Critique of the road professionals obsession with exposure in the Dangerous by Design Report

by B' Spokes
Wednesday, December 03 2014 @ 06:15 AM GMT

B' Spokes: (Today's tangent in trying to call attention to Maryland's high pedestrian fatality rates from national organizations) Normalizing data is an interesting concept, how do you compare different groups with different characteristics? Especially when talking about road safety. It seems the road folks are obsessed with "exposure" and not the toll traffic fatalities has on the general population. Which to me would be like Baltimore saying since we have a lot of guns on the street (exposure) our gun violence is not that high considering the exposure. Or maybe a better analogy would be to say that since we have more people on the street so someone has a better chance of catching a stray bullet therefore our gun violence is not that bad.

Whether or not we have a gun violence problem is based on deaths per population and I'll assert traffic deaths should do the same and not per Vehicle Miles Traveled nor how many are out walking. More or less traffic is its own kind of problem, mixing that in with fatality rates just obnubilates the underlying problems and thus the solutions. People everywhere do the same things, basically go to work and shop. Traveling farther to do the same things does not improve safety nor the quality of life. So why do we accept this as a fair way to normalize fatality rates for comparisons?

Don't get me wrong, places that have a high pedestrian fatality rate with a low mode share are bad and should be highlighted but on the other extreme, are places with a high pedestrian fatality rate and a high mode share good? It comes down to what are we trying to point out, fatality rates or mode share? As I said, both have their issues and both have their solution sets (with some overlap) but why mix the two up?

A new report from the International Transport Forum finds that the United States had more road deaths per capita in 2012 than Canada, Australia, Japan, and all of the European nations that reported data.

Specifically, the US had 10.7 road deaths per 100,000 people. Canada and France both had 5.8. And the United Kingdom was down at 2.8. (The report explains that the per-person death rate is helpful for comparing deaths from various causes.)

http://www.vox.com/xpress/2014/8/25/6064173/road-fatalities-world-map-driving-safety

But then it goes on to say that when the comparison is done by Vehicle Miles Traveled the US looks a lot better. Seriously? The fact that we drive a lot more than the rest of western civilization makes us better?

This obsession with exposure even got into Smart Growth America's report Dangerous By Design. Where their Pedestrian Danger Index (PDI) is modified by pedestrians walking to work mode share. Like a 2% mode share means that pedestrians can be killed at twice the rate to rank the same with a place with a 1% mode share, that is wrong! That's that's taking a 2% change and making it a 200% change. How about normalizing on those who drive to work? More cars (less people walking), more dangerous right? (This would avoid the wild fluctuations where walking is is up to 5 times that of Florida so fatality rates are ranked 5 times better than what they are IMHO.) So the question is, should "exposure" be based on cars that kill or people who walk? (the latter sounds too much like victim blaming to me. Are they really trying to say, "The more people who are out walking naturally the more that are going to get killed." This is the exact opposite of the safety in numbers concept, granted the jury is still out if this is a proven concept but still we cannot assert the opposite across different population characteristics.) Besides Dangerous by Design's methodology makes the New York metro area's high pedestrian fatality rate one of the safest metro areas to walk, this does not feel right to me. The way I would tentatively do the Pedestrian Danger Index by the change in the population that drives, New York's ranking would improve a few notches over a pure pedestrian fatality rate but it still would be high on the list. And for the converse, the Nashville, TN metro area their ranking would be worse by a few notches because so many drive to "justify" their pedestrian fatality rate.

The Dangerous by Design Report takes normalized fatality rates and normalizes them again. So we are normalizing normalized traffic fatalities, something about that just screams of trying to make something bad sound not that bad,

Back to New York Metro area, sure a lot of people in Manhattan walk but think about the Bronx and New York's Vision Zero. I really don't think New York metro deserves a ranking of 48 (with 51 being the least dangerous metro area for pedestrians.)

The biggest problem with using the primary mode of transportation to work it fails to capture the size of the population that is out there waking. Take kids for example which are not in the mode share numbers, to ball park the error, kids make up ~14% of the population. So adding that to those adults that walk would change the range from 1 - 5 (% of adults that walk) to 15 - 20% (of the population that walks), an increase of a third not the 500% that they are using in their math. And that's just one segment of the population that they fail to capture.

Comparisons of the walking share to work is fine, as it is an indicator of how walkable one place is compared to another but using it to determine the size of that population and its "exposure, well that's just wrong, Seriously deaths per population per another population number is supposed to be a meaningful number?

My next point is I looked up the time of day pedestrians were killed here in Maryland and topping the list is what I would call bar closing times, next was lunch time. Neither has anything to do with how people get to work so why are we normalizing on that? In fact Pedestrian fatalities during normal commute times were near the bottom of the list. "the per-person death rate is helpful for comparing deaths from various causes" Life is life everywhere and the rate in which pedestrians die is indicative how safe the streets really are for pedestrians and making bad places look better based on an unproven concept of "exposure" is wrong.

I was biking through Towson during lunch time and there where hoards of people out walking. I am willing to bet over 90% of those drove to work. That is to say how many walking around work centers is not always determined by peoples principle mode on how they got to work in the first place.

My rework of their tables based on pedestrian death rates follows. IMHO excluding this information is wrong. If they want to add tables based on other normalized data fine but I think their math is way off in their current thinking. (Side note: I can understand fatalities per vehicle miles traveled to justify freeways as they eliminated a known danger, intersections. So apply vehicles miles traveled in this instance proves (or disproves) the safety advantages of freeways. But outside of this context diluting fatality rates for a given population with vehicle miles traveled rewards sprawl and penalizes compact development. Exposure (vehicle miles traveled to name one) should not be the universally accepted way to compare diverse populations unless it is part of what we want to test or show. IMHO What the Dangerous by Design Report does is prove that pedestrian "exposure" by those who walk is not a valid way to compare diverse populations )

Table 1

Ranking by Pedestrian Fatality rate	Their ranking by their crazy PDI	Large Metro Areas	Total pedestrian deaths (2003– 2012)	Annual pedestrian deaths per 100,000 (2008– 2012)	Percent of people commuting by foot (2008–2012)	Their Pedestrian Danger Index (2008– 2012)	Percent of people commuting by motorized (2008–2012)	My revised Danger Index
1	2	Tampa-St. Petersburg-Clearwater, FL	874	2.97	1.6	190.13	98.4	292
2	1	Orlando-Kissimmee, FL	583	2.75	1.1	244.28	98.9	272
3	4	Miami-Fort Lauderdale-Pompano Beach, FL	1,539	2.58	1.8	145.33	98.2	253
4	3	Jacksonville, FL	359	2.48	1.4	182.71	98.6	245
5	22	New Orleans-Metairie-Kenner, LA	272	2.09	2.5	84.9	97.5	204
6	9	Phoenix-Mesa-Scottsdale, AZ	840	1.86	1.6	118.64	98.4	183
7	18	San Antonio, TX	373	1.86	1.9	96.87	98.1	182
8	13	Las Vegas-Paradise, NV	413	1.85	1.8	102.67	98.2	182
9	14	Riverside-San Bernardino-Ontario, CA	889	1.81	1.8	102.17	98.2	178
10	27	Los Angeles-Long Beach-Santa Ana, CA	2,435	1.79	2.7	66.91	97.3	174
11	29	San Diego-Carlsbad-San Marcos, CA	576	1.79	2.7	66.02	97.3	174
12	28	Baltimore-Towson, MD	482	1.78	2.7	66.42	97.3	173
13	48	New York-Northern New Jersey-Long Island, NY-NJ-PA	3,384	1.76	6.2	28.43	93.8	165
14	5	Memphis, TN-MS-AR	239	1.72	1.3	131.26	98.7	170
15	7	Houston-Sugar Land-Baytown, TX	1,034	1.7	1.4	119.64	98.6	168
16	23	Sacramento-Arden-Arcade-Roseville, CA	390	1.66	2	81.27	98	163
17	10	Charlotte-Gastonia-Concord, NC-SC	254	1.65	1.5	111.74	98.5	163
18	34	Philadelphia-Camden-Wilmington, PA-NJ-DE-MD	959	1.64	3.7	44.27	96.3	158
19	17	Louisville-Jefferson County, KYIN	200	1.6	1.6	98.48	98.4	157
20	8	Atlanta-Sandy Springs-Marietta, GA	839	1.59	1.3	119.35	98.7	157
21	11	Detroit-Warren-Livonia, MI	713	1.55	1.4	111.63	98.6	153
22	24	Austin-Round Rock, TX	251	1.44	1.8	78.58	98.2	141
23	20	Oklahoma City, OK	177	1.43	1.6	87.16	98.4	141
24	35	Washington-Arlington-Alexandria, DC-VA-MD-WV	843	1.41	3.2	44.06	96.8	136
25	16	Raleigh-Cary, NC*	165	1.37	1.4	100.35	98.6	135
26	47	San Francisco-Oakland-Fremont,CA	633	1.36	4.3	31.44	95.7	130
27	30	San Jose-Sunnyvale-Santa Clara, CA	260	1.35	2.1	65.58	97.9	132
28	6	Birmingham-Hoover, AL*	148	1.33	1.1	125.6	98.9	132
29	19	Richmond, VA	167	1.32	1.4	94.98	98.6	130
30	12	Dallas-Fort Worth-Arlington, TX	900	1.31	1.2	107.54	98.8	129
31	37	Buffalo-Niagara Falls, NY	147	1.29	3	43.06	97	125
32	33	Salt Lake City, UT	132	1.26	2.3	55.28	97.7	123
33	39	Providence-New Bedford-Fall River, RI-MA	198	1.26	3.2	39.94	96.8	122
34	15	Nashville-Davidson-Murfreesboro-Franklin, TN	210	1.25	1.2	100.79	98.8	124
35	31	Denver-Aurora-Broomfield, CO	349	1.24	2.1	58.13	97.9	121
36	26	St. Louis, MO-IL	364	1.22	1.7	69.69	98.3	120
37	32	Columbus, OH	187	1.2	2.1	56.29	97.9	117
38	43	Rochester, NY	121	1.2	3.5	33.97	96.5	116
39	25	Indianapolis-Carmel, IN	199	1.16	1.6	72.98	98.4	114
40	21	Kansas City, MO-KS	228	1.13	1.3	85.74	98.7	112
41	36	Virginia Beach-Norfolk-Newport News, VA-NC	186	1.13	2.6	43.6	97.4	110
42	45	Portland-Vancouver-Beaverton,OR-WA	250	1.12	3.5	32.19	96.5	108
43	38	Hartford-West Hartford-East Hartford, CT	121	1.11	2.7	41.58	97.3	108
44	41	Milwaukee-Waukesha-West Allis, WI	183	1.07	2.8	38.79	97.2	104
45	44	Chicago-Naperville-Joliet, IL-INWI	1,165	1.03	3.1	32.94	96.9	100
46	51	Boston-Cambridge-Quincy, MANH	476	0.99	5.3	18.65	94.7	94
47	49	Seattle-Tacoma-Bellevue, WA	375	0.96	3.6	26.81	96.4	93
48	50	Pittsburgh, PA	234	0.9	3.6	25.1	96.4	87
49	40	Cincinnati-Middletown, OH-KYIN	187	0.84	2.1	39.54	97.9	82
50	42	Cleveland-Elyria-Mentor, OH	142	0.73	2.1	34.37	97.9	71
51	46	Minneapolis-St. Paul-Bloomington, MN-WI	249	0.72	2.2	32.15	97.8	70

Table 5 (Not enough information given to correct their Danger Index);

Ranking by Pedestrian Fatality rate	Ranking by the percentage of traffic fatalities that are pedestrian	Their ranking by their crazy PDI	State	Total traffic fatalities (2003– 2012)	Total pedestrian fatalities (2003– 2012)	Percentage of traffic deaths that were pedestrians (2003–2012)	Annual pedestrian deaths per 100,000 (2003– 2012)	State Pedestrian Danger Index
1	8	1	Florida	29,302	5,189	17.7	2.83	168.6
2	18	12	New Mexico	4,131	504	12.2	2.53	88.5
3	12	8	Arizona	9,960	1,434	14.4	2.34	101.2
4	20	3	Louisiana	8,673	1,030	11.9	2.29	116.6
5	24	4	South Carolina	9,546	1,020	10.7	2.29	110.4
6	1	49	District of Columbia	368	133	36.1	2.26	14.5
7	10	6	Delaware	1,223	194	15.9	2.22	103.6
8	9	13	Nevada	3,322	540	16.3	2.1	85.3
9	4	28	Hawaii	1,269	262	20.6	1.98	35
10	6	15	Maryland	5,799	1,067	18.4	1.88	78.6
11	5	17	California	35,829	6,798	19	1.86	62
12	22	9	North Carolina	14,486	1,683	11	1.84	99.8
13	38	7	Mississippi	7,833	527	6.7	1.8	102.6
14	17	10	Texas	34,107	4,192	12.3	1.74	97.5
15	3	21	New Jersey	6,644	1,501	22.6	1.72	53
16	25	5	Georgia	14,748	1,564	10.6	1.67	104
17	2	39	New York	13,144	3,097	23.6	1.61	24.5
18	35	2	Alabama	10,061	723	7.2	1.55	125.2
19	39	14	Arkansas	6,181	403	6.5	1.41	80
20	37	16	Oklahoma	7,338	513	7	1.4	73.3
21	13	19	Michigan	10,364	1,373	13.2	1.38	59.4
22	21	30	Oregon	4,165	497	11.9	1.33	33
24	33	18	Missouri	9,978	762	7.6	1.29	59.6
23	36	11	Tennessee	11,309	799	7.1	1.29	88.6
26	19	50	Alaska	725	87	12	1.26	13.9
25	40	20	Kentucky	8,496	539	6.3	1.26	58.3
27	23	33	Pennsylvania	14,341	1,555	10.8	1.24	30
28	48	40	Montana	2,334	116	5	1.2	24.2
29	43	26	West Virginia	3,747	219	5.8	1.19	37.1
31	14	31	Illinois	11,429	1,488	13	1.17	32.3
30	26	29	Colorado	5,386	565	10.5	1.17	34.1
32	11	32	Rhode Island	769	121	15.7	1.14	31.1
33	7	43	Massachusetts	4,015	716	17.8	1.1	21.9
34	28	22	Virginia	8,663	841	9.7	1.08	43.6
35	27	25	Utah	2,706	279	10.3	1.07	37.8
36	15	36	Washington	5,391	678	12.6	1.04	28.5
37	44	34	North Dakota	1,217	68	5.6	1.03	28.9
38	31	23	Indiana	8,315	640	7.7	1	43.1
39	46	47	South Dakota	1,559	80	5.1	1	18.4
40	16	27	Connecticut	2,780	351	12.6	0.99	35
41	34	37	Wisconsin	6,870	522	7.6	0.93	27.1
42	51	41	Wyoming	1,550	49	3.2	0.91	23.5
43	29	24	Ohio	11,807	1,012	8.6	0.88	39
44	41	44	Maine	1,716	108	6.3	0.82	20.4
45	49	42	Idaho	2,365	119	5	0.79	22.3
46	47	35	Kansas	4,232	215	5.1	0.77	28.7
47	30	38	Minnesota	4,835	395	8.2	0.76	24.8
48	32	45	New Hampshire	1,294	100	7.7	0.76	19.7
49	45	46	Iowa	4,062	221	5.4	0.73	18.5
50	42	51	Vermont	743	45	6.1	0.72	13
51	50	48	Nebraska	2,362	91	3.9	0.51	16.2