Slugging in Houston

Journal of Public Transportation, Vol. 9, No. 5, 2006 9/8/06

30

and 8 minutes on the Northwest Freeway. The average time spent waiting for a

bus was assumed to be half of the average headway based on the assumption of

random arrivals of transit passengers (Meyer and Miller 2001). Casual carpoolers

saved an average of 2 minutes 36 seconds over transit on the Katy Freeway and 1

minute 36 seconds on the Northwest Freeway.

Other factors besides travel time savings might have influenced the mode choice

of the travelers. Monetary costs (e.g., transit fare, fuel) or trip purpose could have

affected a traveler's decision (Wall 2002). Socioeconomic characteristics could also

have had a major influence on a traveler's decision to casual carpool. Travelers may

have valued the reliability of travel times on the HOV lane. The survey data were

used to determine what, if any, trip and socioeconomic characteristics increased

the likelihood of a traveler choosing to casual carpool on a frequent (3 or more

times per week) basis.

Comparison of Traveler Characteristics by Mode

The survey data were initially examined for significant differences (p ≤ 0.05)

among four groups of travelers based on their primary mode choice: driving on

main lanes, using HOV lane with a traditional carpool, casual carpooling, and tran-

sit. A Chi-Square test assessed significant differences among the binary variables,

and a one-way analysis of variance (ANOVA) examined the continuous variables.

Additionally, a Kruskal-Wallis test determined any significant difference between

groups for the ordinal variables of age, income, and education.

The results of the statistical tests revealed significant differences among travelers

in the four primary morning modes of travel (Table 2). The percentage of respon-

dents on commute, recreation, school, and other trip types was significantly differ-

ent among the four groups. Casual carpoolers were more likely to be on commute

trips. The percentage of respondents ages 25 to 34 and 65+ was significantly differ-

ent among modes. A much higher percentage of casual carpoolers were between

ages 25 and 34. The average household size, percentage of single adult house-

holds and married without children households, and the number of vehicles per

household also differed among modes, with HOV users having significantly larger

households. A difference was also found for those with occupations that were pro-

fessional/managerial, sales, homemaker, self-employed, or retired. Income ranges

of $25,000 to $35,000, $50,000 to $75,000, $100,000 to $200,000, and $200,000 or

more were also different among the four mode choices.

Page 9

Slugging in Houston

31

Table 2. Descriptive Statistics of Surveyed Travelers

1

Significant (p ≤ 0.05) difference when comparing all four modes.

2

Significant (p ≤ 0.05) difference when comparing casual carpooling and transit.

Page 10

Journal of Public Transportation, Vol. 9, No. 5, 2006

32

Next, similar statistical tests were performed to determine significant differences

(p ≤ 0.05) between travelers using just two mode choices: casual carpooling and

transit. These mode choices were specifically examined due to their symbiotic

relationship and the similarity of the modes since travelers on both modes (1)

use park-and-ride lots, (2) have someone else drive, (3) travel on HOV lanes, and

(4) are dropped off relatively close to their work. Also, casual carpoolers are often

former transit users (Beroldo 1990), and in this study more than 90 percent still

used transit for some of their similar trips (Table 3).

The results of the statistical tests (Table 2) revealed several significant differ-

ences between casual carpoolers and transit riders. A higher percentage of casual

carpool passengers were on commute trips and between the ages of 25 and 34,

while a higher percentage of transit riders were between the ages of 55 and 64. A

significantly higher percentage of casual carpoolers had professional/managerial

occupations, while a significantly higher percentage of transit riders had house-

hold incomes between $25,000 and $34,999.

Casual Carpool Passenger Characteristics

The surveys distributed to casual carpool passengers contained a series of ques-

tions that were exclusive to that group. These questions addressed the nature

of each traveler's casual carpooling trip and his or her previous experience using

the mode (Table 3). For this analysis only, both frequent and infrequent casual

carpoolers were examined. The results provided insight into the practice of casual

carpooling in Houston, including what modes were commonly used for return

trips and how frequently respondents joined a casual carpool.

Survey responses indicated that most casual carpool passengers (65.3%) had

never met their travel companions before. However, almost one third indicated

that they had traveled with them once or twice, indicating that a relatively small

community of people used the mode consistently. More than 75 percent of users

noted that they casual carpooled at least three times per week. Passengers also

cited saving money (62.8%) and slow bus service (52.6%) as the two primary rea-

sons for casual carpooling. They indicated that they often use the bus for similar

trips and for the evening return trip. They also noted that money is rarely given to

the driver as compensation, which is consistent with casual carpooling practices

elsewhere in the United States.

Page 11

Slugging in Houston

33

Table 3. Casual Carpool Passenger Characteristics (n = 208)

Note: Some percentages sum to over 100 percent as respondents could choose multiple answers

for some questions

Page 12

Journal of Public Transportation, Vol. 9, No. 5, 2006

34

Mode Choice Model Estimation

To better understand casual carpoolers and the factors that affect their mode

choice, discrete choice model coefficients were estimated for two sets of choices.

The choice between casual carpooling and transit was evaluated with the first

model. The second model examined traveler choice of four modes: casual carpool,

transit, traditional carpool, and driving on GPLs.

Methodology

Both models were estimated as discrete choice models. Discrete choice models

assume that each traveler makes his or her decision based on the utility of each

mode (Ben-Akiva and Lerman 1985). The traveler's ultimate decision will deter-

mined by both the systematic utility based on measured variables and the random

utility of each mode. The model in this analysis was estimated using a logit model,

which assumes that random utilities follow an extreme value distribution (Small

and Winston 1999).

Casual Carpool versus Transit Mode Choice Model

Although many variables were tested when estimating the model coefficients, only

those variables significant at the 95 percent confidence level and not correlated

to other variables were left in the final model. The results of the discrete choice

model are shown in Table 4. For this model, the null choice was casual carpooling.

The utility function derived in the model describes the utility of the transit mode

relative to the casual carpooling mode that had all coefficients equal to zero.

The results of the model highlight some of the factors that describe selected types

of travelers who choose to casual carpool rather than use transit. The constant

coefficient is positive, indicating that all else being equal, travelers were more likely

to choose transit than casual carpooling. This was not surprising as many more

travelers used transit than casual carpools. The results also indicated that having

an income between $25,000 and $35,000 increased the traveler's likelihood to use

transit rather than casual carpooling. However, being on a commute trip, making

a higher number of total trips per week, and/or being between the ages of 25 and

34 increased the traveler's likelihood of forming casual carpools.

Page 13

Slugging in Houston

35

Table 4. Model Coefficient Estimation Results

(Casual Carpooling vs. Transit)

1

1

Base alternative is casual carpooling with utility of zero.

*Significant at the 95 percent confidence level.

Four-Option Mode Choice Model

Several sets of variables were used for testing the four-choice model, using the

main lanes option as the null choice. Only variables significant at the 95 percent

confidence level remained in the final model. The variables used in the model as

well as which mode choice utility functions they were associated with are listed in

Table 5, while the model estimation results are shown in Table 6.

The constants for the HOV, casual carpool, and transit modes were all negative,

indicating that all else being equal, travelers were most likely to drive on the main

lanes. The trip purpose, age, and occupation (professional) variables applied only

to the casual carpooling utility function and indicated a number of factors influ-

enced casual carpoolers' decisions. The coefficient for the trip purpose was positive,

indicating that being on a commute trip increased the likelihood that a traveler

Page 14

Journal of Public Transportation, Vol. 9, No. 5, 2006

36

Table 5. Definitions of Variables Used in Logit Model (All Four Modes)

would choose casual carpooling over the other three modes, which duplicates the

results of the previous model. Professional/managerial or administrative/clerical

occupations also increased a traveler's likelihood to use casual carpooling over the

other three modes. Thus, travelers with weekday jobs with typical workday hours

were more likely to casual carpool. This was not surprising considering the times

during which casual carpooling occurs. Travelers with typical workdays would be

more likely to encounter peak-period congestion if they drove alone on the GPLs.

The results also indicated being between the ages of 55 and 64 reduced a traveler's

likelihood of casual carpooling, which reflected a possible increased willingness

among younger persons to try a newer, less-utilized mode of transportation.

In addition, having an income between $25,000 and $35,000 reduced a traveler's

likelihood of casual carpooling, which was surprising considering the relatively low

expense of that mode. One possible explanation was that low-income persons

already used transit for many of their other trips, and they chose to use transit

Page 15

Slugging in Houston

37

Table 6. Model Coefficient Estimation Results (All Four Modes)

1

Base alternative is driving alone on main lanes with utility of zero.

*Significant at the 95 percent confidence level.

Page 16

Journal of Public Transportation, Vol. 9, No. 5, 2006

38

during the times of casual carpooling as well. Another possible explanation was

subsidized transit passes were available to low-income travelers. Travelers with

subsidized transit passes would have little to no money-savings incentive to casual

carpool. Also, the descriptive statistics indicated that travelers with incomes

between $25,000 and $35,000 were less likely to make commute trips, leading to

less use of casual carpooling because commuting is a primary factor that influ-

ences casual carpool use.

Summary

This research effort examined the use of casual carpooling in Houston, Texas. Sur-

vey results revealed that most casual carpool passengers often used transit for eve-

ning return trips and similar morning trips. Approximately 63 percent used casual

carpooling to save money and about 53 percent used casual carpooling because of

slow bus service. Most casual carpoolers (76%) used this mode three or more times

per week. Casual carpool passengers were significantly more likely to be on com-

mute trips and be between the ages of 25 and 34 (younger), but were significantly

less likely to have household incomes between $25,000 and $35,000.

The results obtained in these analyses provided some information on the charac-

teristics of travelers who chose to casual carpool. This information can be used to

better evaluate HOV/HOT lane use and future lane development considerations.

Casual carpooling has grown in popularity and should be considered when assess-

ing potential corridor improvements. Although potential liability concerns would

likely prevent agencies from actively promoting casual carpooling, they could

encourage it passively by constructing park-and-ride HOV facilities that are con-

ducive to the mode. Casual carpooling has the potential to improve the operation

efficiency of HOV/HOT facilities by improving person movement. Although there

are potential liability concerns, it may eventually become beneficial to promote

casual carpooling as a viable mode alternative.

Acknowledgments

This article reflects the views of the authors, who are responsible for the facts and

the accuracy of the data presented herein. This article was a result of research

conducted in cooperation with the Federal Highway Administration (FHWA),

the Texas Department of Transportation (TxDOT), and the Metropolitan Transit

Authority of Harris County, Texas. The authors gratefully acknowledge the con-

39

tributions of numerous individuals and organizations that made the successful

completion of this article possible.

References

Ben-Akiva, M., and S. Lerman. 1985. Discrete choice analysis: Theory and application

to travel demand. Cambridge, MA: MIT Press.

Beroldo, S. 1990. Casual carpooling in the San Francisco Bay Area. Transportation

Quarterly 44 (1): 133–150.

Burris, M., and J. Appiah. 2004. An examination of Houston's QuickRide partici-

pants by frequency of QuickRide usage. Transportation Research Record 1864,

22–30.

Burris, M., and W. Stockton. 2004. HOT lanes in Houston—Six years of experience.

Journal of Public Transportation 7 (3): 1–21.

Houston TranStar Real Time Traffic Information. http://traffic.tamu.edu. Accessed

June 15, 2004.

LeBlanc, D. E. 1999. Slugging: The commuting alternative for Washington, D.C. East

Point, GA: Forel Publishing.

Meyer, D., and E. Miller. 2001. Urban transportation planning. New York: McGraw-

Hill.

Ojah, M., and M. Burris. 2004. Quantification of casual carpooling in Houston. Texas

Transportation Institute.

Reno, A.T., W. A. Gellert, and A. Verzosa. 1989. Evaluation of Springfield instant

carpooling. Transportation Research Record 1212, 53–62.

RIDES for Bay Area Commuters, Inc. 1999. Casual carpooling 1998 update.

rideshare.511.org/research/pdfs/casualcarpool.99.pdf> Accessed December

8, 2003.

Small, K. A., and C. Winston. 1999. The demand for transportation: Models and

applications. Essays in Transportation Economics and Policy. Washington, DC:

Brookings Institute Press.

Spielberg, F., and P. Shapiro. 2000. Mating habits of slugs: Dynamic carpool forma-

tion in the I-95/I-395 corridor of Northern Virginia. Transportation Research

Record 1711, 31–38.

Page 18

Journal of Public Transportation, Vol. 9, No. 5, 2006