Top: John S. Allen's Home Page
Up: Table of Contents
Previous: John S. Allen's Analysis
Next: Response by Dr. Gregory Rodgers

Calculation of miles/hour figure from the CPSC study

John S. Allen

Two sets of data  are used in Bicycle Use and Hazard Patterns to sample the U. S. population of bicyclists. Data collected by research firm Abt Associates for the CPSC were used to calculate the size of the bicyclist population and yearly mileage. Data collected by the Harris polling organization for Rodale Press, publishers of Bicycling magazine, were used to calculate the size of the bicycling population, monthly mileage, and the number of months ridden per year.

The Rodale data is based only on the 80 percent of bicyclists who had bought their bicycles new, but the age distributions of the two populations are very similar and the two data sets are comparable.

I have undertaken a comparison of the mileage figure from the Rodale data and the hours of use figure from the CPSC data, as follows:

Rodale divided the bicyclists into age groups and "cluster groups." These reflected four levels of bicycling activity, "Infrequent," "Casual," "Moving up" and "Enthusiast." The fraction of bicyclists in each age group and cluster group was reported as follows:

Table 1
Fraction of cyclists by age group and cluster group

Cluster by age group (Tbl. 2, p. 117) All
Enth. Mov. up Casual Infreq.
18 to 30 0.388 0.389 0.386 0.327 0.346
31 to 40 0.361 0.363 0.351 0.338 0.343
41 to 50 0.165 0.158 0.153 0.174 0.169
51 up 0.086 0.090 0.110 0.160 0.143

All ages

0.027 0.073 0.208 0.692 1.000

The monthly mileage for the cluster groups was reported as follows:

Table 2
Monthly mileage by cluster

Cluster Enth. Mov. up Casual Infreq.
Miles/mo (tbl. 17, p. 121) 205.4 80.9 29.8 24.1

The CPSC report includes a graph from Rodale data indicating in which months bicyclists in each cluster group rode. From this and the monthly mileage above, it is possible to determine a yearly mileage for each cluster group, as follows:

Table 3
Months in which riding occurred and miles per year, by cluster


Percentage in each cluster riding in month
(Figure B, p. 133)

  Enth. Mov. up Casual Infreq.
Jan 41 26 16 14
Feb 43 26 17 15
Mar 60 43 28 24
Apr 83 72 60 53
May 94 92 84 78
June 95 94 91 86
July 93 92 83 80
Aug 93 92 83 78
Sept 92 88 78 71
Oct 79 72 58 49
Nov 54 42 27 23
Dec 42 28 17 14
Total months 7.24 6.39 5.35 4.88
Miles per year 1784.93 620.50 191.32 140.99

Recalculating from Table 1 above gives the fraction of bicyclists in each age and cluster group:

Table 4
Fraction of bicyclists in each age and cluster

fraction of cyclists in age and cluster group All clusters
Enth. Mov. up Casual Infreq.
18 to 30 0.010 0.028 0.080 0.226 0.345
31 to 40 0.010 0.026 0.073 0.234 0.343
41 to 50 0.004 0.012 0.032 0.120 0.168
51+ 0.002 0.007 0.023 0.111 0.142
All ages 0.027 0.073 0.208 0.691 0.999

The tables of Rodale data in the CPSC report give monthly mileage figures for the different cluster groups, but not for the different age groups. Multiplying the fraction of bicyclists in each age and cluster group by the yearly mileage for each cluster group assumes that the average mileage within each cluster group is the same regardless of age. Calculating the contribution of each cluster group in this way and comparing this with the CPSC figure for hours of riding per year produces the results in the rightmost column below:

Table 5
Comparison of Rodale mileage and CPSC hours of riding

Age Group Miles per year times fraction of bicyclist population Totals for age group Totals/ frac.  in age group = mi/yr for age group CPSC Hrs/yr. (tbl. 5, p. 45)  

Mi./hr. by age group

Enth. Mov. up Casual Infreq.
18-30 18.699 17.620 15.360 31.903 83.582 241.956 224.160 1.079
31-40 17.398 16.443 13.968 32.976 80.784 235.418 164.400 1.432
41-50 7.952 7.157 6.088 16.976 38.173 226.921 177.000 1.282
51+ 4.145 4.077 4.377 15.610 28.208 197.965 103.400 1.915
All ages, weighted by size of age group 230.748 182.602 1.264

The average speed of American bicyclists according to the CPSC study is, then, 1.264 mile per hour. The assumption that members of a cluster group have the same mileage regardless of age introduces an unknown difference between the reported mileages and the mileages calculated here for each age group. Cluster group membership depended on other factors besides reported mileage, and even if it did not, the distribution of responses within each mileage range would not have to be the same in each age group. However, the errors cancel out in the overall mileage figure for all age groups. Also note that the Rodale data shows declining mileage with age. The increase in the miles/hour figure with age reflects a greater decline in the hours of use in the CPSC data.

My calculations here account for the Rodale data on months in which riding occurred. Both Rodgers and I overlooked this data in our earlier calculations. We assumed 8 months of riding, and the overall average which resulted was 1.55 miles per hour. Accounting for the data on the number of months reduces the yearly mileage substantially.

The Rodale question as to the months in which riding occurred, and the one about how far a respondent rode in a "warm weather month" do not exactly coincide with each other. There are likely some months in which some riding occurred, but not as much as in a "warm weather month." The yearly mileage figures which I have derived from the Rodale data may be somewhat high for this reason.

It is not possible to tell from my comparison whether the CPSC data, the Rodale data, or both are inaccurate. However, as I discussed in my analysis of the CPSC report, the hourly use figure from the CPSC data is clearly wildly inaccurate, based on the absurdly low accident rate it generates and on comparisons with hours of use data for motor vehicles.

The Rodale yearly mileage for "enthusiasts" -- 1785 miles -- agrees reasonably well with that of the 1976 Kaplan study of League of American Wheelmen members -- 2322 miles. This comparison suggests that the Rodale survey's exposure data is reasonably accurate, at least for avid bicyclists. The Kaplan Study indicates more months of riding per year, consistent with the greater mileage and suggesting that one reason for the lower mileage in the Rodale study's "enthusiast" category is that it reached somewhat deeper into the bicycling population.

The Microsoft Excel '97 spreadsheet in which I carried out my calculations of the miles/hour figures is posted for downloading and review. If you do not own Excel, you may still view the spreadsheet by downloading the free Excel viewer from the Microsoft Web site.


Top: John S. Allen's Home Page
Up: Table of Contents
Previous: John S. Allen's Analysis
Next: Response by Dr. Gregory Rodgers

Contents 2001, John S. Allen
Last modified October 21, 2001