How We Built the Metro Phoenix Intersection Study: Data, Definitions, and Verification

This page documents exactly how the Metro Phoenix intersection study was produced, so any figure in it can be traced back to a defined rule. AZ Crash Data publishes methodology alongside every original analysis.

Data Source

The underlying data is crash-level records from the Arizona Department of Transportation''s Accident Location Identification Surveillance System (ACIS), obtained by public records request and covering all reported crashes in calendar years 2023 and 2024. Each record represents one reported crash, with fields including date and time, city, geocoded location, the primary road and crossing feature, whether the crash occurred at an intersection, and totals for persons injured and killed.

Federal law provides confidentiality protections for certain crash data compiled for safety purposes (23 U.S.C. § 407); this analysis uses records released by ADOT through its standard public records process and reports them in aggregate.

Study Universe

The study covers crashes coded to 23 incorporated Metro Phoenix cities and towns: Apache Junction, Avondale, Buckeye, Carefree, Cave Creek, Chandler, El Mirage, Fountain Hills, Gilbert, Glendale, Goodyear, Guadalupe, Litchfield Park, Mesa, Paradise Valley, Peoria, Phoenix, Queen Creek, Scottsdale, Surprise, Tempe, Tolleson, and Youngtown. Crashes in unincorporated county areas are not included. Within that universe, the two years contain 161,753 reported crashes, of which 68,281 are coded as intersection crashes.

How an "Intersection" Is Defined

A crash counts toward an intersection when ADOT''s own intersection indicator marks it as an intersection crash. The intersection itself is identified from the crash record''s geocoded primary road and crossing feature. Because the same physical intersection can appear with its two road names in either order, the two names are treated as an unordered pair — "67th Ave & Indian School Rd" and "Indian School Rd & 67th Ave" are counted as one location.

Surface Intersections vs. Freeway Interchanges

Junctions whose coded road names identify frontage roads, ramps, or non-cardinal frontage alignments are separated into their own table rather than ranked against surface intersections. A single freeway interchange complex spreads its crashes across several coded locations (each frontage-road crossing, each ramp terminal), so mixing interchange junctions into a surface-street ranking would understate some locations and overstate others. Both groups are reported; they are simply reported apart.

Counting Rules

• Crashes = the number of reported crash records coded to the intersection over 2023–2024 combined.
• Injury crashes = crashes in which at least one person was reported injured.
• People injured = the sum of persons injured across all crashes at the location.
• Fatal crashes = crashes in which at least one person died.
• Deaths = the sum of persons killed across all crashes at the location.
• Intersections with equal values on the ranking measure are listed in no particular order; no secondary tiebreaker is implied.

Verification Checks

Three checks were run before publication:

1. Duplicate check. The top-ranked location (99th Ave & Lower Buckeye Rd, 180 crashes) was tested for record duplication: all 180 records carry distinct incident IDs, the 2023 and 2024 tables share no overlapping IDs, and every record carries the correct crash year.
2. Coordinate check. The latitude/longitude of those same records cluster tightly at the physical intersection (a spread of roughly one city block), confirming the crashes are genuinely located there rather than snapped in from elsewhere.
3. External cross-check. The Maricopa Association of Governments'' most recent published regional intersection study (2018–2022 data) independently recorded more total crashes at 99th Ave & Lower Buckeye Rd than at any other ranked intersection — consistent with our 2023–2024 finding that it remains the region''s highest-volume crash location.

Known Limitations

• Reported crashes only. Crashes that are never reported to police do not appear in ACIS, and reporting practices for minor property-damage crashes can vary by agency.
• Geocoding. Locations reflect ADOT''s geocoding of each crash. Crashes very near an intersection can be coded to it, and complex junctions can split across multiple coded names.
• City coding. Each crash carries one city code. Intersections on municipal boundaries may have their crashes divided between two cities'' totals.
• Volume, not risk. These rankings count crashes; they do not adjust for traffic volume and are not an engineering risk score. A high-volume intersection can rank high while being no more hazardous per vehicle than a quieter one.
• Point in time. Signal timing, construction, and roadway changes since 2024 can change current conditions at any listed location.

Questions About the Data

Corrections and questions are welcome — contact information is on our About page. Requests to reproduce tables from this study are granted with attribution to AZ Crash Data and ADOT as the underlying source.