Results are reported for a recent study to identify similaritiesamong the financial, infrastructure, and operational structures of the state highway systems, and to group states according to these similarities. The purpose of the study was to classify states according to similar systems and problems, rather than according to closeness or geography. An additional goal was to determine the degree to which states that are widely separated may have similar problems, thereby suggesting political alliances for various issues. A structure for classifying state highway agency financing is first proposed and described. Then, a data base containing 61 data items for each stateis developed using cross-sectional data from 1984, 1985, and 1986. The source of these data items is primarily "highway statistics", supported by organization studies in the transportation literature. Data consist of measures of size, road and bridge condition, taxes, revenues, disbursements, and agency characteristics. After a review ofdescriptive rankings of states on a number of key variables, the data are then factor-analyzed using varimax rotation. This procedure yields a smaller number of data items found to most clearly separate these states. States are then clustered according to their ratings on these variables. Clusters based on the full data set containing 61variables are compared with clusters based on reduced data sets containing 19, 13, and 7 variables. Results show that with 19 variables, the state highway systems can be quite accurately clustered into five distinct groups as follows: (1) alaska; (2) the far western states, along with michigan, hawaii, florida, west virginia, and northern new england; (3) southern new england and seaboard middle-atlantic; (4) midwestern and southern states, including washington; and (5) very large states (california, texas, new york, and pennsylvania). Within each of these groups, a number of subgroupings identify strongregional and content coalitions. The group structure appears to be reasonably robust under a variety of assumptions, with regional subgroupings particularly strong for new england, the southeastern seaboard, the far west, and big states. It is concluded that while nearest neighbor geographies may be useful for many political liaisons, state highway agencies should also look at broader similarities with states that are not immediate or nearest neighbors. This paper appears in transportation research record no. 1283, Transportation systemsplanning and applications 1990.
Samenvatting