• Lane-changing or turning-movement behavior, and • Segment-level interactions with roadway attributes such as curvature. How such interactions can be modeled to provide key insights is an endogenous issue as those interactions relate to data. To take advantage of multiscale perspectives, for example, infor- mation is gathered at the microlevel on driver behavior through naturalistic data whose level of measurement occurs on a second- by-second basis (2). When one collates this information with mea- surements in the space traversed by the vehicle the driver operates, key questions emerge about how the driving context should be framed dimensionally. While information on the dynamic vehicle level is collected at a microlevel, roadway and roadside information is collected inconsis- tently; that is, where changes occur in the attribute being measured, a new measurement is recorded. A “change” is a perceptible differ- ence recorded at discrete intervals, whereas, in the truest sense of continuous information, measurement of highway attributes should occur at an inch-by-inch or foot-by-foot level. In the meantime, with the available data and the scales at which they occur, the highway engineer or planner is faced with significant decision challenges as they pertain to estimating the societal cost of crashes, the associated reduction and prevention schemes, and the justifications required by legislatures for prioritized investments. It is this confluence of macro- and microchallenges that moti- vated this paper. The goal of this paper is to provide some retrospec- tion on what the transportation community can do commonly with data, what the prospective value of such data is, and what future data might aid insights into the development of robust crash models and associated risk management enterprise schemes. The Washington State Department of Transportation (WSDOT) has embarked on approaches to enterprise risk management (ERM) for better consoli- dation of its liability, financing, and economic assessment of highway programs. Therefore, the authors hope that this paper will provide useful insight for highway agency planners and engineers on key method- ological issues to consider from a data perspective before they embark on full-fledged data collection schemes geared to their respective risk management agendas. The rest of this paper is organized as follows: first comes a dis- cussion of the data available in Washington State about divided highways, including the methodological issues associated with the database. A statistical model of crash severity distributions is presented as an example; the model presentation is followed by a Data-Driven Perspective on Management of Safety Risk at State Agencies Case Study in Washington State John C. Milton, Venky Shankar, Ming-Bang Shyu, Sittipan Sittikariya, and Ram Pendyala 1 Nationally, transportation agencies have embarked on efforts to collect information digitally on highway attributes to help understand factors that contribute to traffic crash occurrences. Instrumented vehicles, data- base modeling efforts, and enhancements in crash-data collection are salient examples of such efforts. This paper provides insights into the prospective value of roadway information as it pertains to statistical analy- sis of severity of crashes. It presents a case study analysis from Washing- ton State that involves divided highway crash data. A statistical model is presented that demonstrates an empirical relationship between key road- way variables and distributions of crash severity. The other notable out- put of this paper involves the contribution of weather information to the distributions of crash severity. While the case study is restricted to divided highways in the northwest part of the United States, the statistical insights from the analysis of severity distributions indicate the prospec- tive value of key data elements in relation to their regular measurement and updates to statewide crash risk management. Nationally, highway agencies have embarked on systematic efforts to collect data digitally on the roadway and roadside network to aid in the understanding of factors contributing to crashes (1). The basic motivating idea behind such systematic efforts is to help provide for a collective source of insights into a variety of components underly- ing these crashes. These factors include driver behavior in a variety of contexts that involve • Roadway and environmental interactions, • Roadside crash severity impacts due to the presence of fixed objects, • Roadside terrain characteristics and interactions with vehicle maneuvering, • Intersection attributes and interactions with vehicle maneuvering such as car following, J. C. Milton, Washington State Department of Transportation, P.O. Box 47418, Olympia, WA 98504. V. Shankar, Department of Civil and Environmental Engi- neering, Pennsylvania State University, 201 Old Main, University Park, PA 16802. M.-B. Shyu, Mirai Associates, Inc. 11410 Northeast 122nd Way, Suite 320, Kirkland, WA 98034-6927. S. Sittikariya, DKS Associates, Inc., 719 Second Avenue, Suite 1250, Seattle, WA 98104-1728. R. Pendyala, Department of Civil and Environmental Engineering, Room ECG 252, Arizona State University, Tempe, AZ 85287-5306. Corresponding author: J. C. Milton, miltonj@wsdot.wa.gov. Transportation Research Record: Journal of the Transportation Research Board, No. 2083, Transportation Research Board of the National Academies, Washington, D.C., 2008, pp. 1–8. DOI: 10.3141/2083-01