Click for next page ( 13


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
13 CHAPTER THREE SURVEY RESULTS Twenty-six states responded to the survey for this synthesis. 30 Days" (9). Still, almost 20% of respondents indicated that The vast majority of responses (23 of 26) came from DOTs or it could take from 91 to 364 days to be entered into their the equivalent. One response was received from a highway statewide system ("Less than 1 year"). Two respondents indi- safety office within the state police agency, a second came cated that it takes more than a year to enter a crash into their from an office of highway safety within the department of pub- system. Again, it is likely that some of those respondents that lic safety, and the final response did not specify the agency. refer to their DOT system as not receiving crash data for The survey responses reflect crash record systems in place more than 90 days (e.g., New York reports more than a year), and/or under development during the early summer of 2004. have crash data readily available from that state's custodial Figure 3 shows the geographic distribution of the responding agency at an earlier time. states, which are shaded on the map. Table 1 gives an over- view of their crash experience. Question 3 asked respondents whether all crashes meet- ing the state threshold are collected and entered into the crash As seen in Figure 3 and Table 1, the responding states are records system. Twenty-two of 26 states (85%) responded from all areas of the country and represent a broad distribu- "Yes." Three states (i.e., California, Connecticut, and Ore- tion of crash experience. It is of interest to note that the data gon) stated that not all reportable crashes are entered in the in Table 1 also came from different years of crash records. system and one did not answer. The oldest data came from Washington State (which plans to have 19962002 data available soon). As of this writing, only Question 4 was a compound question. The first part of 3 of the 26 states had made their 2003 data easily accessible the question asked if users are able to obtain reports from by the public. Because these data are derived from crash the system. The second part asked how this is accomplished. summary data available on the Internet at a particular point Figure 5 summarizes the answers to these questions. in time, it is likely that some of the other states either have 2003 data available for internal use or they have a policy Because it is possible for a system to support more than against providing crash data on the Internet. one level of reporting, the data in Figure 5 show the cumula- tive totals for all answers from each respondent. No single respondent reported that users are unable to obtain reports SURVEY RESPONSES from their crash records system. Twenty of the 26 systems, almost 77%, support ad hoc queries specified by the user and Question 1 of the survey asked respondents whether their 17 (65%) indicated that their crash systems support prede- responses applied to an existing system, a new system cur- fined "canned" reports. Of the 17, only 3 states indicated that rently under implementation, or to a planned future system. canned reports were the highest level of reporting available; Twenty-one of the 26 respondents reported that their answers the remaining 14 crash systems supported both predefined described a current system. Four of the remaining five respon- and ad hoc reporting. Four states have systems designed to dents indicated that they were describing a new system that support users by having them submit requests to trained ana- was currently under implementation. One respondent (Col- lysts. In three of these four cases, this was the only way for orado) said that its responses describe a planned system for users to obtain a report. At least some of the states that which funding is already in place. Because almost 90% of the reported having to submit report requests, such as Iowa, have respondents were from state DOTs, in some cases (e.g., Mis- a university-based center that actively supports these requests. sissippi) there is an existing crash database at the custodial agency, but a separate state DOT crash system of linked data Question 5 asked respondents to indicate whether road- is currently under development. way, vehicle, driver, emergency medical service (EMS), and other sources of data can be linked with the crash data. Question 2 of the survey asked respondents how long it Because it is possible for a system to include linkage to more takes (from the date of the incident to final data entry) to enter than one external data source, Figure 6 shows the cumulative a crash into their central crash database. Figure 4 shows the totals for all answers from each respondent. Twenty of the distribution of answers. As may be seen in the figure, the most 26 respondents (77%) indicated that their crash database has frequent responses were "Within 90 Days" (10) and "Within links to roadway data. This was more than double the number

OCR for page 12
14 FIGURE 3 Geographic distribution of responding states. TABLE 1 CRASH DATA FOR STATES RESPONDING TO THE SURVEY Total Year of State Total Fatal Injury PDO Fatalities Injuries Casualties Data Arizona 134,228 974 46,209 87,045 1,119 74,230 75,349 2002 Arkansas 70,904 557 28,125 42,222 641 52,474 53,115 2002 California 522,562 3,517 201,478 317,567 3,926 305,907 309,833 2001 Colorado 96,990 595 26,208 70,187 655 38,283 38,938 2002 Connecticut 82,787 319 34,448 48,020 343 51,129 51,472 2000 Delaware 20,408 118 6,021 14,269 137 9,967 10,104 2001 Hawaii 10,848 133 6,125 4,590 140 8,620 8,760 2001 Idaho 26,700 261 9,661 16,778 293 14,601 14,894 2003 Illinois 438,990 1,273 87,458 350,259 1,420 127,719 129,139 2002 Iowa 64,361 394 23,763 40,204 445 36,031 36,476 2000 Kansas 78,271 449 18,495 59,327 511 27,059 27,570 2002 Kentucky 130,347 810 32,393 97,144 915 49,329 50,244 2002 Louisiana 160,991 791 48,800 111,400 902 82,800 83,702 2003 Maine 37,251 153 11,538 25,713 165 16,415 16,580 2000 Maryland 104,843 606 38,875 65,362 661 59,517 60,178 2002 Mississippi 91,687 786 24,228 66,673 871 37,174 38,045 2003 Missouri 94,623 822 27,376 66,425 922 42,298 43,220 2002 Montana 23,529 232 6,479 16,818 269 10,083 10,352 2002 New York 306,050 1,431 172,174 132,445 1,554 259,143 260,697 2001 Nevada 62,237 330 20,475 41,432 381 31,522 31,903 2002 Oregon 48,282 388 18,679 29,215 436 27,791 28,227 2002 S. Carolina 108,280 949 32,427 74,904 1,053 52,095 53,148 2002 Virginia 154,848 860 55,041 98,947 942 78,842 79,784 2003 Washington 51,474 318 22,298 28,858 360 34,178 34,538 1996 W. Virginia 49,913 405 16,859 32,649 444 25,788 26,232 2002 Wisconsin 129,072 723 39,634 88,715 805 57,776 58,581 2002

OCR for page 12
15 12 of linkages reported for any other data source; however, it is 10 not unexpected because the respondents were predominantly 10 from the state DOTs. The next most frequent linkages cited 9 were to the vehicle and driver data files. 8 No. of Responses Seven respondents reported linkages to sources of data 6 "other" than those cited in the question. These were identi- 5 fied in the survey responses as: 4 Linkages to annual average daily traffic volume data (3), 2 2 A link to citation data (1), A link to hospital discharge data (1), 0 A link to their bridge inventory (1), and Within 30 Within 90 Less than 1 Over 1 year No source of linked data identified (1). Days Days year FIGURE 4 How long it takes (from the date of the crash) for a The following states reported more than one linkage in report to be entered into a traffic records system. their survey response: 25 20 20 17 No. of Responses 15 10 5 4 0 0 No user reports Request reports Pre-defined reports Can run my own ad- come from the system from trained come from system hoc queries analyst/programmer FIGURE 5 How easily can users obtain reports from the system? How does this process work? 25 20 20 No. of Responses 15 10 9 8 7 5 4 3 1 0 Roadway Vehicle Driver EMS Other None No answer FIGURE 6 Other sources of safety data that are linked to the system. EMS = emergency medical service.

OCR for page 12
16 Colorado--Vehicle and driver. Fourteen of the 26 states that responded to the survey, or Iowa--Roadway, vehicle, and driver. almost 54%, are using GIS and/or GPS map-based systems Maryland--Roadway, vehicle, driver, and EMS. with coordinates. These systems include reading a GPS to Missouri--Roadway and vehicle. obtain coordinates either automatically or manually at a crash Mississippi--Roadway, vehicle, driver, and EMS. site, GIS locator routines to identify a site by pointing to a Nevada--Roadway, vehicle, driver, EMS, and citations. map, and after-the fact locating of a crash on a map based on New York--Roadway, vehicle, and driver. the officer's description of the location. Twelve crash record South Carolina--Roadway, vehicle, driver, EMS, and hos- systems use location-coding schemes based on reference pital discharge. posts or mile markers placed on roadsides, and 16 systems Virginia--Roadway, vehicle, driver, and bridge inventory. are using a document-based mile point and calculated dis- placement methodology for locating a crash. It is interesting to note that there are many more docu- mented linkages of data available in these reporting states Question 7 asked respondents to specify what percentage (e.g., CODES projects); however, the survey respondents did of crashes is located reliably in their system. The answers not report the additional linkages that might be available. ranged from 50% to 100%, with the median response at 94%. Even within the DOTs, for example, Missouri's report of a The mean response was 88.7%. Two respondents reported roadway linkage refers not to a single roadway characteristic that the percentage of crashes reliably located was unknown. file, but rather to their comprehensive transportation manage- In general, the crash records systems that identify locations ment system. This enterprise-wide system is a GIS-based data based on one of the methods of obtaining coordinates are per- system supporting their activities with extensive information ceived as more accurate and descriptive than the crash sys- about traffic, pavement, safety, bridges, and travelways. tems using traditional field and document-based methods. Being able to conduct spatial analysis with crash and related Question 6 asked respondents to tell us what location cod- data with a GIS was cited as a considerable advantage to ing methods are used in their systems. The answers are sorted using the coordinate-based location method. into three basic categories: Question 8 asked respondents for an estimate of the cost to develop their crash records system. Thirteen respondents 1. Locations based on posted locations in the field (e.g., reported a total cost for developing their statewide crash mileposts), records system. Although it was not possible to determine the 2. Document-based systems that assign a calculated loca- system elements included in the total cost, the mean cost of tion code (e.g., mile point, log point), and crash systems reported was just over $861,000. The median 3. Locations in which a latitude and longitude are col- cost was $500,000. The difference in these two measures lected by GPS or a GIS map is used to pinpoint the indicates that some outliers likely affected the mean--in this location. case, one system came in at $3,500,000 and two systems were near $2,000,000. The remaining 10 systems reported Figure 7 summarizes the responses to Question 6. Because much lower costs. Excluding the three multimillion dollar it is possible for a system to include more than one location systems, the mean cost of the systems was approximately coding method, this figure shows the cumulative totals for all $390,000. This estimate is much closer to the median value answers from each respondent. of $400,000 for these 10 systems, indicating that for these 25 20 No. of Responses 16 15 14 12 10 5 0 GIS/GPS Map-Based Posted in Field Document Based FIGURE 7 Location coding methods used.

OCR for page 12
17 lower-priced systems, the measure of central tendency is not for all answers from each respondent when summarized in overly affected by outliers. four broad categories. Twelve of the respondents reported that they did not know Data collection--Ten of the 24 respondents mentioned the cost of developing their systems. Missouri reported a data collection as a feature they like about their crash cost for their enterprise-wide system of $24 million, which records system. Three of these respondents mentioned included numerous data files linked to their crash data. This electronic transfer of crash data into their system and cost was not included in the averages cited previously, as eight of the respondents were particularly pleased with there was no obvious way to apportion the cost of the crash the data edits and quality control in their system. component of the system. Recent costs for large crash sys- Management--Nine of the 24 respondents favorably tems that were not reported in this survey include the Texas mentioned the management and maintenance of their crash system, which is expected to cost approximately $9 mil- crash records system. Of these, five spoke of the bene- lion and the Indiana crash system that has cost approximately fits of their document management system and five spoke $5.5 million. of the ease with which the crash records system could be managed. Question 9 asked respondents for the cost of collecting Linkage--Ten of the 24 respondents were pleased with and entering crash data into their crash records system. Eleven the ability of their crash records systems to link with of the 26 states reported this cost, with 3 providing a cost per other components of the traffic records system. Four crash. For the other eight, cost per crash data were calculated of the 10 respondents were particularly pleased with using the summary data (annual total cost) of the system and their ability to use location or GIS as a means of link- an estimate of the total number of crashes based on the data ing data with roadway and other inventories. Other data reported in Table 1. The costs ranged from a low of $1.53 in components mentioned as linked to crash records sys- California to a high of $38.85 in Washington State. Costs for tems included driver, vehicle, EMS, and hospital dis- Washington State and Oregon (at $19.88/crash) were by far charge data. the highest reported, with the next highest cost per crash reported at $7.61 (estimated for Missouri). It should be noted Analysis and reporting--Approximately two-thirds of that the discrepancies in crash costs reported could be the the respondents (15 of 24) believe that the best feature result of many factors, including inconsistencies in the cost of their crash records system is the ease with which they components counted as part of the estimate, methods used to can do analysis and reporting of the data. The reporting compute the component costs, and actual differences in the responses include query capability, canned reports, ad costs of labor and other items in the various locales. These hoc reports, and exporting of data to other systems. costs fall within the average of $21.00 per crash calculated in the 1998 crash cost study described in chapter two. Question 11 asked respondents what they would change about their crash records system if they could start over. Question 10 asked respondents what features and capa- Twenty-three of the states responded to this open-ended bilities they like about their crash records systems. Twenty- question by listing one or more features. Because most states four of the states responded to this open-ended question by indicated that there were several features that they would like listing one or more features. Because most respondents indi- to change about their systems, the data in Figure 9 show the cated that there were several features that they liked about cumulative totals for all answers from each respondent when their system, the data in Figure 8 show the cumulative totals summarized in four broad categories. 18 16 15 14 No. of Responses 12 10 10 10 9 8 6 4 2 0 Data Collection Management Linkage Analysis & Reporting FIGURE 8 What features and capabilities do you like about your crash records system?