The purpose of this study was to evaluate the validity of data entered on the new State of Michigan Traffic Crash Report form, or UD-10, and the reliability of data entered from this form into the Michigan State Police (MSP) data base. In order to check the validity of crash data reported on the new UD-10, data from eighteen crashes reported on UD-10s were compared to data from a second report on the same crashes, collected by independent investigators for the University of Michigan In-depth Vehicle Occupant Report (UMIVOR). Comparisons were made on 63 data items in 20 crashes for a total of 882 data entries, with UMIVOR used as the validity check and the UD-10 compared to it. Universal agreement between the two reports was found on 34 data items. The remaining 29 items had at least one or more crashes in which UD-10 disagreed with UMIVOR or had missing data. In these 29 data items there were 101 data entries with disagreement or missing data, or 11.4 percent of the total data entries. In order to check the reliability of crash data entered into the Michigan State Police data base, data from the same eighteen UD-10s described earlier were compared to printouts of crash data from the Michigan State Police data base for these UD-10s. The majority of the disagreements occurred in areas of the UD-10 where data was coded by filling in a "bubble" alone (77.6 percent) or by filling in a "bubble" and printing the data (1.6 percent) for a total of 79.2 percent of the disagreements. At this time it is not clear how the errors occurred in reading the optically scanned part of the UD-10. The remaining errors, 20.8 percent, occurred in data printed on the UD-10 and incorrectly entered into the MSP data base. (A)
Samenvatting