Business Research Methods, 14e/Schindler1
>cases
State Farm, the nation’s largest auto insurer, distributed a list of the 10 mostdangerous intersections in the United States based on crashes resulting in claimsby its policyholders. What started as a study to reduce risk turned into an ongoingstudy that directs a major public relations effort: State Farm provides funds forcommunities to further research their dangerous intersections and initiate improve-ments based on the research. This case tells you how the State Farm DangerousIntersections initiative got started and how it is done. www.statefarm.com
>Abstract
>The ScenarioState Farm Insurance has a rich history of proactive safety involvement in auto andappliance design to reduce injury and property loss. In June 2001, State FarmInsurance, Inc., released the second report in its Dangerous Intersection reportingseries. State Farm modeled its program after an initiative by the Insurance Corporationof British Columbia, Canada (ICBC), and the American Automobile Association ofMichigan (AAA) to help position the nation’s largest auto insurer as the most safety-conscious insurer. ICBC had patterned its program on an earlier effort in Victoria,Australia. AAA, in turn, benchmarked its program on the ICBC program. AAAinvited State Farm to help fund one of its intersection studies. State Farm saw this asan opportunity to expand its effort into a nationwide campaign in 1999. “The 2001study is part of a larger effort focused on loss prevention and improving the safety ofintersections around the U.S.A.,” shared State Farm research engineer JohnNepomuceno. State Farm has allocated significant resources as well as funds to theinitiative. Since its inception, every city with an intersection on the overall list ofdangerous intersections is eligible to apply for a $20,000 grant to defray the cost ofa comprehensive traffic engineering study of the intersection. Additionally, each citynamed to the national top 10 dangerous intersection list is eligible for a grant of$100,000 per intersection to defray some of the cost of making improvements. Alltotaled, State Farm offered $4.44 million to the safety initiative in its first year.
Due to its large market share, State Farm is the only U.S. insurer in a position tomine its databases for the requisite information on accidents to come up with aviable U.S. list. But it found that although it had the interest to do so, its data warehousedid not have sufficient information to tally accident rates for intersections. To rectifythis, in 1998 State Farm included a location field as part of the data that its claimsadjusters regularly complete. This location information, in open-text format, indicateswhether the accident took place in an intersection or as part of an incident related toan intersection accident, and identifies the intersection. Following the 1999 study,the fields for identifying intersections were further refined.
In the first study using 1998 data (reported in June 1999) as well as the 2001
State Farm:Dangerous Intersections
Used with permission ofPamela S. Schindler©2001.
Business Research Methods, 14e/Schindler
2
study, State Farm looked at accidents involving only intersecting roads. Theyexcluded any accident that occurred at the intersection of a road and a highwayaccess or egress ramp. State Farm also looked only at accidents where the StateFarm–insured driver was at fault.
Because of the study’s focus on road safety engineering, the first study ignoredaccident severity and made no attempt to isolate demographic (age or gender ofdriver, driving record, etc.) or geographic (weather conditions, population of area,etc.) factors related to the accident. It also looked only at State Farm’s own internalincident reports, not at any public records involving traffic patterns or volume orpolice incident reports. Based on industry market share information, State Farmwas able to estimate the total number of crashes at a given intersection. “There wasgood reason to exclude police reports and traffic counts,” explained Nepomuceno.“The reporting threshold for police filing reports on accidents differs widely fromjurisdiction to jurisdiction. Some will only fill out reports when personal injury orcriminal behavior is involved. Others will fill them out only when a vehicle is damagedto the degree that it needs to be towed from the scene. Still others fill out suchreports on every incident. Traffic volume reports are often prepared infrequentlyand often by independent sources. Not only may the data quality be questionable,but the time period in which the data was collected may not match our 1998 incidentreports in every city involved. Also, when traffic volumes are factored in, low volumeroads with relatively few crashes are often deprioritized. Now that we’re throughwith the 2001 study, we are asking ourselves if intersection volume should be factoredin, and if so, how it can be included without significantly increasing our effort in dataprocessing.”
In the 1998 study, State Farm identified 172 dangerous intersections. The top10 most dangerous intersections in the United States were released publicly(www.statefarm. com). Public affairs staff for each state could request that up to 10intersections be identified for their state. “This was usually determined by the resourcesthat our local public affairs staff were willing to put toward the program,” sharedNepomuceno. “Each state had to recognize a top 10 national intersection, but theycould request that no more be released or that up to 10 intersections within theirstate be released.” As of August 2001, 97 cities (56.4 percent) had applied forState Farm grants.
“While some in the media claimed we had ‘hit a home run’ with the program, wequickly learned that there was a lot more at stake than we had anticipated in generatinggoodwill with transportation engineers,” indicated Nepomuceno. “This is, after all, atraffic safety program and we would not achieve that goal without having thecooperation of the traffic and transportation engineering community. First, whileinitially they lauded us for the attention our listing brought to traffic concerns, we andthey soon discovered that the spotlight generated demand for immediate solutions,solutions that they often didn’t have budgets to implement. Also, from theirperspective, not all accidents are the same; locations with accidents that result ininjuries and death should be given more attention. Some jurisdictions were upsetthat we didn’t consider intersection volume and we didn’t include accident ratedata.1 The fact that the State Farm grants were intended to study the intersection
State Farm: Dangerous Intersections
Business Research Methods, 14e/Schindler
3
more completely wasn’t always seen as a solution to their immediate problems.”To include accident severity, State Farm needed a measurement system for
classifying accidents. For the 2001 study, which used 1999 and 2000 accident data,State Farm calculated a median property damage accident payout (approximately$1,700). Incidents requiring payout of more than the median amount were classifiedas “high severity”; those requiring less, “low severity.” Additionally, State Farm choseto classify each accident using a multipoint scale. Zero was assigned to “no propertydamage, no personal injury” incidents and a higher number was assigned to “Highproperty damage, personal injury” incidents, with numbers in between assigned tolevels of property damage and personal injury (see Exhibit SF 1–1). Accident scoreswere summed to create an aggregate danger index for each intersection. Eachintersection was then weighted by dividing the danger index by State Farm’s marketshare in the area. Of the 224 intersections identified, the top 10 were released to thenational media. Each of those 224 is now eligible for the $20,000 grant to study theintersection to identify specific improvements; the top 10 are also eligible for $100,000grants for improvements. In this second round, State Farm has committed $5.48million to the safety program.
State Farm is making plans to track the success of the Dangerous Intersectionprogram. Once cities notify them of the completion of an intersection’s improvements,State Farm will start tracking accidents for that intersection for a period of one year.The first post-improvement evaluation study is expected in 2002. Additionally, StateFarm is taking steps to learn from the characteristics of the dangerous intersections.Each grant application for an affected city’s study of a dangerous intersection mustinclude:
• Collection and analysis of police report data.• An engineer’s “geometric review”2of the intersection.• A capacity profile of the intersection.• A traffic conflict study.3• A benefit-cost analysis.• A schedule of improvements (short-term, intermediate-term, and long-term).4
State Farm plans to use the new data to identify patterns of problems. This maylead to a model of desired intersection traits against which improvement plans canbe assessed, further increasing the effectiveness of the loss prevention program andmaking life a little easier for the transportation engineers with whom they must partnerto achieve safety success.
Exhibit SF 1–1 Danger CodesNo Personal Injury With Personal Injury
No property damage 0 Y
Low property damage 1 1 + Y
High property damage X X + Y
State Farm: Dangerous Intersections
Business Research Methods, 14e/Schindler4
1 Identify the various constructs and concepts involved in the study.2 What hypothesis might drive the research of one of the cities on the top 10
dangerous intersection list?3 Evaluate the methodology for State Farm’s research.4 If you were State Farm, how would you address the concerns of transpor-
tation engineers?5 If you were State Farm, would you use traffic volume counts as part of the
2003 study? What concerns, other than those expressed by Nepomuceno,do you have?
>>>>>Discussion
>>>>>Sources
>Notes1 Accident rate is calculated by dividing the number of accidents in a given period by the total
traffic volume over the same period.2 The intersection geometry or physical layout of the intersection can play an important role
in influencing driver behavior at intersections. For example, a curve on the approach to anintersection may impede the sight distance to a traffic signal, preventing drivers fromstopping in time. Or, a driver approaching two signalized intersections very closely spacedmay see the traffic signals at both intersections and become confused about which trafficsignal to obey.
3 A traffic conflict study is an observation study of traffic conflicts that do not necessarilyend in an accident but have the potential to do so. Recording and studying observations ofdriver behavior is expected to help traffic engineers understand the same problems thatresult in collisions. Some examples of conflicts are the sound of sharply applied breaks;sudden, unsignaled lane changes; or drivers hitting their horns.
4 Short-term improvements might include sign changes, changes in lane markings, or signal-timing changes. Most short-term improvements can be implemented in less than two years.Intermediate-term improvements might involve lane widening, the addition of turn lanes,etc., and be accomplished within 2–5 years. Examples of long-term improvements includegrade separation of the intersecting roads and can take 5–10 years to implement.
This case is based on information provided by John Nepomuceno in interviews that tookplace on August 9, 2001, and September 13, 2001. Other sources include: “Miami Area IntersectionTops State Farm List of Most Dangerous in the United States,” State Farm press release, June27, 2001 (http://www.statefarm. com/media/release/danger00.htm); “Research” (http://www.statefarm.com/media/methods.htm); and “State Farm 1999 Dangerous Intersection NationalStatus List” (http://www.statefarm.com/media/statustop.htm); and “State Farm’s DangerousIntersection Initiative,” Institute of Transportation Engineers press release, June 27, 2001(http://www.ite.org/press_release.htm).
State Farm: Dangerous Intersections
