Dewar, R. E. & Ells, J. G. (1974). Methods for the evaluation of traffic signs. Final report, submitted to Ministry of Transport.

(Conclusions & recommendations.) The series of experiments comprising the present research project was formulated to develop and compare several methods for evaluating the perception of traffic signs. The major conclusions based on the results of this research are as follows.

Legibility distance of traffic signs can best be determined by having them viewed by the subject while driving under normal conditions. Only this technique should be used to obtain criterion measures against which to validate other methods of sign legibility. The speed at which the subject drives the vehicle is of relatively little importance in determining legibility distance.

A modified on-the-road technique, using miniature traffic signs viewed by the subject driving at a reduced speed has been found to predict legibility distance under normal driving conditions very well. It would be an adequate field method for traffic sign evaluation.

Time lapse photography of traffic flow is another valuable tool for recording drivers' responses to signs. Marked variations occur in traffic flow and behavior patterns from day to day, and from one location to another. Therefore, care must be taken to record an adequate sample of driving behavior.

Comprehension, or intrinsic meaning, of unfamiliar symbols can be adequately measured by the use of either a multiple choice questionnaire or an open-ended question format. In the case of the former method, care should be taken to construct all erroneous choices so that they are perceived as plausable answers, otherwise a 4-choice format may in effect be a 2 or 3-choice format. Under such circumstances the probabilities of making a correct guess increase.

The multiple-choice format, with choices related to action to be taken or meaning (whichever is appropriate to the individual sign) permits an adequate measurement of comprehension of ambiguous messages.

The semantic differential technique provides an index of subjective meaning of traffic sign symbols. This measure (all four semantic differential factors) is highly correlated with intrinsic meaning and would be a valuable adjunct in the measurement of meaning of unfamiliar symbolic messages.

The population stereotypes technique is useful for determining existing expectations with regard to symbolic representation of messages. Such information would be helpful in guiding the design of symbolic traffic signs.

Unfamiliar symbolic messages with little intrinsic meaning are difficult to learn and remember. Some symbols, after being learned, are forgotten within a few weeks. This illustrates the need for an extensive education campaign whenever a new symbol is introduced into a signing system.

Subjective indices of preferences for symbolic messages and judgements of clarity of their meaning provide little useful information in the evaluation of traffic sign symbols.

Glance legibility has been used in traffic sign research by a number of experimenters in the past. However, the present research indicates that this measure is unrelated to legibility distance. Therefore, its validity as a tool for evaluating traffic sign perception must be questioned.

Verbal reaction time is a good method for determining the speed with which subjects can extract the meaning from a traffic sign. This laboratory method is a valid predictor of legibility distance of verbal signs, but in the case of symbolic messages a valid prediction can be obtained only under conditions which involve the type of attention demands and visual distractions found in the driving task.