Davis has produced a mathematical function that gives the cycling suitability rating as the sum of a series of terms for various conditions: motor traffic volume in the outside lane, motor traffic speed, outside lane width, the condition of the pavement, and the condition of the location. In this system the higher the rating the worse the suitability for cycling. The maximum possible value and the approximate average value in this study for each term are given in Table 1.
|Lane Traffic Volume||4.0||1.4|
One can see that although each term is given equal weight in the formula, the
range of values of each term does not give equal weight for the characteristic
in the total score. If we look at the range from average to maximum as a
reasonable measure of the relative effectiveness of that characteristic in
influencing the score, we see that traffic volume has a range of 2.6, lane width
a range of 1.0, and motor speed a range of 0.4. These are the effective weights
given to each of these factors. Davis thought that he gave equal weight to each
factor, but this analysis shows that a different interpretation is more
The individual factors that contribute to pavement smoothness and to locational conditions also should be questioned, both for the rating of each and of the factors that are not included.
The factors for pavement smoothness are: cracking, patching, weathering, potholes, rough edge, curb and gutter, rough RR crosssing, and drainage grates. Certainly all of these things, if properly defined, detract from cycling suitability, but how many points off for how many defects in what length of road is a great question. Among many satisfactory drain grates one can find one that is very dangerous; do we count the average or the worst, or how does one rate the roadway? If the factor value for patching is 0.25 (as is given), does one patch in a mile rate as 0.25, or as 0.01? Does it make a difference if the patch is satisfactory or not? These are not explained. A pavement factor that is left out is concrete vs asphalt; concrete often has bumpy expansion joints and sometimes has dangerous ones parallel to travel.
The adverse factors for location conditions are: angle parking, parallel parking, right turn lanes, severe grades, moderate grades, frequent curves, short sight distance, numerous driveways, industrial land use, and commercial land use. Favorable factors (producing a negative contribution to the score) are: physical median, center turn lane, and paved shoulder. In contrast to the listing of right-turn-only lanes as a detriment to cycling, they are a benefit for cyclists, because they move the right-turning motorists to the cyclist's right before the motorists turn right. That avoids the motorist-right-turn car-bike collision, which is about number 2 or 3 in the frequency of motorist-caused car-bike collisions. There is no listing for the beneficial presence of traffic signals with protected left turn phases, which eliminate, where installed, the motorist-left-turn car-bike collision, which is the most frequent of motorist-caused car-bike collisions. As an additional benefit, they also make left turns easier for cyclists. As for listing paved shoulder here, if it is suitable for cycling (smoothness, cleanliness, etc), then it should be listed as part of outside lane width, while if it is not suitable for cycling it shouldn't be listed at all. are frequent curves detrimental to cycling? Some of the nicest roads I know have frequent curves. Furthermore, there is no consideration of the features that are very important because they speed or delay cyclists. That is, a street that is protected from cross traffic is much safer and faster than one that is not. I have partaken in a discussion with people who maintain, as probably most of the subjects in this study do, that residential streets are much safer than arterial streets. That may be true for slow cyclists, but it is not true for fast cyclists, who should slow down for every intersection if they are not to take dangerous and unlawful risks. Being protected by stop signs is generally ideal, but no street is so protected for all of its length. Practically all streets that are protected by stop signs eventually come to intersections with unfavorable stop signs or traffic signals, and generally those traffic signals have an unfavorable green proportion because the traffic on the cross street is more numerous or faster. Therefore, the system should consider the disbenefit of intersections in general, with the value of being protected by stop signs and the disbenefits of facing unfavorable stop signs, and have a system for evaluating the effect of traffic signals.
Davis then surveyed a set of road segments that formed an easily cycled route, and scored them according to his system. He then had a sample of cyclists ride this route and report their scores for eleven different factors concerned with roadway and traffic. (See List 1) The scores were all rating on a 1 to 10 scale, not in any way a reflection of actual values. He then analyzed the cyclists' scores for each factor and compared them to the ratings he produced by using his system.
List 1: Rated Characteristics
1: Pavement surface
2: Lane/intersection configuration
3: Outside lane width
4: Curves, sight distance, visibility
5: Hills, terrain
6: Adjacent land use
7: Traffic volume
8: Motor vehicle speed
9: Parked vehicles
10: Driveway traffic
11: Traffic signals, signs
The first thing that Davis recognized, probably in advance, was that the subjects could not rate length of trip and time for trip, as they were not going to any particular destination and had no particular schedule to meet. He covered this by asking his subjects to rank the relative importance of the four factors: Length of route, travel time, roadway conditions, and traffic conditions. Starting with the least important, their ratings were travel time, length of route, roadway conditions, and traffic conditions. Therefore, concluded Davis, such a rating system, if validated by the responses of subjects on test rides, would reflect two factors most important to cyclists, traffic conditions and roadway conditions.
I think that this is dubious. Commuting cyclists whom I know tend to evaluate routes according to two criteria: the one that takes the least time, for when time is important, and the one that provides the most physical conditioning within the available time, for when that is desired. I think that rating routes for cycling suitability must consider at least their transportational utility. That may be combined with a rating of the road's conditions according to some system such as Davis's, but it should not be ignored.
Davis provides comparisons of three ratings: the calculated rating, the overall general rating given by the cyclists at first, and the average of the ratings given by the cyclists for each category. The calculated ratings varied most, the general ratings varied less, and the averaged ratings varied very little. One would expect the averaged ratings to vary less than any individual ratings, as actually happened. However, the cyclist's general ratings were not a component of the average. This shows that the cyclists did not accurately average the ratings that they later wrote down and went more on general impressions. The general ratings and the averaged ratings varied consistently together, while the calculated ratings in 3 out of the 8 instances varied in the opposite direction from the other two. This shows that the cyclists used a different system than the formula originally hypothesized.
Davis did not provide a comparison between the actual values and the cyclists' evaluations. Indeed, the two systems were incompatible. However, it is possible to compare the cyclists' ratings against the three numerically defined characteristics: traffic volume, traffic speed, and lane width. I compared the average of the cyclists' scores against the given conditions; the correlation coefficients, assuming linear relationships, are given in Table 2.
These show that the cyclists' satisfactions were strongly related to traffic
volume and traffic speed, but that cyclists had no consistent way of judging
whether the roadway width was or was not adequate. Each cyclist may have known
how wide the outside lane actually was, but if so the cyclists disagreed about
how satisfactory it was. Consider the facts. The narrowest road, with 8 foot
lanes, carried only 750 cars per lane per day at 20 mph, while the next
narrowest lane, 10 feet, carried, in one example, 5,000 cars per lane per day at
40 mph. With variations like that in the actual conditions, it is no wonder that
the cyclists could make no consistent relationship between satisfaction and lane
width. Davis recognized the deficiency, but put no numbers to it.
This study tells us that average cyclists pay fairly accurate attention to the volume and speed of same direction motor traffic in assessing their estimate of the suitability of cycling routes. It implies, but does not prove, that they pay much less attention to anything else. However, more can be shown than Davis did. If the average cyclist scores of satisfaction with motor-vehicle speeds and volumes are plotted on the same plot as Davis's Figure 3, Rating Index Comparison, the following information appears. In five out of the eight routes, the level of satisfaction with speed is identical to that with volume, and in the remaining three instances the values are only one unit apart. Furthermore, if the average of the two is plotted its swings are always in the same direction as the average rating and the general rating, but are more extreme. In other words, all the other factors, in total, serve to dilute the effect of motor-vehicle traffic on the opinions of average cyclists.
However, while the opinions of average cyclists express reality in the sense of being a fairly accurate measure of same-direction motor traffic, there is no indication at all that they are attached to reality in any substantive way. Cyclists pay great attention to same-direction motor traffic, but what is the actual effect of same-direction motor traffic upon cyclists? The effect is very small. The most significant effect is that it makes them allow more time when preparing to turn left, because adequate gaps are less frequent as the speed and volume of overtaking traffic increase. There is also an increase in the frequency with which cyclists encounter right-turning motorists., which is presumably proportional to the number of motorists who overtake the cyclist. However, I venture to suggest that these considerations play no part in the opinion of average cyclists, and they certainly were not taken into consideration in the study.
What then were the cyclists considering? Davis did not ask. I venture that the cyclists considered the high volume roads dangerous, and the only danger that they were considering was the probability of being hit from behind. That probably is proportional to the number of overtaking motorists, but it is only a minute cause of casualties to cyclists under urban, daylight conditions.
In short, there is no real reason to consider that the speed and volume of same-direction motor traffic is a real measure of the cycling suitability of most routes. However, there is great reason to consider the effect of this cyclist-inferiority superstition on the type of planning and of programs that we conduct for cycling transportation, and to consider how to produce a rational program in the light of this present false superstition. So far I see every indication that government is catering to the superstition and doing nothing at all about producing a rational program.