ONGOING UWMADISON DVCRELATED RESEARCH Some Preliminary Results UWMadison

ONGOING UW-MADISON DVC-RELATED RESEARCH Some Preliminary Results UW-Madison Pyle Center February 4, 2003

TWO STUDIES • Evaluation of Warning Sign Locations and Nearby DVC Patterns • Development of Wisconsin County. Level DVC Prediction Models

SIGN/CRASH STUDY • DVCs Near Signs have Never been Systematically Considered • This is Not a Study of Crossing Sign Effectiveness • Assumption: Signs are in Locations with Higher than Typical DVC Numbers • Five WI Counties Chosen after a DVC and Carcass Removal Data Evaluation • Considered 3 Years of DVC Data near 72 Signs Grouped within 30 Roadway Segments

SIGN/CRASH STUDY (Cont. ) • Segment Group Definition: Ø The Ends of Each Segment were 2 Miles from the Last Signs in Each Direction Included within a Group Ø All the Signs in a Roadway Segment Group were within 2 Miles of Each other or the End of the Segment

SIGN/CRASH STUDY (Cont. ) • Preliminary Analysis Conclusions: Ø DVCs per Mile and DVC Rate (per VMT) Found to be Statistically Higher Between Signs Ø 28 Groups had Higher than State DVC Average Rate Between Signs Ø “Between” Sign Rate was Approximately 4 Times Higher than the State Average and “Outside” Rate was 2 Times Higher than the State Average • Preliminary Thoughts on Interpretation

COUNTY MODEL STUDY • At Least Two County Models Exist in Other States • Data from 12 WI Counties Eliminated from Use Lack of Data Ø Unusual Annual Variability in Reported DVCs Ø Deviations from Typical County and Statewide Variability Ø Deviations from Typical County and Statewide Trends Ø

COUNTY MODEL STUDY (Cont. ) • Data from 54 Counties used in Development and 6 for Ongoing Validation • Illinois Model Applied and Determined to be Inappropriate for Wisconsin • Three WI Models Developed with Almost 20 Input Variables Considered along with Their Interactions and Higher Order Forms

COUNTY MODEL STUDY (Cont. ) • Some of the County Input Variables Considered: Ø Deer Density and Populations Ø Human Density and Populations Ø Traffic Volume (VMT and AADT) Ø Roadway Length and Density Ø Acreages: Recreational Areas, Farmland, Woodland Timberland Ø Average Season Snow Depth

COUNTY MODEL STUDY (Cont. ) • County Models Developed: Ø DVCs per Land Area Ø DVC Rate (per VMT) Ø DVCs per Year Ø R 2 = 0. 64 to 0. 82

COUNTY MODEL STUDY (Cont. ) • Some Variables with Significance: Ø Deer Density Ø Human Density Ø VMT Ø Roadway Density Ø Snow Depth Ø Timberland/Woodland/Recreational Acreages • Models Might be Useful to Predict County & State Numbers and Determine Impacts of Physical Changes

CONCLUSIONS • Research Projects are Additional Benefit of UW-Madison Clearinghouse • Research Work will be Finalized Soon • Expected that Results will be Presented within Region/Nationally and Shared on Website • Students Graduating with Safety Expertise in this Area another Benefit of UW-Madison Clearinghouse

QUESTIONS AND DISCUSSION