8. Experiments in the Interest of Electric Railway Service
The problem of securing competent motormen for urban rail systems is evidenced by the number of accidents that occur as a result of their performance, as determined by the courts: liability payments made as a result of accidents in which railway employees are at fault (as opposed to the actions of a pedestrian or driver who was struck by a train) constitute 13% of the gross earnings of some railway companies, resulting from up to fifty thousand accident indemnity cases each year.
A commission to investigate the cause of the problem noted that accidents were frequent at certain hours of the day, and among crews who had been working a certain number of hours - or in effect, that the crews were too fatigued to give adequate attention to their work. In psychological terms, it has to do with attentiveness.
While it stands to reason that a tired or overworked motorman may become careless, there were also instances in which accidents occurred with relatively fresh crews. As such, attention it is not merely the diminished ability to give attention as a result of overwork - but the inability to give attention may be a mental capacity even at the start of a shift.
The physics of motion and the limitations of the machinery are largely fixed elements: a rail car of a certain weight travelling at a certain speed cannot be brought instantly to a halt, given the quality and condition of the braking mechanism. But one factor that can be attributed to human operator is the reaction time it takes to apply the brake - and this can be significant.
When a playing child or an inattentive adult moves suddenly onto the tracks, a tenth of a section in reaction time may decide his fate. An inattentive brakeman will be slow to recognize the need to initiate the necessary action, and will be the cause of more accidents. It's noted that lethargic individuals do not fare well in this position - but even among average individuals, reaction times differ.
A closer analysis of the situation follows: driving an electric train through busy city streets is extremely challenging, as there are all manner of people and vehicles who are constantly in motion - moving onto the tracks in front of a moving train from the right or the left, is a manner that is often unpredictable. One solution is to drive the car more slowly, but this is unsatisfactory as drivers are required to keep to a timetable - drivers who fail to do so are sacked. But driving at an adequate speed to keep to the timetable means that there is less time to react when a pedestrian or vehicle crosses in front of the train.
(EN: Perhaps I'm in a pessimistic mood, but I strongly suspect that management was unreasonable - demanding that drivers make their rounds quickly while driving the trains slowly, and blaming the conductor for being able to do what is plainly impossible, sacking some to threaten others. And if ever there was any accommodation to plan timetables in a manner that would allow the drivers to travel at safe speeds, it was likely done by a clerk who had never operated a train.)
He speaks of a failed experiment in which models and sketches were used to illustrate and test the capacity to avoid accidents. This was routine done in the naval courts, which used small ship models on a table in the courtroom - but the problem is that when operating a ship, a helmsman doesn't have that birds-eye view and full knowledge of the relative position of the vessels - as such the use of models in a laboratory environment is inaccurate and frustrating to participants.
He elaborates in some detail an experiment designed to be more accurate that created a board of 26 squares, used numbers to indicate objects in the street (1 is a pedestrian, 2 is a horse, 3 is an automobile) that moved a number of squares that reflect their velocity. And measured how many "moves" it took for a motorman to recognize that one of those objects were going to cross the tracks in front of the train. (EN: In the present day, we might use a computer simulation to do this - but what strikes me is that this is simply less inaccurate and lies on its own set of assumptions, and the outcome will be tainted by flaws in the model.)
The test was performed with a group of motormen who had twenty or more years of service and whose records were practically without accident, and this was compared to the results of another group of men who were less experience and whose records included "many more or less important" collisions or accidents, as well as a third group whose performance was neither especially good nor especially bad.
The results were calculated as the number of errors the subject would make in suggesting which objects would or would not cross the tracks, as well as the speed at which the test would be completed. As a result, he identified a number of behaviors characteristics that could be correlated to performance: a good motorman pays attention to both sides of the tracks, predicts the speed and direction of multiple objects, gives attention to objects that are both close and further away, etc.
His suggestion in the end is applying this very same test to men who seek to gain employment as motormen, and eliminating those candidates who make too many mistakes or take too long to complete the test. IT is his sense that spending the time to apply this test could significantly reduce the amount of accidents.
(EN: And while this suggestion seems reasonable, the author does not provide any proof. If he were to complete the case study by citing an actual reduction in accidents by a railway that utilized his screening test, the case would be far more convincing.)