Commut Stress Ride Novaco
Commut Stress Ride Novaco
Commut Stress Ride Novaco
of California Center
The University of California Transportation Center (UCTC) is one of ten regional units mandated by Congress and established in Fall 1988 to support research, education, and training in surface transportation The UCCenter serves federal Region IX and is supported by matching grants from the U.S. Department of Transportation, the California Department of Transportation (Caltrans), and the University. Based on the Berkeley Campus, UCTCdraws upon existing capabilities and resources of the Institutes of Transportation Studies at Berkeley, Davis, Irvine, and Los Angeles; the institute of Urban and Regional Development at Berkeley; and several academic departments at the Berkeley, Davis, Irvine, and Los Angeles campuses. Faculty and students on other University of California campuses may participate in
Center activities. Researchers at other universities within the region also have opportunities to coUabomte with UC faculty on selected studies. UCTCseducational and research programs are focused on strategic planning for improving metropolitan accessibility, with emphasis on the special conditions in Region IX. Particular attention is directed to strategies for using transportation as an instrument of economic development, while also accommodating to the regions persistent expansion and while maintaining and enhancing the quality of life there. The Center distributes reports on its research in working papers, monographs, and in reprints of published articles. It also publishes Access, a magazine presenting summaries of selected studies. For a list of publications in print, write to the address below.
University of California Tramportation Center 108 Naval ArchitectureBuilding Berkeley,C.alifomia94720 Tel: 510/643-7378 FAX:510/643-5456
Thecontentsof this report reflect the views of the authorwho is responsible for the facts and accuracy of the data presentedherein. Thecontentsdo not neee~axtly reflect the official views or policies of the State of Californiaor the U.S.Department of Transportation. This report does not constitute a standard, specification,or regulation.
Commuting Stress, Ridesharing, and Gender: Analyses from the 1993 State of the Commute Study Southern California
in
Raymond W. Novaco
Schoolof Social Ecologyand Institute of Transportation Studies University of California at Irvine Irvine, CA92717
Cheryl Collier
Commuter Transportation Services, Inc. 3550Wilshire Blvd., Suite 300 Los Angeles, CA90010
Working Paper January 1994 presented at the 73rd annual meetingof the TransportationResearchBoard,Washington, D.C.
UCTC No. 208 The University of California Transportation Center University of California at Berkeley
Abstract A stressful nature of exposure to traffic congestion in automobile commuting has been
research that has been measurement-intensive but study examined commuting stress in
sample (N = 2591) in southern California through associated with distance and the stressful effects of
duration of the commute, controlling for age and income. As predicted, long distance commutes (> 20 miles) were further much greater commuting stress
as women in such
commutes perceive
effects of ridesharing were found, as full-time ridesharers congestion and more satisfied
In analyses of prospective
it was found that the perception of ones commuteas having a negative impact on family life had a very significant effect on the inclination to try carpooling and train/rail, associated with distance itself. congestion that is internalized beyond the effect
Commutingstress is discussed as an external cost of traffic by the solo driver. Marketing strategies by highlighting for alternative modesof
Commuting Stress, Ridesharing, and Gender:. Analyses from the 1993 State of the Commute Survey in Southern California
Among the external costs believed to be associated with traffic congestion are the human stress effects. Remaining attached to the private automobiletravel modeand constrained by the availability of affordable housing, workers endure congested commutes and absorb the stressful consequences.Indeed, the stressful effects of chronic exposure to traffic congestion and other dem~mds of long-distance driving in commuting between home and work have been
demonstratedin a series of studies (most recently, Novaco,1992; Novaco,Stokols, & Milanesi, 1990; Novaco, Kliewer, & Broquet, 1991). In these studies, understood to be stressful behavioral constraint traffic congestion has been
thus constituting
frustrating condition. As such, it elevates physiologicalarousal, elicits negative emotionalstates, and impairs cognitive performance. This research has found that high impedance commuting, indo:ed by objective and subjective dimensions, has adverse effects on blood pressure, mood, frustration tolerance, illness occasions, workabsences,job stability, and overall life satisfaction. The methodology used in this research program on commuting stress has been a
measurement-intensive, quasi-experimental field site testing procedure that has incorporated manycontrol variables as covariates in the analyses. However,this methodological rigor has comeat the expenseof assurances regarding generalizability, due to the relatively small samples sizes (each of these previous projects has involved approximately 100 participants) location (all studies were conducted with companiesin one city). The present study, therefore, examined commutingstress associated with automobile travel in a muchlarger geographic area with a large representative sample of commuters. For this purpose, several items pertaining to commuting stress were added to the southern California and the
of this survey on previous years. The newly added items sought to index commuting exposure
stress in terms of the aversiveness of the commuteand the negative impact of traffic on work and home life. Ridesharing has been promoted as an alternative pollution, and stress.
air
The present study also examined the merit of ridesharing in comparison effects of long distance commuting. National trends in in the workforce, in the availability of
with solo driving regarding the stress commuting patterns have indicated
that increases
automobiles, and the shift of jobs to suburban locations have significantly by private automobile (Pisarski,
increased commuting
1987). From 1960 to 1980, travel to work by private automobile while use of public transit congestion and air quality declined from 12.6% to 5.2% management regulations have
prompted ridesharing programs, Teal (1987) has shown with national data that the large majority of drive-alone commuters lack any transportation or economic motivation to carpool. nonexistent, gasoline remains inexpensive, is illustrated by the fact increase in
In California,
and nearly everyone who needs a car has one. The impact of the latter that from 1980 to 1989, many California registered increased increased
For example, the population of San Francisco County while the number of registered automobiles
(12%) people and 183,643 (25.4%) autos; for Los Angeles County 1,205,900 (16.2%) people 870,191 (23.6%) autos; and for Orange County 357,990 (18o6%) people and 330,621 (30.8%) Regarding reducing the demandfor road space, transportation alternatives to solo driving
reduction mandated by Regulation XV(Giuliano, Hwang,and Wachs, 1992; Wachs& Giuliano, 1992) found that AverageVehicle Ridership increased from 1.213 to 1.246 (although there was considerable variation across sites) and that "(t)he number of work.sites meetingthe target actually decreased during the first year of the program"(p. 9). Increased carpooling accounted for virtually all of the increase in AVR. While the Giuliano et al. data from employment sites found the solo driving modeto be at 71%around mid-1991,the Collier and Christiansen (1992b) "State of the Commute" survey found the drive alone rate to be 77%at regulated sites across southern California counties, continuing the national trend of solo-driving commuting noted by Pisarsld (1987) and Liss (1991). To the extent that ridesharing mitigates commuting stress, marketing strategies might then be devised that highlight this benefit and be targeted toward the population sectors that are most at risk or whoare otherwise sensitized to stress themes. Regardingthe question of at risk or sensitized population sectors, the previous research on commuting stress has examined the hypothesized moderating role of gender, whereby domain were expected to be greater for females
commutingon routes of high physical impedance. Reasoning from a convergence of findings in the three areas of travel behavior, workersstress physiology, and workeffects on family life, Novacoet al. (1991) expected that female high impedancecommuters(>20 miles; > 50 minutes on PMcommute)would be highest on residential domain stress measures. This hypothesis was
strongly confirmed across multiple measures and analyses, which rigorously controlled for potentially confounding factors as covariates. The significant effects for high impedance females, relative to their male counterparts, were obtained for measuresof dysphoria, general spirits, satisfaction with location, desire to move, and ratings of the home physical environment;although not statistically significant, the
results were in the expected direction for negative mood at home,satisfaction with dwelling, and
satisfaction
appraised their commutes more negativel3) than did menin the samecondition, despite these women have higher family incomes and not differing in education, marital status, or home ownership; nor did they differ in the objective characteristics of their commutes. However, these high physical impedance females reported considerably more constraint particularly thart did men, jams,
for the AM commute - they reported being delayed more often by traffic
however, have more complex travel segments than did the high impedancemen in that study, hence Novaco et al. (1991) speculated that differential responsibilities) might be an explanatory factor. While the effort to explain previously obtained gender effects is an important research agenda, it is also imperative that the question of gender differences in commuting stress be examinedwith a muchlarger representative sample. Hence, this issue is pursued in the present study with the 1993 State of the Commute Survey. Gender effects were examined in various statistical designs with commuting mode(solo driving, part-time ridesharing, and full-time role strain (work and household
METHOD Survey Desi_~n and Procedure The State of the Commute is an annual study conducted by CommuterTransportation Services, Inc. (CTS). The 1993 State of the Commute study is based on a telephone survey 2,591 commuterswithin Los Angeles, Orange, Riverside, San Bemardino,and Verttura Counties. Respondents surveyed included only commuters whowork full-time whomthe home is a primary work site. and excluded those for on
about traffic
congestion, alternative
travel modes,
CTScontracted Interviewing Services of America, Inc. (ISA) to draw a sample based random-digit dialing using their copy of GENESYSs sampling program. This method, rather
than directories,
is used because of the high proportion of unlisted telephone numbers in the Los telephone
Angeles area. Random-digit dialing avoids the bias introduced by using only listed numbers An extensive cleaning and validation numbers in GENESYSs database probability were assigned
process was undertaken to ensure that all phone to the correct number. The survey questionnaire was area code and to increase the
pretested by interviewers from ISA. Since the majority of the survey questions were consistent with previous surveys, only minor formatting changes were made. using a Computer Assisted
Telephone Interviewing system (CATI), contacted respondents in the five-coun~ area. The use of a CATIsystem assures strict allows for extensive quality adherence to skip patterns, control. No interviews travel eliminates key entry errors, and
were conducted between November 19 and around the Thanksgiving holiday. and Spanish versions A
patterns
of the
were available
Five hundred
were completed in each county in order to make county comparisons with sample sizes of 500. A two percent sampling error
Regkmally, a total of 2,591 interviews was used in the analysis. is no:finally associated with sample sizes of 2,500.
Each interview began with the screener question, in your household work full-time
was based on the person who had the most recent birthday. avoid the possible bias of surveying a disproportionate
they are most likely to answer the telephone. Once interviewing had been completed, responses were weighted by the number of eligible respondents within each household. This ensures that Furthermore, for the analysis at
small households are not over-represented in sample statistics. the regional level, data were additionally
Commu ring Stress Measu res Four survey items constituted the commuting stress indices. "Commutingsatisfaction,"
rated on a nine-point scale, has been a item in previous State of the Commute surveys (Collier & Christiansen, 1991 & 1992), and a similar item has been a component of the subjective Thus, it is here incorporated
a stress index. Three other items, rated on five-point scales, were newly composedfor the 1993 State of the Commute study: "How often do you feel bothered by traffic congestion in
commuting to or from work?"; "After your commute to work, how often do you feel a need to wind down and relax before starting work?"; and "Some people say that dealing with traffic on
their commutehome from work has a negative effect on their home life. true for you?". The first
of these new items is intended to assess the aversiveness has been a principle components factor
in the subjective
impedance measures of Novaco et al. (1990; 1991). The other two items aim to assess work and home domain impacts which are part of the commuting stress construct. While it is less than
optimal to operationalize the construct with these four simple items, the pragmatics of survey research demandsimplicity.
Hypotheses and Analytical Procedures 1. Commuting stress indices were expected to be significantly correlated with distance and duration of the commute,controlling for age and income. This was examined by simple
correlation and in multiple regressions with the control variables. 2. Consistent with the concept of impedance, commuteduration was expected to be a stronger predictor of stress than would be commutedistance Commutetime to work was expected to be the strongest predictor of the workarrival stress measure("need to wind down"), whereas commutetime homewas expected to be the strongest predictor of the homestress measure("negative impact on family life"). 3. Following the rationale and results of the Novacoet al. (1991) study, females
commutinga long distance (20+ miles) were predicted to have higher commutingstress than men.- i.e. females would be less satisfied with their commute,be more bothered by traffic
congestion, report a greater need to wind downon arrival at work, and perceive a greater negative impact on their family life. This prediction was tested in a 2 X 2 (distance Xgender) analy,~is of variance. In addition to the interest in replication of the previous research, distance rather than duration is used as the commuting condition factor because it is a more stable attribute of the commute. While drivers may indeed vary their route, commutedistance
fluctuates less than does duration which is affected not only by road conditions but also by ridesharing. 4. Ridesharingis expectedto buffer the stress effects of commuting, especially in the case of long distance commutes(20+ miles), comparing full-time ridesharing to solo driving.
predictions were madefor partial ridesharers. This wasexamined in a 3 X 2 analysis of variance design (commute modeX distance), and post-hoc comparisons(Scheffe tests) of the solo driver and fulltime ridesharer meansfor the stress indices.
RESULTS Consistent with Hypothesis 1, the commuting stress indices are significantly related to the distance and duration attributes of the commute,which are muchmorestrongly associated
with the stress measuresthan are age and income. These correlations are given in Table 1 for the full sample. Table 2 partitions the sample according to automobile commute mode(solo
driving, partial ridesharing, and full-time ridesharing) giving the correlations of the stress measures with miles and minutes to work. It can be seen that for solo drivers the magnitude of each correlation is stronger than the correspondingcoefficient in the full sample, except one for whichit is the same. Except for the part-time ridesharers, the magnitudeof the correlations is generally stronger for females than for males. There is some indication in the set of coeffidents in Table 2 that full-time ridesharing attenuates the correlation between commute attributes and the stress measures, but this is more properly assessed in the ANOVA tests of
group meansreported later The differential effects stipulated in Hypotheses 1 and 2 were tested by multiple
regressions performed with age and incomeas control variables entered on the first step, and then the distance and duration measures entered as predictors in separate equations. For "commute satisfaction", time to workaccounts for 12.2%of the variance (R2 change = .122, T
16.3, df = 3,2157, p < .0001) above that are associated with the covariates of age and income, which together account for 1%. The R2 change effects for time home and for distance
respectively are .108 and .075, whichboth are also highly significant (p < .0001). The effects
follow a similar pattern for the other stress variables regressed with the covariates of age and income: ~oothered by traffic" is most strongly related to time to work (R2 change = .134, T = 18.5, df = .3,2193, p < .0001); *need to wind downon work arrival" is most strongly related to time to work (R2 change = .063, T = 12.2, df = 3,2181, p < .0001); and "negative impact on family life" is most strongly related to time home(R2 change = .057, T = 11.5, df = 3,2176, p < .0001). These findings are supportive of our predictions of differential effects.
Analyses of the effect of distance ("low" versus "high") was examined in a 2x2 ANOVA design with gender. The means, standard deviations, and ANOVA results are presented in
Table 3. The distance effect is very highly significant for all of the stress indices. Significant gender differences men. The interaction on arrival were found only for commuting satisfaction; womenare more satisfied than
showing the moderating influence of gender on the effect produced commutes perceive much greater commuting stress
The hypothesized mitigating influence of ridesharing distance are presented in Table 4. There is a significant "satisfaction", "bothered", and "need to wind-down" indices,
on the
given in the table. [At this time we are not presenting the results of a three-way analysis that included gender because of the complexity of the interactions.] Regarding the two-way analysis
(distaace X mode), because the differences between means on the stress variables are partly due to the part-time ridesharers, with the full-time post-hoc Scheffe tests were performed to compare the solo drivers so as to examine Hypothesis 4. Summingacross compared to the solo drivers, are significantly distance higher in
ridesharers,
commuting satisfaction and less bothered by traffic congestion (p < .05 for both Scheffe tests). Thus, Hypothesis 4 was only partially confirmed.
Determinants of Prospective Ridesharing Wefurther examined the effects endorsement of alternative of distance, gender, and stress on solo drivers"
grouped according to distance ranges (1-6 miles, 7-14 miles, 15-29 miles, and 30+ miles) and gender, whostated that they would "definitely try" the various alternative commuting modes (carpool, vanpool, bus, train). These distance ranges were selected to optimize the distribution of respondents. The survey question, asked for each mode, was "Would you consider
commutingby ~ for 1 or 2 days a week to see if you like it?" The response options were "definitely try," "might try", and "not try" (If commute distance was less than 21 miles, the vanpool question was not asked. As the chi-square tests given in Table 5 indicate, there are significant effects for distance and gender for carpooling and for train/rail. try carpooling and rail modes increases significantly Thedisposition to
especially for women in the very long distance range. In contrast, menin the 30+ mile commutes have a decreased inclination to try the alternative modes, whencomparedto those in the 15-29 mile range. Finding this significant effect for distance on the disposition to try alternative commute modes, we then examinedwhether the experience of stress would add to this inclination. Given the findings of previous research on homeenvironment consequences of commutingstress (Novaco, et al. 1991), the "negative impact on family life" index was of particular interest. Selecting for long distance solo drivers having commutes greater than 15 miles (for comparison with the data in Table 5 and to get a sufficient N for the analysis), this subset of respondents
10
was then partitioned into those reporting "low negative impact" (a recoding of "not at all", "a little", and "somewhat" responses) versus those reporting "high negative impact" (a recoding
"fairly much" and "very much"responses). These low/high grouping were then crosstabulated with the disposition to try commutingalternatives, performed separately for each commute
mode, for each gender, and across genders. In Table 6 are the percentages of respondents in each columncategory whoendorsed the "definitely try" response, separately tabulated for each commute mode alternative. Theeffect of the stress variable is significant for both males and for females in the case of carpooling and for train/raiL The chi-square tests are given in the table. Theefl.:ect of family life impactis especially strong for femaleswith regard to carpooling. Nearly 48%of the women solo drivers in long distance commutes whoperceive that exposure to traffic congestion has a negative impact on their family life indicate that they would definitely try carpoc,ling. This is a considerably greater percentage than that found for the long distance condit!ion itself.
DISCUSSION Wehave found that commuting stress is significantly associated with the distance and
duration of commuting, controlling for age and income. This study then replicates with a large representative sample of southern California commuterssomeof the main research findings of previous research on this topic conducted with small samples in one city. The finding that commuteduration was more strongly related to the stress measures is consistent with the concept of impedance, as developed in the previous work done by Novacoand his colleagues. In other analyses with this data set, we are examiningdegree of impedance in terms of variation in commuteduration at fixed distance points (shorter vs. longer time to travel the same
11
distance), and our preliminary findings are strongly supportive of the stress propositions. This will be addressed in a subsequent paper. The moderating effect of gender still remains to be understood, because we did have
some mixed results regarding our gender hypothesis and because understanding the explanatory factors requires further analysis of the data set. Contrary to our predictions, women over all were more satisfied with their commutesthan were men, although this effect for commute
satisfaction is primarily in the shorter distance condition. In contrast, a number olF our analyses found stronger stress effects for women than for men. In the long distance commutes,women report a greater need to wind downupon arrival at work and perceive greater negative impact on family life. The factors that might explain these effects remain to be examined. Travel
elements of the commute itself, differential sensitivity to commute aversiveness, and role strain are amongthe areas for examination. It is knownthat womenscommute trips tend to be more complexthan those of men, and we need to disentangle variables associated with child care and other householdresponsibilities. Someevidence was found that supported full-time ridesharing as a buffer of commuting stress, but such results occurred only for two of the four stress indices. The failure to find greater support for a ridesharing effect mayin part be a function of the few stress measures used, whichwas determinedby feasibility. It can also be expected that people select into their
commute modesand psychologically adapt to them. Curiously, there were some indications that part-time ridesharers maybe acutely sensitive to commuting stress and maybe unsatisfied with their ability to mitigate it. The characteristics of this group need to be examinedmorefully, particularly as demographic and household variables may be entangled w~ith part-time
ridesharing. The significant effects for higher commuting satisfaction and being less bothered
12
by traffic efforts.
ridesharers
program
to try
consequences. Concern with the quality of family life and it would seem to be efficacious to the psychological stress
theme in
to call attention
13
Refglr~c~
Collier, C. & Christiansen, T. (1992a). The state of the commute in southern California (1991). Transportation Research Record, 1338, 73-81. Collier, C. & Christiansen, T. (1992b). 1992 State of the commute in southern California. Transportation Research Record, paper #930623. Collier, C. & Christiansen, T. (1993). Public transit 1993: Bus, paratransit, and ridesharingo Transportation Research Record, 1390, Giuliano, G., Hwang,K., & Wachs,M. (1992). Mandatorytrip reduction in southern California: First Year results. Manuscriptsubmitted for publication. Liss, So (1991). 1990 Nationwidepersonal transportation survey, early results. Federal Highway Administration, USDOT. Access (University of California Transportation
Novaco, Ro W., Kliewer, W., & Broquet, A. (1991). Home environment consequences of commute travel impedance. American Journal of CommuniW P sycholowy, ~ 881-909. Novaco, R. W., Stokols, & Milanesi, L. (1990). Objective and subjeaive determinants of travel impedance as determinants of commuting stress. American journal of Community
231-257.
Pisarski, A E. (1987). Commuting in America: National report on commutingpatterns and
trends. Eno Foundation. Wachs, M. & Giuliano, G. (1992). Compulsoryridesharing in Los Angeles. Access (University of California Transportation Center), 1, 13-14.
14
t !
m.
t
~=
m i
t t
t t
e~
A i
~z
<
~E
--5
~
)!
=L)
m
t
m.
t
t%
a~
O
om u
v
tl,
)!
!
t I
t~
~0
I
~o
I t !
f,q
m i
m.
|
8
V
0 0
E
0 ,..t
e-
8 0
<
IIi
b~
b~
~f
~ ~l
ii
~0
~t
w
{3%
! o
-|; 0
~P
A" o e o
V
t
_.~
..
o~
c~
~C~v 0 0
0 0
S
0
.-~ ~l"~
~j
=~
o~
Zo
Zo
~,~.~
QJ
~te~m
II .a~
~2
$7_~
II .-., I~
"O ~
._~ N
O,,1~
LV .
~.~v ~
~ii
..~ Z.~
N~e . ~
~I
~ ~ ~
..~ o~
~8~s
II :~ (~1
b~
~P
L~
b,
L~ b,,
.=
0
eM!
~.= ~~
b,
L%
I:%
u
~a qa
"-
~.i q
~ ~
L~
b,
C~
.~ ..=% ~
o~ m
0
0
Oo
L)
;>
E--
co
~0
c~
t~ ~0
u ~m m