2003 National Truck Fleet and Bus Fleet Safety Surveys: Final Report Contact: Richard Gruberg richard.gruberg@dot.gov (202) 366-2959 Analysis Division, Office of Research and Analysis Federal Motor Carrier Safety Administration U.S. Department of Transportation Publication No. FMCSA-RI-06-044 July 2006 Contents Executive Summary ii Introduction 1 Background 1 Methodology 4 Truck Survey 4 Bus Survey 5 Results: Out-of-Service Rates 7 Truck Survey 7 Bus Survey 10 Additional Data Analyses 13 Association Between Driver and Vehicle OOS Violations 13 Most Common OOS Violations 14 Discussion 16 Acknowledgments 18 Appendix A. Methodology Used To Produce Weighted National Estimates 19 Data Tables 1. 2003 National Truck Fleet Safety Survey: Inspections Requested and Achieved by State and Location Type 4 2. 2003 National Bus Fleet Safety Survey: Sites Selected and Sample Sizes Achieved by State 6 3. 2003 National Truck Fleet Safety Survey: Unweighted OOS Rates by State and Type of Inspection Site 7 4. 2003 National Truck Fleet Safety Survey: Weighted OOS Rates and 2003 MCMIS OOS Rates by State 8 5. Unweighted and Weighted National Estimates of OOS Rates from the 2003 National Truck Fleet Safety Survey and 2003 MCMIS Data 8 6. 2003 National Bus Fleet Safety Survey: Unweighted OOS Rates by State and Inspection Location 11 7. Unweighted and Weighted Estimates of OOS Rates from the 2003 National Bus Fleet Safety Survey and 2003 MCMIS Data 12 8. 2003 National Truck Fleet and Bus Fleet Safety Surveys: Association Between Driver and Vehicle OOS Violations 14 9. 2003 National Truck Fleet Safety Survey: Most Common Vehicle and Driver OOS Violations Found 15 10. 2003 National Bus Fleet Safety Survey, Tour/Charter Bus Sample: Most Common Vehicle and Driver OOS Violations Found 15 11. 2003 National Bus Fleet Safety Survey, Intercity Bus Sample: Most Common Vehicle and Driver OOS Violations Found 16 2003 National Truck Fleet and Bus Fleet Safety Surveys: Final Report Executive Summary This report presents findings from the 2003 National Truck Fleet and Bus Fleet Safety Surveys, conducted by the Federal Motor Carrier Safety Administration (FMCSA) during the summer and fall of 2003. In the two studies, nontargeted (random) inspections were used to estimate driver and vehicle out-of-service (OOS) rates for both buses and large trucks regulated by FMCSA, based on Level 1 (i.e., full driver and vehicle) inspections. The "OOS rate" is defined here as the proportion of vehicles operating on the Nation's highways with at least one violation of the FMCSA regulations severe enough to require the vehicle to be placed out-of-service either for a prescribed period of time (in the case of driver-related violations) or until vehicle defects are corrected (in the case of vehicle- related violations). For the 2003 National Truck Fleet Safety Survey, inspections were conducted at fixed weigh stations (permanent facilities) and at various nonfixed (makeshift) inspection sites on major highways. For the 2003 Bus Fleet Safety Survey, inspections were conducted at major tour bus and charter bus destination sites and at intercity bus terminals across the country. For the 2003 National Truck Fleet and Bus Fleet Safety Surveys, 2,835 truck inspections were conducted in 10 States, and 834 bus inspections (651 at destination sites and 183 at intercity terminals) were conducted in 25 States. For the truck survey, data collection began in late summer 2003 and continued into the fall. Data collection for the bus survey began in July 2003 and continued throughout the summer; in a few States, data collection took place in the fall. For the truck survey, States were selected on the basis of geographic location and participation and cooperation in previous studies, so that the major geographic regions of the country would be represented. The 10 States participating in the study were California, Colorado, Connecticut, Georgia, Illinois, Maryland, New Jersey, New Mexico, North Carolina, and Washington. The participating States were requested to perform nontargeted inspections at both fixed and nonfixed locations in order to capture a representative sample of trucks traveling on the Nation's highways. Inspectors were also asked to estimate the daily volumes of truck traffic passing through each site (both fixed and nonfixed locations) for use in the estimation of national OOS rates, where locations with heavier truck volumes were weighted more heavily. For the bus survey, the tour bus and charter bus destination sites and intercity bus terminals initially selected were the same sites that were used in the 2001 National Bus Fleet Safety Survey. Sites that were found to be nonviable in 2001 were eliminated, and substitute locations were used when possible. Destination sites were required to have a volume of at least 16 buses per day in order to be eligible for selection, and intercity terminals were required to have a volume of at least 11 arrivals per day. The estimated vehicle OOS rate from the 2003 truck survey was 28 percent, and the estimated driver OOS rate was 5 percent (weighted national estimates). The estimated vehicle and driver OOS rates from the 2003 survey of tour bus and charter bus destination sites were 26 percent and 3 percent, respectively. For the intercity bus terminal survey, the numbers of buses and drivers in the final sample were considered too small to support national estimates of OOS rates. In both the truck and bus samples, an association was found between the presence of driver OOS violations and the presence of vehicle OOS violations. In the truck survey, inspections that found at least one vehicle OOS violation were nearly two times more likely to find a driver OOS violation than were inspections that found no vehicle OOS violations. Conversely, inspections that found at least one driver OOS violation were more than 1.5 times more likely to find a vehicle OOS violation than were inspections that found no driver OOS violations. In the bus survey, inspections that found at least one vehicle OOS violation were four times more likely to find a driver OOS violation than were inspections that found no vehicle violations, and inspections that found at least one driver OOS violation were nearly three times more likely to find a vehicle OOS violation than were inspections that found no driver violations. Brake-related violations were the most common type of OOS violation found in both the truck and bus surveys, constituting roughly 40 percent of all OOS violations (both vehicle and driver) in the truck sample, 25 percent in the tour/charter bus destination sample, and 35 percent in the bus terminal sample of intercity coaches. The most common driver OOS violations found were "No Record of Duty Status" and "Failure to Retain Logs for 7 Days," which together constituted 48 percent of all driver OOS violations in the truck sample and 70 percent of all driver OOS violations in the bus destination sample. Results from the 2003 truck survey are reasonably consistent with the results of previous surveys. The 1996 National Truck Fleet Safety Survey, which was the most comprehensive OOS rate survey conducted before 2003, produced a weighted national estimate of 29 percent, as compared with the 28-percent weighted rate for vehicle OOS violations in the 2003 survey. The 1996 truck survey also produced a weighted estimate for the national driver OOS rate that was similar to the 5-percent rate in the 2003 survey. The OOS rates estimated from the 2003 truck survey track closely with the rates derived from all Level 1 inspections conducted for FMCSA in 2003 and maintained in the Agency's Motor Carrier Management Information System (MCMIS). The similarity of the results suggests that the routine targeting of vehicles during the inspection process has not, historically, contributed a significant bias to OOS rates calculated from MCMIS inspection data. Likewise, their similarity suggests that conducting inspections primarily at fixed weigh stations does not contribute a significant bias to the MCMIS OOS rates. To the extent that such bias is minimal, the OOS rates calculated from MCMIS Level 1 inspection data accurately reflect the safety fitness of the trucks and drivers that operate on our highways and are regulated by FMCSA. A question that remains unanswered is why both the OOS rates from the National Truck Fleet Safety Survey and the unweighted rates from MCMIS vary dramatically from State to State. One possibility is that the disparities arise from differences in the inspection procedures employed in different States; if this is the case, such disparities may be contributing a bias both to the MCMIS estimates and to the truck survey estimates. The results of the 2003 National Bus Fleet Safety Survey differ somewhat from the results of some of FMCSA's earlier surveys. For the tour/charter bus component, the weighted national average driver OOS rate calculated from the 2003 survey results (3 percent) is consistent with those from the 1999 and 2001 surveys (3 percent and 4 percent, respectively); however, the weighted vehicle OOS rate from the 2003 survey (26 percent) differs from both the 1999 and 2001 results (23 percent and 13 percent, respectively). The weighting methodology used for the tour/charter bus component of the 2003 bus survey was different from, and may be more accurate than, the methodologies used for the 1999 and 2001 surveys. The weighted national estimate of the driver OOS rate from the tour/charter bus component of the 2003 bus survey is comparable with the rate obtained from MCMIS, using all 2003 Level 1 inspection data for buses, motor coaches, and school buses (driver OOS rate of 3 percent in both cases). Although the driver OOS rate from the 2003 intercity bus terminal sample is slightly higher (5 percent), it is an unweighted estimate and should not be viewed as nationally representative. In contrast, the vehicle OOS rates calculated from the 2003 bus survey are considerably higher than the corresponding rates calculated from MCMIS data. The unweighted national estimate of the average vehicle OOS rate for all passenger buses based on 2003 Level 1 inspection data from MCMIS is 15 percent, as compared with a weighted estimate of 26 percent from the 2003 tour bus sample and an unweighted estimate of 23 percent from the intercity bus sample. Introduction This report presents findings from the 2003 National Truck Fleet and Bus Fleet Safety Surveys, conducted by the Federal Motor Carrier Safety Administration (FMCSA) during the summer and fall of 2003. In these studies, commercial motor vehicles were randomly selected for roadside inspections at both fixed and nonfixed locations in order to assess their level of compliance with the Federal Motor Carrier Safety Regulations (FMCSRs). Nontargeted inspections were used to estimate driver and vehicle out-of-service (OOS) rates for both buses and large trucks regulated by FMCSA, based on Level 1 (i.e., full driver and vehicle) inspections. In this report, the OOS rate is defined as the proportion of vehicles operating on the Nation's highways with at least one FMCSR violation severe enough to require the vehicle to be placed out-of-service either for a prescribed period of time (in the case of driver-related violations) or until vehicle defects are corrected (in the case of vehicle-related violations). For the 2003 National Truck Fleet Safety Survey (TFSS), inspections were conducted at fixed weigh stations (permanent facilities) and at various nonfixed (makeshift) inspection sites on major highways. For the 2003 National Bus Fleet Safety Survey (BFSS), inspections were conducted at major tour bus and charter bus destination sites and at intercity bus terminals across the country. Background Since the mid-1950s, the U.S. Department of Transportation (DOT), in conjunction with State enforcement authorities, has routinely inspected commercial motor vehicles in operation on U.S. highways. Currently the inspections are performed as part of FMCSA's Motor Carrier Safety Assistance Program (MCSAP). Inspections take place at permanent and portable truck weigh stations, port of entry facilities, carrier terminals, hazardous waste stations, and other locations. They are conducted in accordance with standards developed by the Agency in cooperation with the Commercial Vehicle Safety Alliance (CVSA). The inspection standards provide national uniform inspection procedures and criteria for identifying OOS violations. Five different kinds of inspection are performed under MCSAP. The most comprehensive is the Level 1 inspection, as used in this study, which involves a full examination of the vehicle and an inspection of all driver records. The results of all MCSAP-sponsored inspections are uploaded into an FMCSA data management system, the Motor Carrier Management Information System (MCMIS). Vehicles are frequently targeted for inspection under the MCSAP program. For example, whenever a cursory auditory and visual examination of a vehicle entering an inspection site reveals potential violations, the vehicle is inspected. In addition, many MCSAP inspectors have remote access to FMCSA's Inspection Selection System (ISS) software package, which allows them to enter a carrier's U.S. DOT Number and receive a recommendation as to whether the vehicle should be inspected, based on the carrier's inspection history and safety record. Consequently, the totality of inspections maintained in MCMIS are not a random sample and, generally, do not provide a reliable basis for calculating "true" OOS rates for the national truck and bus fleets regulated by the Agency. Beginning as far back as 1969, representatives of the trucking industry have alleged that published DOT data on roadside inspections are not representative of the true state of vehicles on the road. In response, DOT began conducting controlled studies to determine whether its published OOS rates differed substantially from rates based on a statistical sample of trucks. The first such study, implemented during the summer of 1971, involved a random sample of 1,172 trucks. The study found that 28 percent of the trucks inspected had OOS violations. In 1995, DOT decided to reassess the baseline OOS rate for the national truck fleet by means of a random sample. A total of 11 States participated in the study, which was implemented in the summer of 1996. The 1996 survey involved more than 10,000 truck inspections, including some 1,300 inspections of trucks carrying hazardous materials (HM). The estimated OOS rates from the study were 5 percent for drivers and 29 percent for vehicles. The truck fleet survey was repeated in 1998, but only State-level estimates were produced from the survey data at that time. Because only 4 percent of the data came from inspections in the Northeast, the sample was not considered nationally representative. Subsequently, truck inspection data from the 1998 survey were used to derive unweighted estimates of 27 percent and 7 percent for national vehicle and driver OOS rates, respectively. In 1999 and again in 2001, the Agency conducted similar surveys for buses, using both tour/charter bus destination sites and intercity bus terminals to perform inspections. The 1999 survey, involving 295 bus inspections at destination sites and 142 inspections at intercity bus terminals, yielded vehicle and driver OOS rates for tour buses of 23 percent and 3 percent, respectively, and vehicle and driver OOS rates for intercity buses of 18 percent and 9 percent, respectively. The 2001 survey, which involved 611 bus inspections at destination sites and 151 inspections at intercity bus terminals, yielded vehicle and driver OOS rates for tour buses of 13 percent and 4 percent, respectively, and vehicle and driver OOS rates for intercity buses of 18 percent and 0.3 percent, respectively. As indicated by these data, the results of FMCSA's bus fleet safety surveys have been less consistent over time than the results of its truck fleet surveys. For the 2003 TFSS and BFSS, 2,835 truck inspections were conducted in 10 States, and 834 bus inspections (651 at destination sites and 183 at intercity terminals) were conducted in 25 States. For the truck survey, data collection began in late summer 2003 and continued into the fall. Data collection for the bus survey began in July 2003 and continued throughout the summer; in a few States, data collection took place in the fall. Methodology Truck Survey For the 2003 TFSS, States were selected on the basis of geographic location and participation and cooperation in previous studies, so that the major geographic regions of the country would be represented. The 10 States participating in the study were California, Colorado, Connecticut, Georgia, Illinois, Maryland, New Jersey, New Mexico, North Carolina, and Washington. The participating States were requested to perform nontargeted (random) inspections both at weigh stations (permanent facilities) and at nonfixed (makeshift) inspections sites on major highways in order to capture a representative sample of trucks traveling on the Nation's highways. Inspectors were also asked to estimate the daily volumes of truck traffic passing through each site (both fixed and nonfixed locations) for use in the estimation of national OOS rates, where locations with heavier truck volumes were weighted more heavily. Table 1 shows the numbers of inspections requested and achieved in each State. Table 1. 2003 National Truck Fleet Safety Survey: Inspections Requested and Achieved by State and Location Type State Inspections Requested Inspections Achieved Fixed Sites Nonfixed Sites Total Fixed Sites Nonfixed Sites Total California 200 150 350 409 30 439 Colorado 100 60 160 100 88 188 Connecticut 150 80 230 156 147 303 Georgia 150 80 230 255 242 497 Illinois 150 80 230 179 85 264 Maryland 150 80 230 * * 236 New Jersey 100 80 180 119 83 202 New Mexico 100 60 160 152 132 284 North Carolina 200 150 350 77 74 151 Washington 50 80 230 175 96 271 Total 1,350 900 2,350 1,622 977 2,835 *For Maryland, inspection data could not be separated by type of site. Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet Safety Survey. As shown in Table 1, FMCSA did not receive estimates of daily traffic volumes at the truck inspection locations in Maryland. In addition, roughly one-half of the truck inspection data received from New Jersey did not include traffic volume information. Therefore, OOS rates calculated from the survey data for those two States were not weighted by truck volume. National estimates of vehicle and driver OOS rates were calculated by taking a weighted average of all the estimated OOS rates from the participating States, using 2003 estimates of large truck vehicle miles traveled (VMT) in each State as the weighting factors. Bus Survey A total of 648 inspections at tour/charter bus destination sites throughout the country were performed for the 2003 BFSS, as well as 183 inspections at intercity bus terminals. The tour bus destination sites and intercity bus terminals initially selected were the same sites selected for the 2001 bus fleet survey. (For the 2001 survey, separate master lists of tour bus destination sites with at least 16 buses per day and intercity bus terminals with at least 8 buses arriving per day had been developed. Based on those master lists, random samples of tour bus destination sites and intercity bus terminals were selected for the 2001 survey.) Before the 2003 BFSS was implemented, FMCSA division offices throughout the country were asked to review the master list of tour bus destinations developed in 2001 and provide feedback on the viability of the sites listed in their States. Tour bus sites not expected to yield 16 buses were categorized as nonviable. Based on the feedback, a small number of site substitutions were made for previously selected sites that appeared nonviable in 2001. For each of the tour bus destinations selected in their States, MCSAP inspectors were instructed to inspect at least 16 buses. Table 2 shows the number of survey sites selected for the study, by State, and the number of inspections achieved at each site. For both the truck and bus surveys, unweighted and weighted estimates of driver and vehicle OOS rates were produced. In addition, the data were used for various exploratory analyses. Table 2. 2003 National Bus Fleet Safety Survey: Sites Selected and Sample Sizes Achieved by State State Site Name Number of Inspections Achieved Tour Bus/ Charter Bus Destination Sites Intercity Bus Terminals AZ Grand Canyon 30 X CA Disneyland 16 X CA Yosemite 13 X CA Los Angeles 19 X CO Denver 40 X CT Foxwood 18 X DC National Mall 11 X DC DC Terminal 7 X FL Disneyworld 2 X FL Universal Studios 5 X FL Seaworld 9 X GA Six Flags 16 X GA Atlanta 8 X IL Six Flags 9 X LA Bossier Casino 16 X MD Six Flags 16 X MI Greektown Casino 6 X MI Detroit 8 X MN Metrodome 17 X MN Minneapolis 7 X MO Branson 25 X NJ Six Flags 52 X NJ Atlantic City 115 X NJ Atlantic City 8 X NY New York City 13 X NY Flushing 5 X NY Port Authority 8 X NY Albany 8 X NY Buffalo 8 X OH Cedar Point 20 X OH King's Island 17 X OH Cleveland 8 X RI Providence 8 X SC Hilton Head 21 X SD Mt. Rushmore 10 X TN Grand Ole Opry 40 X TX Six Flags 40 X TX Dallas 22 X TX McAllen 8 X TX Houston 8 X UT Bryce National Park 35 X WA Seattle Airport 17 X WI Miller Stadium 17 X WI Milwaukee 8 X WY Yellowstone 37 X Source: Federal Motor Carrier Safety Administration, 2003 National Bus Fleet Safety Survey. Results: Out-of-Service Rates Truck Survey Table 3 shows the raw unweighted OOS rates for each State participating in the 2003 TFSS, for both fixed and nonfixed inspection locations. Large disparities in the unweighted OOS rates can be seen across States. For example, Connecticut and Georgia have rather high unweighted OOS rates, whereas California's are considerably lower. It is not clear whether such discrepancies are a result of actual differences in the safety fitness of the vehicles and drivers from State to State, or can be explained by variations in inspection procedures in different States. Table 3. 2003 National Truck Fleet Safety Survey: Unweighted OOS Rates by State and Type of Inspection Site State Drivers Vehicles Fixed Sites Nonfixed Sites Fixed Sites Nonfixed Sites Sample Size Driver OOS Rate Sample Size Driver OOS Rate Sample Size Vehicle OOS Rate Sample Size Vehicle OOS Rate CA 409 2% 30 0% 409 21% 30 7% CO 100 2% 88 5% 100 22% 88 22% CT 156 12% 147 6% 156 46% 147 57% GA 255 8% 242 5% 255 37% 242 48% IL 179 5% 85 6% 179 41% 85 18% MDa — — — — — — — — NJ 119 0% 83 2% 119 29% 83 24% NM 152 6% 132 3% 152 34% 132 36% NC 77 5% 74 5% 77 24% 74 26% WA 175 3% 96 0% 175 28% 96 31% aFor Maryland, inspection data could not be separated by type of site. Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet Safety Survey. There are also large discrepancies in the OOS rates calculated for fixed and nonfixed sites in several States, including Connecticut (both driver and vehicle), Georgia (vehicle only), Illinois (vehicle only), and California (vehicle only). In the case of California, the extremely low vehicle OOS rate for the nonfixed sites may be due to the small sample size (30 inspections). For the three other States, possible explanations for the disparity are less apparent, although small sample sizes may again play a role. Weighted OOS rates for each State, using the estimated truck volume at each site as the weighting factor, are shown in Table 4, along with the 2003 Level 1 OOS rates from MCMIS for comparison. State-level weighted estimates from the survey track fairly closely with the MCMIS Level 1 OOS rates in most cases. As noted for Table 3, however, in some cases there are wide disparities in OOS rates from State to State. Weighted national estimates of driver and vehicle OOS rates from the 2003 TFSS are shown in Table 5. For comparison, calendar year 2003 Level 1 driver and vehicle national OOS rates based on MCMIS data are also shown. To produce the estimates, the following steps were followed: Table 4. 2003 National Truck Fleet Safety Survey: Weighted OOS Rates and 2003 MCMIS OOS Rates by State State Drivers Vehicles Fleet Survey MCMIS Fleet Survey MCMIS Sample Size Driver OOS Rate Driver OOS Rate Sample Size Vehicle OOS Rate Vehicle OOS Rate CA 439 1% 2% 439 15% 21% CO 188 5% 7% 188 26% 34% CT 303 14% 13% 303 52% 49% GA 497 9% 11% 497 35% 40% IL 264 7% 8% 264 42% 38% MDa 236 7% 6% 236 34% 34% NJa 202 1% 4% 202 27% 31% NM 284 6% 6% 284 32% 28% NC 151 7% 5% 151 31% 19% WA 271 4% 3% 271 28% 24% aFor Maryland and New Jersey, only unweighted estimates are available. Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet Safety Survey. Table 5. Unweighted and Weighted National Estimates of OOS Rates from the 2003 National Truck Fleet Safety Survey and 2003 MCMIS Data Source of National Estimate Useable Observations Driver OOS Rate Vehicle OOS Rate All 2003 TFSS Data, Unweighted 2,834 drivers 2,831 vehicles 5% 33% 2003 TFSS Data, Weighted 2,407 drivers 2,402 vehicles 5% 28% 2003 MCMIS Data 1,020,076 drivers 1,020,076 vehicles 5% 29% Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet Safety Survey and Motor Carrier Management Information System. 1. OOS rates were first calculated for each inspection site used in the study (sites with fewer than eight inspections or with zero OOS violations were pooled together). 2. Next, the total number of trucks passing through each site in an "out-of-service" condition was estimated by multiplying the percentage of trucks found to have OOS violations during the inspection period by the estimated truck volume for each location. 3. Next, the OOS rate for each State was estimated by summing the estimated numbers of trucks passing through each of the sites in an "out-of-service" condition (step 2, above) and dividing by the estimated total number of trucks passing through all sites, as shown in Table 4. 4. The weighted national estimate was then produced by taking a weighted average of the State OOS rates estimated in step 3, above, using estimates of large truck VMT in each of the States as the weighting factors. A more detailed description of the methodology used to produce the weighted, nationally representative estimates is provided in Appendix A. The weighted national estimates from the 2003 TFSS are reasonably consistent with the 2003 Level 1 OOS rates obtained from MCMIS. The similarity of the TFSS and MCMIS results suggests that neither the targeting of vehicles for MCSAP inspections nor conducting the inspections primarily at fixed weigh stations (the primary source of inspection data feeding into MCMIS) contributes a significant bias to the OOS rates obtained from the MCMIS file. A question that remains unanswered is why both the weighted OOS rates from the TFSS and the unweighted rates from MCMIS vary dramatically from State to State. For example, driver and vehicle OOS rates from California are consistently low, whereas driver and vehicle OOS rates from Connecticut are consistently high. One possibility is that the disparities arise from differences in the inspection procedures employed in different States; if this is the case, such disparities may be contributing a bias both to the MCMIS estimates and to the TFSS estimates. Bus Survey Table 6 shows the raw unweighted OOS rates for each inspection location in the 2003 BFSS. Overall estimates from the study, both unweighted and weighted (nationally representative), are shown in Table 7, along with the 2003 Level 1 OOS rates for buses from MCMIS for comparison. In the case of the bus terminal component of the sample, weighted estimates are not shown, because the sample size was too small to yield reliable weighted estimates. State-level estimates were not produced for the bus study. For the truck study, State-level estimates were used to weight the data in order to arrive at a national estimate. Weighting by State VMT was considered necessary in the truck study because of the wide variation in truck VMT by State; for example, the estimate of 2003 large truck VMT in California is more than double that for any other State in the study. In the case of the bus study, however, variations in tour/charter bus VMT among the participating States were assumed to be less significant than variations in large truck VMT. In addition, State-level data on bus VMT are not readily available. Nationally weighted estimates were produced for the tour/charter bus sample only, using the following steps: 1. OOS rates were first calculated for each inspection site used in the study. 2. Next, the total number of buses passing through each site in an "out-of-service" condition was estimated by multiplying the percentage of buses found to have OOS violations during the inspection period by the estimated bus volume (daily bus count) for each location. 3. The overall OOS rate was then estimated by summing the estimated numbers of buses passing through each of the sites in an "out-of-service" condition (step 2, above) and dividing by the estimated total number of buses passing through all sites, based on the daily bus counts. Table 6. 2003 National Bus Fleet Safety Survey: Unweighted OOS Rates by State and Inspection Location State Inspection Location Daily Volume Vehicles Inspected Vehicle OOS Rate Drivers Inspected Driver OOS Rate AZ Grand Canyon 29 29 17% 30 0% CA Disneyland 50 15 7% 16 0% CA Yosemite 38 12 0% 13 0% CA Los Angeles Terminal 59 19 0% 0 — CO Denver Terminal 22 39 33% 11 0% CT Foxwood Casino 75 18 11% 18 17% DC National Mall 200 11 36% 11 0% DC DC Terminal 60 7 57% 7 0% FL Disneyworld 50 2 0% 2 0% FL Seaworld 50 5 20% 5 0% FL Universal 50 9 11% 9 0% GA Six Flags 60 16 50% 16 25% GA Atlanta Terminal 60 8 37% 8 13% IL Six Flags 9 9 89% 9 22% LA Bossier City 16 16 13% 16 0% MD Six Flags 25 16 25% 16 0% MI Greektown Casino 6 6 33% 6 0% MI Detroit Terminal 28 8 38% 8 0% MN Metrodome 7 17 6% 17 0% MN Minneapolis Terminal 21 7 57% 7 0% MO Branson 23 25 16% 25 4% NJ Atlantic City 900 115 29% 115 3% NJ Morey's Pier 20 3 0% 3 0% NJ Six Flags 50 52 19% 52 0% NJ Atlantic City Terminal 58 8 13% 8 0% NY NYC Attractions 100 13 23% 13 0% NY Flushing 25 5 0% 5 0% NY Port Authority 360 8 0% 8 13% NY Albany Terminal 65 8 63% 8 0% NY Buffalo Terminal 42 8 13% 8 0% OH Cedar Point 27 20 5% 20 0% OH King's Island 12 17 12% 17 12% OH Cleveland Terminal 26 8 25% 0 — RI Providence Terminal 31 8 0% 8 0% SC Hilton Head 25 21 0% 21 0% SD Mt. Rushmore 8 10 30% 10 0% TN Grand Ole Opry 50 40 13% 5 0% TX Six Flags 18 40 25% 21 14% TX Dallas Terminal 72 22 0% 0 — TX McAllen Terminal 71 8 0% 8 25% TX Houston Terminal 21 8 50% 8 13% UT Bryce Canyon 10 35 23% 35 0% WA Sea-Tac Airport 17 17 0% 17 0% WI Brewer's Stadium 17 17 0% 17 0% WI Milwaukee Terminal 15 8 13% 0 — WY Yellowstone 20 37 5% 37 3% Source: Federal Motor Carrier Safety Administration, 2003 National Bus Fleet Safety Survey. Table 7. Unweighted and Weighted Estimates of OOS Rates from the 2003 National Bus Fleet Safety Survey and 2003 MCMIS Data Source of Estimate Useable Observations Driver OOS Rate Vehicle OOS Rate Bus Fleet Safety Survey Tour/Charter Bus Sample Unweighted (All Data) 597 drivers 648 vehicles 3% 19% Weighted National Estimate 597 drivers 648 vehicles 3% 26% Intercity Bus Samplea Unweighted (All Data) 97 drivers 182 vehicles 5% 23% 2003 MCMIS Level 1 OOS Rates: Motorcoaches, Buses, School Buses 3,123 inspections 3% 15% aNo weighted national estimates were produced for this category due to the limited sample size. Sources: Federal Motor Carrier Safety Administration, 2003 National Bus Fleet Safety Survey and Motor Carrier Management Information System. As shown in Table 7, the driver OOS rates from the tour/charter bus component of the survey are similar to those calculated from all 2003 Level 1 bus inspections in MCMIS: a 3-percent driver OOS rate was estimated both for the tour/charter bus sample and for the 2003 MCMIS Level 1 inspections of buses, motor coaches, and school buses. A 5-percent driver OOS rate (unweighted) was estimated for the intercity bus component of the survey; however, the estimate should not be considered nationally representative. Vehicle OOS rates calculated from the 2003 BFSS data are considerably higher than those calculated from Level 1 MCMIS inspection data: 26 percent for the tour bus sample and 23 percent (unweighted) for the intercity bus sample, as compared with 15 percent for the Level 1 MCMIS inspection data for buses, motorcoaches, and school buses. Comparison of the results from Table 5 and Table 7 shows that both the weighted and unweighted driver and vehicle OOS rate estimates for the tour/charter bus sample are similar to, but slightly lower than, those for the truck sample in the 2003 fleet surveys. The weighted driver OOS rate estimated from the tour/charter bus sample is 3 percent, compared with the estimate of 5 percent from the truck sample; and the weighted vehicle OOS rate from the tour/charter bus sample is 26 percent, compared with 28 percent from the truck sample. Only unweighted estimates of OOS rates were produced for the intercity bus sample because of the limited number of inspections performed; however, the unweighted driver OOS rate for the intercity bus sample (5 percent) is the same as the weighted estimate for the truck sample. The unweighted vehicle OOS rate for the intercity bus sample is 23 percent. Additional Data Analyses In addition to the estimation of driver and vehicle OOS rates, several other exploratory analyses were performed on the truck and bus survey databases. Findings from two of these analyses—the association between diver and vehicle OOS violations and the most common OOS violations found—are discussed below. Association Between Driver and Vehicle OOS Violations In the truck sample and the tour/charter bus sample from the 2003 fleet safety surveys, associations were found between vehicle and driver OOS violations (Table 8). In the truck study, 7 percent of the inspections that found at least one vehicle OOS violation also found at least one driver OOS violation, whereas only about 4 percent of the inspections that found no vehicle OOS violations found at least one driver OOS violation (based on unweighted data). Thus, trucks with at least one vehicle OOS violation were nearly twice as likely as trucks with no vehicle OOS violations (7/4) to have a driver OOS violation. Conversely, 49 percent of the inspections that found at least one driver OOS violation also found at least one vehicle OOS violation, whereas only 32 percent of the inspections that found no driver OOS violations found at least one vehicle OOS violation. Hence, trucks with at least one driver OOS violation were more than 1.5 times as likely as trucks with no driver OOS violations (49/32) to have a vehicle OOS violation. Similarly, in the bus study, 8 percent of the inspections in the tour/charter bus sample that found at least one vehicle OOS violation also found at least one driver OOS violation, but only 2 percent of the inspections that found no vehicle OOS violations found at least one driver OOS violation (based on unweighted data). Thus, tour/charter buses with at least one vehicle OOS violation were four times as likely as buses with no vehicle OOS violations (8/2) to have a driver OOS violation. Conversely, 47 percent of the inspections that found at least one driver OOS violation also found at least one vehicle OOS violation, but only 18 percent of the inspections that found no driver OOS violations found at least one vehicle OOS violation. In this case, buses with at least one driver OOS violation were nearly three times more likely to have a vehicle OOS violation than were buses with no driver OOS violations (47/18). Table 8. 2003 National Truck Fleet and Bus Fleet Safety Surveys: Association Between Driver and Vehicle OOS Violations Source of Estimate Driver OOS Rate Vehicle OOS Rate Inspections With No Vehicle OOS Violations Found (Unweighted) Inspections With at Least One Vehicle OOS Violation Found Inspections With No Driver OOS Violations Found (Unweighted) Inspections With at Least One Driver OOS Violation Found Truck Survey 4% 7% 32% 49% Bus Survey Tour/Charter Bus Sample 2% 8% 18% 47% Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet and Bus Fleet Safety Surveys. Most Common OOS Violations The most common OOS violations found in the 2003 truck and bus safety surveys are listed in Table 9 (trucks), Table 10 (tour/charter buses), and Table 1l (intercity buses). Because similar violations were found in the three samples, the results are discussed together. In both the truck and bus studies, brake-related violations were the most common type of OOS violation, accounting for 39 percent of all OOS violations (both vehicle and driver) in the truck survey sample (Table 9), 26 percent of all OOS violations in the tour/charter bus sample (Table 10), and 36 percent of all OOS violations in the intercity bus sample (Table 11). The most common driver OOS violations found were "No Record of Duty Status" and "Failure to Retain Logs for 7 Days," which together constituted 48 percent of all driver OOS violations in the truck study (Table 9) and 70 percent of all driver OOS violations in the tour/charter bus sample (Table 10). Table 9. 2003 National Truck Fleet Safety Survey: Most Common Vehicle and Driver OOS Violations Found Violation Code Violation Description Percentage of Total Violations Found Vehicle All Vehicle OOS Violations All OOS Violations 396.3A1BA Brake Out of Adjustment 20% 19% 396.3A1B Brakes (General) 8% 8% 396.3 (a) (1) Inspection, Repair, and Maintenance 8% 8% 393.48 (a) Inoperative/Defective Brakes 8% 7% 393.47 Inadequate Brake Lining 6% 5% Driver All Driver OOS Violations All OOS Violations 395.8A No Record of Duty Status 25% 2% 395.8K2 Failure to Retain Logs for 7 Days 23% 2% 393.23A2 No Valid Commercial Driver's License 8% <1% 395.3A1 10 Hour Rule Violation 8% <1% 395.8E False Record of Duty Status 7% <1% Source: Federal Motor Carrier Safety Administration, 2003 National Truck Fleet Safety Survey. Table 10. 2003 National Bus Fleet Safety Survey, Tour/Charter Bus Sample: Most Common Vehicle and Driver OOS Violations Found Violation Code Violation Description Percentage of Total Violations Found Vehicle All Vehicle OOS Violations All OOS Violations 396.3 (a) (1) Inspection, Repair, and Maintenance 14% 13% 396.3A1BA Brake Out of Adjustment 14% 12% 396.3A1B Brakes (General) 12% 11% 393.61 (b) (2) No/Defective Emergency Exits 8% 7% 393.25 (f) Stop Lamp Violations 4% 4% 393.51 Defective Brake Warning 4% 3% Driver All Driver OOS Violations All OOS Violations 395.8K2 Failure to Retain Logs for 7 Days 40% 4% 395.8A No Record of Duty Status 30% 3% 393.23A2 No Valid Commercial Driver's License 15% 1% 391.11B5 Commercial Driver's License From More Than One State 10% 1% Source: Federal Motor Carrier Safety Administration, 2003 National Bus Fleet Safety Survey. Table 11. 2003 National Bus Fleet Safety Survey, Intercity Bus Sample: Most Common Vehicle and Driver OOS Violations Found Violation Code Violation Description Number of Violations Found Percentage of Total Violations Found Vehicle 396.3A1BA Brake out of Adjustment 18 31% 396.3A1 Inspection, Repair, and Maintenance 5 8% 396.3A1B Brakes (General) 3 5% 393.209D Steering System 3 5% 393.61B Window Escape Inoperative 3 5% 393.25F Stop Lamp Violations 3 5% Driver 391.11B5 Commercial Driver's License From More Than One State 4 7% Source: Federal Motor Carrier Safety Administration, 2003 National Bus Fleet Safety Survey. Discussion Results from the 2003 TFSS are reasonably consistent with the results of previous surveys. The 1996 truck fleet survey, which was the most comprehensive OOS rate survey conducted before 2003, produced a weighted national estimate of 29 percent for the vehicle OOS rate, as compared to 28 percent in the 2003 TFSS. The 1996 truck survey also produced a weighted estimate for the national driver OOS rate that was similar to the 5-percent rate in the 2003 survey. The OOS rates estimated from the 2003 TFSS track closely with the rates derived from all 2003 Level 1 inspections in the MCMIS database. The similarity of the results suggests that the routine targeting of vehicles during the inspection process has not, historically, contributed a significant bias to OOS rates calculated from MCMIS inspection data. Likewise, their similarity suggests that conducting inspections primarily at fixed weigh stations does not contribute a significant bias to the MCMIS OOS rates. To the extent that such bias is minimal, the OOS rates calculated from MCMIS Level 1 inspection data accurately reflect the safety fitness of the trucks and drivers that operate on U.S. highways and are regulated by FMCSA. A question that remains unanswered is why both the OOS rates from the 2003 TFSS and the unweighted rates from MCMIS vary dramatically from State to State. For example, the unweighted vehicle OOS rate from more than 300 TFSS inspections conducted in Connecticut in 2003 was 52 percent, as compared with an unweighted vehicle OOS rate of 27 percent from more than 200 TFSS inspections in New Jersey. One possibility is that the disparities arise from differences in the inspection procedures employed in different States; if this is the case, such disparities may be contributing a bias both to the MCMIS estimates and to the truck survey estimates. More research is needed to determine whether variations in inspection protocols from State to State are in fact contributing to the large disparities. The results of the 2003 BFSS differ somewhat from the results of some of FMCSA's earlier surveys. For the tour/charter bus component, the weighted national driver OOS rate calculated from the 2003 survey results (3 percent) is consistent with those from the 1999 and 2001 surveys (3 percent and 4 percent, respectively); however, the weighted vehicle OOS rate from the 2003 survey (26 percent) differs from both the 1999 and 2001 results (23 percent and 13 percent, respectively). In addition, the vehicle OOS rates calculated from the 2003 BFSS are considerably higher than the corresponding MCMIS rate for all passenger buses. The vehicle OOS rate based on 2003 Level 1 bus inspection data from MCMIS is 15 percent, as compared with a weighted estimate of 26 percent from the 2003 tour bus sample and an unweighted estimate of 23 percent from the intercity bus sample. The weighting scheme used for the tour/charter bus component of the 2003 BFSS differs from the methodology that was used for the 1999 and 2001 surveys. For the 1999 and 2001 studies, it was assumed that the master list of tour bus destination locations (the sampling frame) was comprehensive. As a result, traditional sampling weights were developed, based on a stratified sampling design, using geographic region (Northeast, South, Midwest, and West) as the basis for stratification. That approach assumed that the sampling frame was not missing any popular tour bus destinations, and that the destination locations listed on the frame for each stratum were similar in terms of the types of buses they attracted. To the extent that either assumption was incorrect, the resulting estimates may have been inaccurate. The weighting scheme used for the tour bus destination locations in the 2003 BFSS is based on a simpler methodology (simple random sampling from an infinite population) that makes fewer assumptions about the sampling frame or about the similarity of the locations being sampled. Acknowledgments Under FMCSA's Motor Carrier Safety Assistance Program (MCSAP), the truck and bus inspection data used for these studies were collected by State and local commercial motor vehicle inspectors in Arizona, California, Colorado, Connecticut, the District of Columbia, Florida, Georgia, Illinois, Louisiana, Maryland, Michigan, Minnesota, Missouri, New Jersey, New Mexico, New York, North Carolina, Ohio, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Utah, Washington, Wisconsin, and Wyoming. Participation of States in the 2003 National Truck Fleet and Bus Fleet Safety Surveys was entirely voluntary, and FMCSA extends its appreciation to these states. Cleaning of the inspection data was performed under contract by IBM Business Consulting Services. IBM also performed many of the analyses discussed in this report, including analyses of the most common violations found and the calculation of driver and vehicle OOS rates for the bus study. OOS rates for the truck study were calculated by FMCSA analysts. Appendix A Methodology Used To Produce Weighted National Estimates This appendix describes the methods used to compute the weighted, nationally representative OOS rates for the 2003 National Truck Fleet and Bus Fleet Safety Surveys, including: (1) estimation procedures for the truck survey, (2) estimation procedures for the tour/charter bus destination site component of the bus survey, and (3) estimation procedures for the intercity bus terminal component of the bus survey. Truck Survey For the truck survey, OOS rates were first estimated for each State, using the following formula: (1) Ri = ( ? Vi,j × Ri,j ) / ? Vi,j , where: Ri = the estimated percentage of trucks in State i that are in out-of-service condition, Ri,j = percentage of inspections from site j in State i that resulted in an out-of- service order, and Vi,j = the estimated 8-hour truck volume for site j in State i. Next, an overall national OOS rate was estimated, using a weighted average of the State rates calculated in Equation (1) above, with data on truck VMT in 2003 as the weighting factor: (2) R = ( ? VMTi × Ri,j ) / ? VMTi , where: R = estimated national percentage of trucks that are in out-of-service condition, and VMTi = estimated truck vehicle miles traveled in State i during 2003. Bus Survey For the bus survey, the procedure for producing the weighted OOS rates for tour/charter buses was similar to that used for the trucks, except that weighting by State VMT was not used. The procedure is shown in the following formula: (3) R = ( ? Vj × Rj ) / ? Vj , where: R = the estimated national percentage of buses that are in out-of-service condition, Rj = the percentage of buses from site j that are in out-of-service condition, and Vj = the estimated daily bus volume for site j. v 20