Early in December, Peter Jacobsen wrote in discussion groups concerned with cycling safety  that "traffic safety education for children does not work ... . very intensive, real-world, on the street training alters children's behavior only slightly," citing the paper "Prevention of Pedestrian Injuries to Children: Effectiveness of a School Training Program," by Rivara, et al.  Jacobsen uses this to argue that we must traffic calm our streets if we are to reduce accidents. "We, as bicyclists, could do a lot for the safety of bicyclists by actively promoting traffic calming."
The implication of Jacobsen's words is that children cannot be taught how to ride properly. I have just read the referenced paper. The content of the study does not fit his either his description or its implication. The study was not "very intensive, real-world, on the street training," it considered largely children too young for cycling, yet even so it showed considerable improvement in behavior.
The children in the main test group were in kindergarten to grade three. Some students from grade four were used as auxiliary teachers for the younger ones.
The program was done twice, the first time without parental involvement, the second time (in different schools) with parental involvement.
The paper describes the curriculum for K-3 as follows. "A 15-minute introduction, six 30- to 40- minute lessons, and a final follow-up assembly in which the main street-crossing procedures were reviewed and prizes presented. Students received instruction on the basic pedestrian safety skills during lessons 1 through 4. These included learning to recognize and avoid pedestrian hazards; making eye contact with a driver; crossing at a blind spot; crossing at corners; and identification of traffic signs, signals, and safe walking zones. After being taught the basic skills, children practiced in an outdoor setting during the final two lessons. This included videotaping and peer critiquing of performance, and practice using real and simulated streets."
The evaluation consisted of pre and post observations of the children's behavior on the streets near their school upon leaving school. 
Only four items were observed:
1: Did the child walk in the roadway or on the sidewalk (or shoulder if no sidewalk)?
2: Did the child stop before crossing the street?
3: Did the child look right-left-right before crossing?
4: Did the child continue to look while crossing?
1: The program with parental involvement performed much better than the program without. Substantially all children used the sidewalk before the studies so no improvement could exist.
2: Stopping before crossing showed minor changes (3 improvement, 1 degradation) that were not statistically significant, and averaged about 1/3 of the times.
3: Looking before crossing showed minor, statistically insignificant, improvement without parental involvement, and approximate doubling, from about 1/4 to about 1/2, with parental involvement.
4: Continuing to look showed an approximate doubling without parental involvement and an approximate tripling with parental involvement, in each case starting from a base of about 1/6.
In my experience, children in grade three are the youngest that can learn the first stages of proper cycling, appropriate for 2-lane residential roads, when taught in groups. The students of grade 3 level whom I taught consisted of some that had just completed grade 2 and some who had just completed grade 3. I noticed a distinct average difference in their ability to learn, and concluded that some maturation that was highly desirable for traffic cycling was occurring at about that age. Since the second skill that I taught was entering the roadway, which requires the same ability as a pedestrian crossing the roadway, I think it likely that children below grade 3 are likely to find it difficult to learn that skill when taught in groups.  However, experience has shown that when taught individually by well-informed family members, children of age 6 can learn traffic-cycling skills. The considerable difference in success of Rivara's program without and with parental involvement is another indication of the value of individual instruction at these ages.
So far as the relevance of this study to training of cyclists is concerned, most of the subjects of this study were too young for any meaningful inferences to be drawn.
The only way in which the word "intensive" can properly be applied to such a program is that it was an intensive waste of time. Over two-thirds of the program was devoted to lecturing, and much of the lecturing was about abstract concepts that are far above the students' mental capacity. One example, crossing a street at a blind spot. I don't know how to explain in words in class the concept of safe sight distance to six-year-olds. More time is wasted on concepts that are both intellectually difficult and useless, such as making eye contact with drivers. What rule can you give about that subject? Look the driver in the eye and step off the curb in front of him?
When I was considering how to teach cycling to eight-year-olds, I worked out what behaviors they had to exhibit, what knowledge they had to have to be able to exhibit that behavior (exhibition of the correct behavior is skill), and how to teach that knowledge to them.  It turns out that little of the knowledge is used in verbal form, and that much of that which could be taught to adults in verbal form cannot be taught to young children because they have had no experience that allows them to connect the words with the real world.
The children need practice, practice, practice in real traffic to develop the appropriate skills. When teaching traffic cycling to grade three students, I teach them only three things: ride on the right side of the road; when approaching a road that is bigger than the one they are on, yield to traffic on that road; when moving laterally on the roadway, yield to traffic you must cross. Teaching the students to perform those operations with very few errors requires fifteen hours of practice. Conservatively, I would say that teaching the only one of those skills that pedestrians require for crossing the road would require about five hours.
Rivara's program gave only four hours of instruction and wasted two thirds of that time on ancillary subjects.
One peculiar point is that for some observation times more children continued to look than the number that looked once: at one time, 37% continued to look whereas only 21% had started to look, and 32% vs 30% at another time. The only reasonable inference is that children who looked only once were counted as Looking, while children who looked more than once were not counted as Looking but were counted as Continuing to Look. Therefore, the numbers of Look and Continuing to Look must be added to discover the number who looked. However, when one does that, in one case the pair of numbers (62% and 53%) add to more than 100%. Neither assumption works all the time, so that the true situation is unknown.
When parents were involved in the training, the proportion of looking rose from about 1/6 to 1/2 for grades K-1, from about 1/10 to 1/4 for grades 2-4. That is not satisfactory for safe performance, but it shows that the training, such as it was, made a significant improvement.
Children in K-1 grades showed a much greater tendency to stop, both before and after the program, than did children in 2-4 grades. Both groups showed some increase in stopping after the program, but the improvements were not statistically significant, although for the older children, the rate improved from 21% to 33%. 
Strictly speaking, Rivara's program should be evaluated only for what it was, a program for teaching very young children to operate safely as pedestrians. However, since this study has been advanced as an argument in cycling affairs, it is necessary to also evaluate its relevance to cycling affairs.
Children of the ages tested are obviously walking about their neighborhoods and to school. Testing children of these ages is obviously appropriate. The number of subjects was probably insufficient, since changes of 50% in an activity (stopping) were not statistically significant. The observing and scoring system had faults that produced irreconcilable statistics in some cases (looking and continuing to look), leaving the actual state of affairs unknown and unknowable.
However, the strongest criticism should be placed on the training program, its relevance to the actual requirements, its suitability for young children, and its relevance to the behavior that was tested. At these ages, the most frequent type of traffic accident that is caused by the child itself is crossing the street without yielding to traffic on the street. The behavior that needs to be taught is how to yield to crossing traffic. All the significant testing was devoted to evaluating this effect. 
Groups of children in the third grade can be taught to do this extremely well (95% or better) even when riding bicycles. This program should have achieved an equal level of performance, at least with the children of grades 2-4. It did not because it did not devote sufficient time to training the children to cross the street properly, while spending the majority of its time on other subjects. Furthermore, it is very likely that the teaching was not only misdirected but too abstractly designed. Children of these ages cannot learn about traffic from words, and probably they can learn only a little from pictures. Older persons, with some experience of traffic, can learn from words and from pictures, because they already have experience with the subject being discussed or shown. Children of this age must learn from the actual experience because they have no mental experiences through which to apply the discussion or even the pictures.
Learning how to yield to crossing traffic requires the development of judgement regarding speed and distance. Third grade students can learn this, but they require considerable practice in doing it, and this can only be done in actual traffic. Rivara's program occupied more than 6 class periods, yet only the last 2 of those were outside, and only some undefined part of those were on real streets in real traffic.
The particular behaviors observed are also subject to criticism. I stated that the requirement is yielding to crossing traffic. The behaviors that were observed were stopping, starting to look, and continuing to look. These are parts of yielding, but they are not all necessary parts at any one time. Consider the observed stopping behavior; older children stopped less frequently than younger children. That may be ingrained bad habit, or it may be a more realistic evaluation of the need for stopping. When you can see that nothing is coming, it is not necessary to stop. Even these children are smart enough to realize that. What must be taught is to look in such a way that they can see if traffic is coming, and to stop if it is. That has to be taught on the street in real traffic, but in Rivara's program it was taught in classroom as crossing at a blind location. Adults can learn this concept even by reading it, as in Effective Cycling, but children require repeated practice.
Much the same applies to continuing to look. If the initial look determines that no traffic is coming, then it is not really necessary to continue to look. On a 2-lane street, which is the kind on which the observations were made, there is little chance for anyone to change their decision part-way across. Once started, you continue. If traffic is in sight from the right that will reach your location before you complete crossing, you should not start. If you cannot see sufficiently far along the roadway to cross before traffic that is now out of sight will reach you, then you should not cross at that location.
It is surprising that the program produced results as good as it did, considering the inadequacy of the training and the waste of time that that training involved.
Jacobsen has argued that Rivara's study demonstrates that cyclists should be advocating traffic calming instead of training because, so he argues, Rivara's study demonstrates that "very intensive, real-world, on the street training alters children's behavior only slightly". As discussed when considering Rivara's program purely as a pedestrian program, it demonstrates significant improvement even from a very poor training program.
Furthermore, the bulk of Rivara's subjects were too young to be considered capable of cycling in traffic without the supervision of a competent cyclist. The demonstration more than ten years ago that children as old as the older group of Rivara's subjects can demonstrate competent performance of the particular cycling skills that are required for 2-lane residential streets showed the falsity of Jacobsen's argument long before he made it.
Furthermore, Jacobsen's argument was not limited to children of these ages. As one read it, it applied to all child cyclists, which is an even greater stretch of propaganda.
In addition to the criticism of the Rivara, Booth, Bergman, Rogers and Weiss study as applying to its stated subjects, young child pedestrians, as stated above, there is no reason whatever to consider that it has any relevance to questions of cycling, whether by children or by adults.
Jacobsen argues that public policy regarding the use of the roads should be guided by the supposed inability of children to yield to crossing motor traffic. He and other people in the cycling forums have described a program, based on those of some European nations, of road designs that reduce motoring speed and of legal changes to place responsibility on the motorist in such dart-out accidents. "Should we require children to watch out for adults? In front of their own home? ... No."
The first point is that Jacobsen's argument applies only to a very restricted range of ages, from the age at which we would let children run about unsupervised to the age at which they can learn how to cross easy streets properly. Probably the age range is five through seven, three years.
The evidence on which Jacobsen bases his argument shows that even a poor training program that includes parental involvement produces significant improvement even with children of these ages. A good program ought to do better.
Furthermore, Jacobsen's argument applies only to child pedestrians, not to child cyclists. No person, child or adult, should be cycling without supervision unless they can exhibit the first three basic traffic-cycling behaviors: riding on the right side of the roadway, yielding to crossing traffic, and yielding when changing lateral position. As has been demonstrated, children of the third-grade level can learn these skills.
Therefore, a traffic-calming policy should be evaluated on the basis of the restricted portion of the community which it would benefit.
There are several methods of traffic calming. Some prevent through motor traffic, thus reducing the traffic on that street to that originating on that section of that street. Some make it dangerous to drive fast by reducing sight lines. Some make it uncomfortable or dangerous to drive fast by shaking up the vehicles. Small traffic circles used liberally reduce speed in their area. While slowing motor traffic, all of these have disadvantageous side effects that require considerable study. Jacobsen recommends speed humps.
In my opinion, traffic calming does not produce a net benefit for cyclists in most American urban areas, and in some areas with some methods it produces considerable disadvantages for cyclists. Some methods of traffic calming also produce considerable disadvantages for pedestrians by requiring longer walking distances. Whether traffic calming of residential areas would make it safe to leave young children unsupervised in street areas is something that needs to be determined when considering a policy of traffic-calming, but that is the substance of Jacobsen's argument.
The proposed changes in legal responsibility present considerable difficulty. The present law is that anyone entering the roadway has to yield to traffic already on the roadway. The law proposed by the traffic-calming advocates would require traffic on the roadway to yield to children (but possibly not to adults) who enter the roadway. The physical situation is that every parked car can hide a child. I haven't worked it out in detail, but I expect that the maximum speed at which it would be possible to stop before hitting a child emerging from behind a parked car is about two miles per hour. (That depends on the shape of the parked car and how far away the moving car can be driven from the parked car, which depends on the width of the moving car and the width of the roadway.) Some have written that this is the law in some continental European nations. I do not know this. However, it is generally recognized that their legal systems differ greatly from ours. Even if a motorist in, say, the Netherlands, is traveling at, say, 15 kph, hits a child who runs out in front of his car from a hidden location, what is the consequence to that motorist? If the proposed law were in force in the USA, the motorist would likely be financially responsible for lifetime medical care for the child, matter of some ten million dollars or so.
It would be difficult to make the law applicable only when a child was hit. Supposing that the victim were an adult. Then the law would consider the victim negligent for not obeying the law. However, if the motorist never saw the victim in time to prevent the collision, then the motorist could never claim that he saw that the victim was not a child in time to prevent the accident. The motorist was driving negligently because he could not stop if, indeed, a child had appeared, as is evident from the fact that he hit the victim, no matter what the age of the victim. Since the motorist was driving negligently fast, then he must have been at least partly responsible. Furthermore, the victim had the right to expect that the motorist would not be driving so fast, for that is, by definition, negligently fast. So the victim, adult or child, has the right to act as a child because the others using the road have the duty to act as if a child were always present.
I rather doubt that the American voter would let himself be made liable for injuries caused by hitting someone he had no chance of seeing.
The above argument applies to cyclists as well as to motorists, and in a worse way. How many cyclists have had to dodge away from a cyclist riding on the wrong side of the street who appears past a parked vehicle? The encounter would be like that. A cyclist who collided with a child under these conditions would not only injure the child and be held financially responsible for the child's injuries, but would be injured himself and would be financially responsible for his own injuries as well. Whether a voter is a motorist or a cyclist, such a law would be bad news.
1 firstname.lastname@example.org, email@example.com
2 Pediatrics Vol 88 No 4, Oct 1991
3 After-school observations enabled those children who had received the training to be positively identified when among other children.
4 Forester, John; Elementary-level Cyclist Training Program: Objectives, Techniques, and Results; 1981
5 That analysis became very useful. Its results now are the five basic traffic operating principles that are the basis for teaching Effective Cycling to both adults and children, and presumably would be equally useful when teaching motor driving.
6 When such an improvement in the sample is not statistically significant, it is reasonable to conclude that the study should have been designed with more subjects.
7 Substantially no children walked in the roadway before the program. Although this was observed, there was nothing the program could do to make a significant improvement in an activity that was already done very well.
Return to: John Forester's Home Page Up: Education