A conceptual approach to estimating the rlsk or rear-end collisions at a signalized Intersection Is presented. It ls argued that the creation or a large option mne Increases the range or the Indecision mne, the direction lmpllcatlon or which ls an Increase In the rlsk or rear-end collisions. With the aid or field data collected for two warning Intervals (3 and 6 sec) before the red llght, a large option zone ls shown to Increase the variance underlying the stopping probablllty curve, and thus to determine a larger range for the Indecision mne. Data from urban Intersections support the basic argument that a long warning period causes a significant Increase In the number of rear-end colllslons.

At present, a general consensus appears to exist in the literature about the effect of traffic signals on rear-end collisions. Most of the researchers apparently have concluded that signalizing an intersection significantly increases the number of rear-end collisions. For example, in a sample of 34 urban intersections Hakkert and Mahalel (1) found that after the introduction of a traffic signal control, the annual number of rear-end collisions increased from 33 to 77." In a similar study of 31 intersections in Milwaukee, Short et al.(2) found an increase of 37 percent in the number of such accidents. King and Goldblatt (3) observed the same phenomenon of increased rear-end collisions in a statistical analysis of US accident data nationwide. In addition to the fact that the number of rear-end accidents increases after the introduction of traffic signals at an intersection, it is typical that the highest number of accidents at signalized intersections are rear-end collisions. A statistical analysis conducted by the author of almost all signalized intersections in Israel (600) indicated that, over a 2-year period (1983 to 1985), about 39 percent of all accidents were rear-end collisions, compared with about 27 percent that were rightangle collisions. A similar result was identified by Hanna et al.(4) in a study of signalized-intersection accidents in rural communities in Virginia. They found that 43 percent of all accidents were rear-end collisions and 37 percent were right-angle collisions.