest tree that is within one standard fault of the minimum-toll subtree.

v.2. Logistic regression

Logistic regression models were also applied to identify the factors affecting drivers' choices of HOT lanes and carpooling in the I-85 corridor. In the model, the response variable has only two possible outcomes, whether the respondent usually uses HOT lanes or whether they practice not. When Yi is an independent Bernoulli random variable for the i th observation with an expected value Due east{Yi }, the logistic regression model with chiliad predictor variables, known constants ten and coefficients to be estimated β is expressed as follows (Kutner, Nachtsheim, Neter, & Li, 2005): E Y i = exp ⁡ ( β 0 + β 1 x i ane + … + β g x ik ) 1 + exp ⁡ ( β 0 + β i 10 i 1 + … + β thou x ik ) .

The interpretation of the estimated regression coefficients in the fitted logistic response function is non straightforward as in a linear regression model. The reason is that the effect of a unit increase in predictor variables varies depending on the location of the starting point on the predictor variable calibration (Kutner et al., 2005). Thus, the odds ratio, which is computed past taking the exponent value of the estimated coefficient, is used for associating the outcome with explanatory variables. Odds ratios higher up one indicate that the event is more likely to occur while odds ratios smaller than 1 indicate lower chances of the upshot to occur.

Kim (2009) showed that logistic regression models tin exist more efficiently developed past utilizing the results of the tree-based regression and classification technique. This is because classification trees may reveal statistically meaningful interactions between the explanatory variables, helping analysts identify which interaction effects should be entered in the regression models. In particular, the approach is substantially helpful when numerous and circuitous interaction effects may exist.

6. Results

vi.i. HOT lane choice classification trees

Classification tree analyses considering the eleven factors were performed using MATLAB 2015b to assess HOT lane choices. Firstly, the costs of the models by the tree size were estimated based on the 10-fold cross-validation arroyo to identify the best tree size. The graphs in Figure 1 show the estimated costs for the two models using all the 11 factors and excluding the variable of the perception of the effectiveness of the HOT lanes. The graphs imply that the perception variable has a substantially potent explanatory power, which can be explained in two means. Showtime, when all the eleven factors are considered, the cost is minimized at two terminal nodes, with the perception being the single variable dividing the choices. 2nd, the misclassification errors (costs) become much larger when the perception variable is excluded, which tin be easily identified by comparing the costs in the two graphs in Figure 1.

Figure 1. Nomenclature tree cost (error) by tree size for HOT lane choice. (a) Models for all the 11 factors; (b) models without the perception variable. Note: The dashed line indicates one standard error of the minimum-toll subtree.

The estimated costs in Figure one indicate that the best tree for the model with the perception variable has only two terminal nodes. The 2-node model, nonetheless, may fail to capture the of import aspects of the choices because of its simplicity. Thus, a classification tree with six terminal nodes, the second best tree in terms of the cost, was developed as an alternative for explaining the lane choices. The adult tree with 6 terminal nodes is shown in Effigy ii, illustrating that v variables are disquisitional factors: perception of benefit, age, quondam HOV user, piece of work kickoff time and number of children. The tree implies that the respondents who practice non perceive the HOT lanes have improved their own commute conditions are more probable to choose regular GP lanes instead of the HOT lanes. Of the respondents who perceive the positive effects of the HOT lanes, the ones in their age of 40s are more likely to choose the HOT lanes. In add-on, the model implies that the respondents who usually used HOV lanes are more likely to use the HOT lanes. The respondents who usually start to work between 7 and 9 a.one thousand. and have children, likely time-constrained commuters during morning peak hours, are also found to take a stronger tendency to choose the HOT lanes.

Effigy ii. Classification tree with six final nodes for the HOT lane choice (including the perception variable).

Considering of the ascendant touch of the perception variable, the influences of other variables may exist curtained. Thus, an test of a classification tree without the perception variable is as well of interest. Every bit suggested by the model costs shown in Figure 1, a tree with 4 terminal nodes was synthetic as the best model for the perception-excluded information. Figure 3 illustrates the tree depicted by three variables: historic period, education, and number of children. Unlike the previous model, the educational activity variable is constitute to be an of import factor; the respondents with a available'southward degree or higher are more than likely to choose the HOT lanes. It is conjectured that the instruction level may reverberate the financial ability of the respondents to pay a price.

Effigy 3. Classification tree for HOT lane choice (excluding the perception variable).

6.two. HOT lane option logistic regression

It seems that the constructed classification trees successfully identified potential relationships between lane choices and influential factors. Yet, they exercise not announced to exist sufficient to show the factors' statistical significances in a measurable way. To overcome this situation, further investigations were attempted by developing three logistic regression models using IBM SPSS Statistics 23 with different independent variables: models with main effects only (Model one), with both principal and interaction furnishings (Model 2), and without the perception variable (Model three). The results of the classification copse were fully utilized when identifying the appropriate forms of independent variables. More specifically, the chief result variables were re-divers based on the cut points revealed by the classification trees, treating them as categorical variables. As a result, all eleven factors were simply classified into binary cases. Moreover, potentially influential interaction terms were identified in an efficient and effective manner based on the resultant classification trees. In other words, without the data provided past the classification copse, potential 55 interaction terms, all combinations of two from eleven main furnishings, should accept been considered for specifying the interaction models. Still, the interactions of the factors revealed by the classification copse significantly reduce the number of interaction terms to be entered in the model to a practically implementable level.

The resultant logistic regression models were summarized in Tabular array 5, where only statistically significant variables at a significance level of 0.10 were captured based on a backward stepwise procedure, eliminating variables that do non add explanatory ability to the model. This stepwise procedure is beneficial to systematically exclude correlated independent variables (Kutner et al., 2005). The table also shows the Hosmer-Lemeshow goodness-of-fit statistics, of which p-values are at least 0.181, implying that the estimated models properly follow the key property of the logistic response office at a significant level of 0.05. The Nagelkerke R-square statistics suggest that Model 2 considering primary and interaction effects together has the strongest explanatory power amidst the iii models although the deviation from Model 1 appears to be marginal. This justifies the inclusion of the interaction effects. Moreover, the inclusion can be beneficial in that the interaction terms tin can nowadays useful perspectives for those aiming at improving HOT lanes use. Meanwhile, the model excluding the perception variable has the least explanatory power, indicating the variable's influential impact on the lane choices as already revealed in the nomenclature tree analyses.

Table 5. Logistic regression models for the HOT lane choices.

The model considering only main furnishings captured iv statistically meaning variables: the perception, former HOV user, commute altitude, and age. In item, the odds ratio for the perception variable indicates that respondents are near eleven times more than likely to utilize the HOT lanes when they positively perceive the effectiveness of the HOT lanes. Age as well appears to exist influential in the lane selection conclusion; respondents in their 40s are 2.eight times more likely to choose the HOT lanes than respondents in other historic period groups. In addition, commute distance, which was non a significant gene in the classification tree analyses, was establish to exist a critical one although its bear upon is rather weaker compared to the other iii factors. The longer-distance commuters, particularly longer than thirty miles (48 km), take a stronger trend to employ the HOT lanes. This finding may be ascribed to the attribute that the longer travelers tin can gain more travel fourth dimension-saving benefits by traveling on the HOT lanes during congested peak hours.

When the interaction effects are considered, five variables, including two main effects (the perception and commute distance) and three interaction terms (combinations of sometime HOV user, age, and perception) are plant to be significant. Interestingly, the odds ratio for the perception variable decreases by virtually half from 10.880 to 5.246, compared to Model one, although perception is withal significantly meaningful in explaining the choices. It seems that the explanatory ability of the perception variable is dispersed over the two interaction terms combined with onetime HOV user and age. In fact, the interaction terms, former HOV user by perception, and age by perception, have relatively loftier odds ratios, iv.901 and 5.318, respectively. Model ii besides shows that two main effects of quondam HOV user and age in 40s are no longer pregnant by themselves. Instead, they appear to be significant merely when they are combined with other factors, implying a simple consideration of chief effects may neglect to fully capture the characteristics of the data. A potential benefit of the model with interaction terms seems to be its enhanced capability to predict the choices more specifically.

The model excluding the perception variable reveals boosted significant variables not shown in the previous models: household income and historic period by education. HOT lane positive perception may exist related to some extent to these variables, perhaps tied to employment in some style. In the later department, a model is presented to evidence the relationship between the perception and other variables. However, the substantially lowered explanatory power of the model measured by Nagelkerke R2 (from 0.364 to 0.145) indicates that the variables cannot fully supercede the perception variable in explaining the lane choices. This aspect may justify the use of the perception variable for the model evolution. The estimated model shows that the respondents with a high income and a college didactics degree are more likely to choose the HOT lanes. In detail, the respondents in their 40s and with a bachelor'due south caste or higher are found to have iii.5 times more than chances to use the HOT lanes.

half dozen.3. Carpool pick classification trees

Classification copse were developed to analyze the commuters' carpool choices using the selected 332 samples. As illustrated in the HOT lane selection models, the all-time tree size was first identified using the cost functions of the trees. The cost changes of the trees considering all the 11 factors, shown in Effigy iv, indicate that a tree with two or five terminal nodes may exist adequate for explaining the carpool selection behavior. When the two-terminal node tree was considered, the one-time carpooler variable was found to exist the unmarried factor predicting carpool choice, given that the majority of prior carpoolers are still carpooling. Indeed, the cost graph in Figure 4 illustrates that the cost becomes much larger when the sometime carpooler variable was excluded. Although the two-final node tree is meaningful, its simplicity may fail to provide sufficient data on the information construction. Thus, a five final node tree was selected for analyzing the data. Meanwhile, when constructing a tree without the former carpooler variable, seven terminal nodes were considered as suggested by the tree costs in Figure 4.

Figure four. Nomenclature tree cost (error) by tree size for carpool selection. (a) Models for all 11 factors; (b) models without the old carpool user variable. Note: The dashed line indicates one standard mistake of the minimum-price subtree.

Effigy five illustrates a tree with five terminal nodes depicted by iv factors: former carpooler, number of workers, historic period, and income. The tree strongly supports that respondents are less probable to carpool unless they were already carpoolers before the HOT implementation. In fact, the importance of the former carpooler variable was expected past the sample characteristics; 89% of the carpoolers are the former carpoolers (see Table 4). The figure besides shows that even amid the old carpoolers, respondents in their 40s whose households accept a single worker are more likely to bulldoze alone for their commutes.

Figure five. Classification tree for carpool selection.

Information technology is plausible that under a situation in which i single variable explains virtually all of the variability, the furnishings of other important factors can be obscured. Thus, further analyses were conducted past developing an additional classification tree without the former carpooler variable. Interestingly, the adult tree with vii last nodes, shown in Effigy 6, illustrates that the perception of the effectiveness of the HOT lanes has the strongest impact on the carpool pick decision, pointing out that the respondents who accept a positive perception of the HOT lanes are less likely to carpool. It should be noted that the positive perception of the HOT lanes is also correlated with higher chances of using the HOT lanes, as identified in the HOT lane choice models. Thus, HOT lane use and carpool choices are negatively associated, at least for the I-85 HOT corridor commuters. The tree revealed that the number of vehicles for commuting tin play a role in determining the conclusion. However, the influence of vehicle ownership varies depending on subtree factors such equally gender and number of workers. The tree structure also indicates that females are more likely to carpool.

Effigy half-dozen. Nomenclature Tree for Carpool Selection (Excluding the Sometime Carpooler Variable).

6.four. Carpool pick logistic regression

Further analyses were performed to examine the factors that may influence carpool choice using logistic regression models. The model specifications and procedures were identical to those of the HOT lane choice models, except that the number of worker variable was reclassified to have three groups (one, 2, and 3+), reflecting the cut points suggested in the classification trees. In improver, the number of vehicles available for commuting was not used as an explanatory variable, because the variable was institute to be significantly correlated with the number of household workers (Pearson's correlation coefficient = 0.625).

The model considering only principal effects revealed that the former carpooler variable is the single dominant cistron at a significance level of 0.05 with a Nagelkerke R2 value of 0.611. (This model is not reported for conserving space.) This may be incurred by the data characteristics; a majority of the carpoolers are the erstwhile carpoolers (56 out of 63). This characteristic became more pronounced in another model which considers both the main and interaction furnishings. (This model is as well not reported for conserving infinite.) Indeed, the estimated parameters in that model seem to exist inflated, implying the maximum likelihood estimates are not properly obtained. This situation clearly indicates that the data have a separation problem which occasionally happens in logistic or probit regressions (Heinze & Schemper, 2002). In other words, the erstwhile carpooler variable separates the carpool choices near completely except for 7 cases. When separation occurs, ii approaches are oftentimes employed: 1) "mechanical" measures including increasing sample size, combining the category, with similar ones and omitting the category, and two) statistical measures such as Firth'southward penalized maximum likelihood method (Gim & Ko, 2017).

This study developed a carpool choice model by omitting the former carpooler variable (one of the common "mechanical measures") to be consequent with the HOT lane selection models. As well, it was conjectured that the use of this influential variable would obscure the impacts of other important factors, in particular for a pocket-size sample size data prepare. Future studies may consider other alternative approaches for this modeling. Table 6 illustrates the event of the estimated model, pointing out three main effects and two interaction effects that are statistically pregnant. The model suggests that the respondents who are in their 40s, start to piece of work betwixt seven and nine a.one thousand., and have two or more workers in their households are more probable to carpool. Combined with the finding that the respondents in their 40s are prone to apply the HOT lanes more, this result implies that they are also more than probable to apply the HOT lanes in carpool mode.

Tabular array 6. Logistic regression models for carpool option (excluding the one-time carpool user variable).

Every bit institute in the classification tree, the interaction effects reveal that the participants who have a positive perception of the HOT lanes have a weaker tendency to carpool, which may statistically back up that HOT lanes may negatively influence carpooling. The perception variable is found to interact with historic period (40s) and the number of workers (two-worker households) and their impacts seem to be substantial as suggested by the magnitudes of the estimated parameters (−2.182 and −1.176). The resultant Nagelkerke Rii value of 0.165 suggests that the model lacks the ability to strongly predict the carpool choices. Future studies are encouraged to contain more factors including travelers' perceptions and attitudes into the model for better understanding carpool beliefs.

6.five. The perception model

It was suspected that the perception of the HOT lanes might accept associations with other factors. To examine this, ordered probit models were developed, considering that the perception was measured by a five-point Likert calibration from one (definitely not improved) to five (definitely improved). The model was adult based on the data set of the HOT lane choice model and the car ownership variable was excluded due to its strong correlation with the number of household workers. Table 7 presents the resultant models, illustrating the six factors that are statistically meaning at a level of 0.10: gender, number of household workers, income, former HOV user, work showtime time, and commute altitude. Interestingly, the former HOV users appear to negatively perceive the HOT lanes, implying the HOT implementation might not be preferred by them, and thus may influence the breakup of carpools. The HOT implementation is also negatively perceived by commuters who usually start their work between 7 and 9 a.m., which may be ascribed to decreased travel speeds even in the HOT lanes during morning peak hours. Further studies are encouraged to explore these phenomena in more detail for ameliorate interpretations.

Table 7. Ordered Probit Models for the Perception of Improved Commute Atmospheric condition.

Despite the appearance of the significant variables, the overall explanatory power of the perception model seems unsatisfactory every bit suggested by the low value of Nagelkerke R2 (0.159), suggesting the lack of capability of the model to predict HOT lane perception using the variables. This state of affairs may justify the inclusion of the perception variable in the choice models together with other variables. More than research in this area is definitely warranted.

7. Conclusions

The understanding of commuters' responses to HOT installations is important in that it can help transportation agencies identify operational strategies designed to maximize the sustainable use of HOT facilities. This study explores Atlanta'southward HOT lane implementation and carpool choices over the I-85 HOT corridors using data collected through a questionnaire-based survey. The self-administered mail-out/mail service-back survey asked respondents most their lane choices (HOT or regular GP lanes) and carpool choices, both before and after the HOT lane installation, likewise equally overall trip patterns and demographic data. This survey is meaningful in that it was designed as the starting time attempt to assess carpool behavior after the installation of a conversion of an HOV lane to a HOT lane. As expected, the retrieval rate of the survey was low (about v%), and a significant number of the retrieved surveys were not usable for developing certain statistical models, due to missing values and multiple answers for the same questions. Although low sample size restricted this study from fully utilizing respondents' various behavioral responses before and after the HOT installation, the binary option models via classification trees and logistic regressions produced interpretable results that help explain the commuters' lane and carpool choices.

The HOT lane choice models showed that the perception of the effectiveness of the HOT lanes, a unique variable rarely treated in HOT behavior studies, exerts the strongest impact on the choices. More specifically, commuters are more likely to choose HOT lanes when they perceive HOT lanes have improved their own commute conditions. This finding implies that HOT operators should maintain an acceptable level of HOT lane performance for maximizing the utilization of the lanes. The models also suggested that HOT lane choices can be affected by commuters' socio-economic characteristics. Commuters in their 40s, commuters with higher income, and commuters with higher education levels are more likely to cull the HOT lanes. This suggests that commuters with a loftier value of time are more likely to utilize HOT lanes, as expected. The importance of the age variable was also illustrated in the SR91 Limited Lanes study, but in the SR91 report the age group in the 50s showed a stronger tendency to apply HOT lanes (Li, 2001). Apropos trip patterns, commuters making longer trips were found to more likely choose HOT lanes. This situation appears to be intuitively correct, in that these travelers may have a stronger intention to save their travel times. The touch on of trip length was non found significant in Li'southward (2001) written report. The models pointed out that former HOV lane users tended to choose HOT lanes, suggesting many one-time HOV lane users might opt to use HOT lanes even afterwards the HOT conversion. However, it is non clear how they use HOT lanes: paying a toll or HOV3+. Futurity studies are encouraged to investigate these choices in detail for a better understanding of commuters' behavior.

Regarding carpool choices, the selected data prepare showed that most carpoolers afterwards the HOT installation were composed of former carpoolers. Weak carpool formation was noted, even after the HOT conversion, which is in the same vein as the conclusion of Burris et al. (2014). As well, the developed models revealed that the former carpooler variable dominated the consequence on the carpool pick. This attribute clearly indicates that before-and-after carpooling behavior should be considered together to firmly understand drivers' behavior in particular when the HOT lanes are converted from HOV lanes. Statistical models also showed that commuters' socioeconomic characteristics could affect the carpool choice. Commuters in their 40s, commuters that have two or more workers in their households, and commuters that start work betwixt 7 and 9 a.thou. are more likely to carpool. In addition, females have a higher tendency to carpool, which conforms to the findings of studies conducted in France (Delhomme & Gheorghiu, 2016) and in Dallas-Fort Worth and Houston in Texas (Li et al., 2007). Still, the models also indicate that commuters who have a positive perception of the HOT lanes are less likely to carpool. In particular, the synthetic classification tree revealed that perception was the near of import factor when the former carpooler variable was excluded. Based upon the survey data, this HOT project did not enhance carpooling as originally expected by the project proponents, which was also confirmed by vehicle occupancy evaluation in the previous before-after study (Guensler et al., 2013). This may also mean that carpools could continue to break-up every bit the operation of HOT lanes continues to better. From the perspective that college vehicle occupancy can generally enhance the sustainability level of transportation systems, this finding appears frustrating in spite of a potential sample bias that the postal service-out/post-back survey tin can retain. Policy makers may need to rethink strategies designed to increase carpool formation and retention as they implement HOT projects throughout the region.

Complementing previous studies, this study has enhanced the agreement of HOT lane and carpool choices on HOT corridors in particular by revealing the stiff association between perception and mode/lane choices. However, the findings obtained from the binary pick observations seem to still exit numerous unexplained behavioral responses of the commuters, which might accept been overcome with larger samples and more complete survey responses. A sufficient sample may be able to provide researchers with more chances to examine their complex controlling mechanisms. It is also uncertain what actually happened nigh travelers' departure times and routes later the HOT lane installation. This stresses the importance of a better before-and-after information collection for the next managed lane conversion. In addition, the limited number of factors considered can explain just a small-scale portion of HOT lane or carpool manner decision-making processes. Indeed, the explanatory power of the lane choice model was at most 0.36 in terms of Nagelkerke R2 . Work place TDM options and toll pricing effects, which were not captured in the survey, may deserve to be considered for the boosted factors. Time to come study efforts are encouraged to capture larger samples and explore boosted variables for developing improved models.

mckinleyexcied82.blogspot.com

Source: https://www.tandfonline.com/doi/full/10.1080/15568318.2019.1663961