Listening to Phoenix’s Bicycle Collision Summary

[Updated Jan 2015: There are more recent summaries posted at; direct links (unfortunately, many of these links now are 404 or link to the wrong report): bike (2010) ; ped(2012) ; all traffic(2012); bike (2013) (see section below for 2013 numbers) also an interesting MV collision rate study spanning 2006-2010.  I did a quick glance at the bike and numbers are very similar to the 2007 which is detailed below… including the tendency of Phoenix PD to mis-characterize collisions at driveways and crosswalks as the fault of the bicyclist ]
[Update 2023: I see a bicycle summary for year 2020 , see below, which was released in 2022; ]

Phoenix, and many other entities issue a report, usually called something like a Bicyclist Collision Summary. For some background, complaints, and links to others, see Understanding Collision Summaries.

At hand, I have the most recent, 2007, report from the City of Phoenix, which can be still be found hereLocal copy in case that link breaks)

To put some gross figures into context, in 2007 in City of Phoenix (excludes freeway; which are normally investigated by DPS) there were around 33,000 collisions of all types; and of that number there were 440 bicycle-MV, and 624 ped-MV collisions. (source: 2009 Phoenix Traffic Collision Summary (link dead as of early 2012), which tracks 5 years of data for comparison purposes).

[ by way of a brief update: I haven’t looked closely at it but now there is 2010 version for overall traffic, but no bicycle-specific update; the trend over the past several years since 2006 has been a rather sharp reduction in the overall number of crashes; while the number of bicycle-MV has remain quite steady around 450. The overall crash number has declined steadily from a high of 35,200 in 2006 to 21,900 in 2010. Recession? A more recent update shows remarkably consistent number of Phoenix bicyclist injuries 2009-2016 in the mid-400s  ]

The data, at first glance, appears to have been simply extracted from ADOT’s ALISS (later called the SDM, Safety Data Mart) database, But later conversations with Joe Perez, CoP Bike/Ped coord, indicate he eyeballs and tallies each Bike/ped crash report to get the Bike/ped summary reports as published. Note that these figures do NOT include freeways (see “Additional Information”); thus there would be appreciably more MV collisions, but virtually no more cyclists, and only a handful of additional peds actually occurring within Phoenix city limits. (that’s not completely true because sometimes DPS investigates collisions at interchanges; still it’s a small percentage of bike or ped crashes)

What does the Bicycle Collision Summary Tell us?

Other than my standard complaints, e.g. there is no exposure data, and nothing is split by seriosness, the most compelling thing I’ve see is the data within graphical representations of where the bicyclist was located when the crash occurred (pages 10 and 11). I’ve recreated the graphical data and organized it into a table, below.

VC (Vehicular cycling, also called EC. Effective Cycling) tells us that “Cyclists fare best when they act and are treated as drivers of vehicles”. To expand just a little, vehicular cyclists ride in visible, predictable manner, following the rules of the road; these methods have (long) been shown to minimize crash-risk to cyclists.

If there is more compelling data showing the vehicular cyclists are likely to avoid a MV collision, I’m not aware of it.

Based solely on the position in the road, I’ve categorized whether or not the collision happened while a cyclist was potentially cycling in a VC manner. A stunning 90% of the collisions involved non-VC position. The weasel words actually imply that the 90% figure is even higher — because position alone cannot tell us if the cylcist was following the rules of the road; e.g. a cyclist in the street, riding with traffic and running a red light, or stop sign is in the correct position, but most certainly not following VC principles.

90% is so high, it makes me wonder if there’s some flaw in the data or methodology.?

A word about sidewalk cycling. Riding on the sidewalk, and subsequently through crosswalks and across driveways is certainly legal in Phoenix (though this topic could apparently fill volumes, be sure to check here; sidewalk cycling itself varies by locality) but is also certainly leads to many collisions at driveway and crosswalks, particularly when the cyclist is riding counter-flow to adjacent traffic. Cyclists involved in this sort of collision are frequently wrongly cited for inapplicable violations; and are frequently wrongly “faulted” (listed as Unit #1);  this is a result of poor police training — though without looking closely at any particular crash reports it’s hard to tell exactly what’s going on. (we’ve looked at a large number of 2009 reports)


Raw Data Subjective categorization
Cyclist position when collision occurred Num of col lisions VC position? # of GOOD Side walk?
Near Inter section In street(Bike lane), with traffic 4 GOOD 4
In street(Bike lane), against traffic 5 BAD
In Street, with traffic 5 GOOD 5
In Street, against traffic 18 BAD
In Street(within intersection) with traffic 18 GOOD 18
In Street(within intersection) against traffic 15 BAD
Driveway (on sidewalk), with traffic 8 BAD 8
Driveway (on sidewalk), against traffic 27 BAD 27
Crosswalk, unmarked, with traffic * 15 BAD 15
Crosswalk, unmarked, against traffic * 37 BAD 37
Crosswalk, marked, with traffic 52 BAD 52
Crosswalk, marked, against traffic 82 BAD 82
Crossing mid-block 20 BAD
Not near inters ection Crossing mid-block 26 BAD
In street(Bike lane), with traffic 4 GOOD 4
In street(Bike lane), against traffic 0 BAD
In Street, with traffic 11 GOOD 11
In Street, against traffic 6 BAD
Driveway (on sidewalk), with traffic 10 BAD 10
Driveway (on sidewalk), against traffic 77 BAD 77
440 42 308
10% 70%

* In the report, this figure was listed as 52 total; I got the breakdown of 37 against from Joe Perez, city bike/ped coordinator.

Against Traffic and Fault Assignment

60% of all reported Bike-MV in Phoenix were bicyclist against the flow of traffic (5 + 18 + 15 + 27+37 + 82 + 6 + 77 = 267 of 440 total), the large majority of these on the sidewalk (i.e. at cross walks and driveways), and not in the road. Conversely, 29%( 4 + 5 + 18 + 8 + 15 + 52 + 4 + 11 + 10 = 127) were traveling with the flow of traffic. Note there’s a third category,  10% (20+26) were neither with nor against traffic; as they were “crossing mid-block”; possibly, or perhaps typically in a pedestrian-style maneuver.

By the way, only 10% of the total were in street against traffic (5 + 18 + 15 + 6 = 44 / 440). All figures from the table above.

In street against traffic cases should almost certainly be faulted to the cyclist.

However, the number crosswalk+driveway collisions where the cyclist was going the “wrong way” is HUGE; 223 (51%). “Wrong Way” is in quotes because there is no such thing legally as going the wrong way on a sidewalk or along a crosswalk or driveway. These are the most likely to be reported as cyclist at fault in the OTHER category — this is wrong, yet police routinely make this mistake.

This potentially skews the ostensibly (fault assigned by police) high ratio of cyclist at fault vs. motorist at fault (over 2 to 1), as presented on page 7 of the 2007 report as pie charts:

There were 84 cases where the cyclist was faulted for “OTHER”  (52% x 36.8% x 440); this is in addition to 30 “WRONG WAY” cyclist-faulted cases (52% x 13% x 440). Most if not practically all of the 84 should be assigned fault to the motorist and not the cyclist; the motorist turned across the path of cyclist where the motorist failed to yield.

For more about Most At Fault (MaF) see arizona-agency-ncic-numbers and scroll down a little.

Other Data

It would seem to be sensible and relatively inexpensive to extract consistent summaries for any and every locality — though for whatever (historical?) reasons those that I have see appear to present the data differently, and even present different data. Again, check out Understanding Collision Summaries for links to ADOT, Mesa, and Phoenix summaries.

Particularly interesting comparisons would be Tempe, where counter-flow sidewalk cycling is illegal, and Tucson, where generally speaking all sidewalk cycling is illegal. Of course, to make valid comparisons you would have to wonder or know how enforcement (or lack thereof) affects the number and direction of sidewalk cyclists.

2013 Data #2013

The 2013 Bicyclist collision summary (that link now links to the 2016 summary, 2013 is apparently lost), unfortunately, did not break out cyclist direction. There were 485 total bicyclist collisions in 2013. It also makes a wrong/misleading statement in the Facts-at-a-glance, on page 4: “29% of all collisions occurred in an intersection crosswalk”. That should have either said “in a marked crosswalk”, or more accurately, the correct answer appears to be 42% — unmarked crosswalks are crosswalks nonetheless.

The breakdown of where the bicyclist was is on p. 20, here are the three categories of what was referred to above as “sidewalk?” (i.e. the bicyclist was cycling on the sidewalk just before the collision):

  • 141 Crosswalk (implies marked crosswalk) [29%]
  • 62 Unmarked Crosswalks [13%]
  • 113 Driveway [23%]

For a grand total of 65% of collisions in 2013 were sidewalk-related; so a tad down from 2007 when it was 70%. These collisions included 3 bicyclist fatalities. I didn’t include the one incident that occurred at a mid-block marked crosswalk.

2020 Data #2020

Note: 2020 was “pandemic” year. 2020 Bicyclist Collision report.

Unfortunately, the 2020 report continues the problem noted on the 2013 and 2016 reports, which is that the “Collisions by Location” no longer break out cyclist’s direction, i.e. with vs. against traffic. This is valuable information to help people understand the basics of traffic safety.

It also seems to be missing a fatality. Crash database records show 5 but they only say 4. (all, by the way, “midblock crossing” location); i haven’t tried to figure that out, but all 5 are listed in the bicyclist fatality grid for 2020.

The overall trend towards far fewer reported crashes bears looking into; I would suspect it’s less reporting (vs. actually fewer crashes), but that’s just a hunch. It appeared that around 2014 the trend in many cities around the state was a dramatic decline; but not Phoenix at that time. I haven’t revisited those numbers in a few years, so maybe it’s just that Phoenix is lagging at the under-reporting. As noted here, overall fatalities were VERY high in 2021, even as the overall number of reported crashes is histrically very low.

Post Script

The title of this article is a play on the title of a 1996 paper written by Jeffrey A. Hiles,  Listening to Bike Lanes, which is well worth reading. (the old html version is at; and there’s an image of the original .pdf at

4 thoughts on “Listening to Phoenix’s Bicycle Collision Summary”

  1. As an aside — the numbers in the overall collision summary do NOT include collisions that occur on freeways, or even possibly ones occuring at freeway interchanges. These are investigated by DPS (ncic 799) vs. Phoenix PD (ncic 723). So doing a query like this:

    select eInjurySeverity,OfficerNcic,count(*) from 2012_incident where CityID=214 group by 1, 2;

    showed 116 fatal collisions via PPD, vs. 28 by DPS. The Phoenix 2012 Collision summary lists only 105 fatal collisions (and 109 people killed).

    FATAL 700 1
    FATAL 713 1
    FATAL 723 116
    FATAL 799 28

    This may or may not be an issue in the bicyclist or ped summary; I’m not really sure exactly how they, CoP, generates their data… they do say “Collisions that occur on the freeway system or private property are not included in this report. SDM data may not match statistics reported by other sources”. SDM is (also) the data I have expressed above, so it’s not clear how so many fatals are “missing”, i.e. between 11 and 13. There are a few peds killed on freeways, and usually zero bicyclists; there are, however, typically several bike and several ped fatals per year at interchanges; so not on the freeway per se but will have been investigated by the DPS.

Leave a Reply

Your email address will not be published. Required fields are marked *