Pulscar P0301 analysis — method Pulscar analysis of NHTSA public data, September 2026. Built 19 September 2026. Licence: CC BY 4.0. Credit "Pulscar" and link to https://www.pulscar.io/blog/p0301-code SOURCES - NHTSA ODI complaint flat files COMPLAINTS_RECEIVED_2015-2019 / 2020-2024 / 2025-2026. Field dictionary: https://static.nhtsa.gov/odi/ffdd/cmpl/CMPL.txt 777,040 distinct vehicle complaints (ODINO, product type V) received 1 January 2015 to 8 September 2026. - NHTSA manufacturer communication (TSB) flat files TSBS_RECEIVED_2015-2019 / 2020-2024 / 2025-2026. Field dictionary: https://static.nhtsa.gov/odi/ffdd/tsbs/TSBS.txt - Bulletin PDFs from static.nhtsa.gov/odi/tsbs//MC--.pdf, read as text. CASE SENSITIVITY Complaint narratives are ALL CAPITALS through 2020 and mixed case after. Every text query below is case-insensitive (SQL LIKE on a lower-cased column, or a Python regular expression with re.I). QUERY 1 — the complaint set descr matched on the whole word \bP0301\b (case-insensitive) -> 315 complaints. This is the same rule and the same count the Pulscar Check Engine Code Index publishes. A looser rule (the characters P0301 anywhere) returns 336. All 21 extra rows were read: 19 are genuine P0301 reports whose code carries a scan-tool suffix (P030100), is written in a range (P0301-P0304) or is run together with the next word ("P0301and"); 1 is a Lexus row writing P0301300; 1 names cylinder 3 in the same sentence. So the 315 is a floor, conservative by about six percent, and we say so rather than quietly widening the net. FALSE-POSITIVE READING: 30 complaints drawn at random (Python random.sample, seed 20260919) were read in full before any count was published. All 30 are genuine reports of a cylinder 1 misfire code. 0 of 30 false positives. SEVERITY Across the 315: towed 2.54%, crash 0.95%, fire 0.95%, 2 injuries, 0 deaths. All-complaint baselines across 777,040: towed 5.11%, crash 4.37%, fire 1.82%. MILEAGE 84 of 315 state a usable odometer (rows with 0 or 400,000 and above dropped). Median 80,320.5; quartiles 63,250 and 110,000; 10 at or under 36,000; 17 at or under 60,000; 26 at or above 100,000. The Check Engine Code Index publishes 80,641 for the same 84 reports because it quotes the upper of the two middle values; this file averages the pair, as a median of an even-numbered set conventionally is. The two middle readings are 80,000 and 80,641. The gap changes nothing. QUERY 2 — is cylinder 1 special? (p0301-cylinder-code-comparison.csv) For each of P0301 to P0308, counted over every complaint whose narrative contains the string "P030": (a) complaints naming that code as a whole word; (b) complaints naming that code and NO other cylinder code and NOT P0300 — the true lone-cylinder population; (c) distinct NHTSA manufacturer-communication documents whose summary names it as a whole word. Result: P0302 leads on (a) with 392 and on (b) with 267; P0301 is 315 and 174. Cylinders 5 to 8 are rarer because most engines in the file have four cylinders. QUERY 3 — what owners name (p0301-what-owners-name.csv) Regular expressions, case-insensitive, applied to the whole narrative. Key ones: coolant or antifreeze: coolant|antifreeze head gasket: head gasket piston rings: piston ring|oil ring spark plug: spark ?plug ignition coil: \bcoil fuel injector: injector valvetrain: \bvalve(s)?\b|lifter|camshaft|rocker|lash adjuster compression or leak-down: compression|leak.?down|leakdown engine or block replaced: engine (was )?(replaced|replacement)|new engine |short block|long block|replace the engine a recall, extension or settlement: recall|warranty extension|extended warranty|settlement |class action|customer satisfaction ignition parts reported as fitted: (replac\w*|chang\w*|new|install\w*|put in)[^.]{0,70}(spark ?plug|ignition coil|\bcoil) |(spark ?plug|ignition coil|\bcoil)[^.]{0,70}(replac\w*|chang\w*|were new|installed) persistence word: still|again|came back|returned|reoccur|recur|persist|did n.t (fix|help|solve|resolve) |no difference|no luck|made no difference|same code Groups overlap; a complaint counted in one may be counted in others. These measure vocabulary, not diagnosis: "spark plug named" means an owner typed it. Headline results: 85 of 315 (27.0%) report ignition parts fitted, and 54 of those 85 (63.5%) also contain a persistence word. THAT IS SELECTION BIAS, NOT A FAILURE RATE: an owner whose $40 spark plug fixed the car does not file a federal defect complaint. The figure describes the complaint file, and the article says so in the same sentence. Only 6 of 315 describe swapping a coil or part between cylinders, and 20 of 315 (6.3%) mention a compression or leak-down test. QUERY 4 — by make (p0301-complaints-by-make.csv, p0301-top-models.csv) The same phrase groups computed per make and per manufacturer group. Groups used: Ford group = Ford, Lincoln, Mercury; Honda group = Honda, Acura; GM = Chevrolet, GMC, Buick, Cadillac; Toyota group = Toyota, Lexus; Hyundai; Kia; Stellantis = Jeep, Dodge, Chrysler, Ram, Fiat. Headline results: Ford group 137 complaints, coolant or head gasket named in 84.7%, piston rings in 0%, an engine or block replacement in 48.9%. Honda group 81 complaints, piston rings 25.9%, fuel injectors 28.4%, spark plugs 43.2%, coolant or head gasket 3.7%, engine replacement 7.4%, a recall, extension or settlement 60.5%. Everyone except Ford and Honda: 97 complaints, coils 26.8%, plugs 28.9%, coolant 10.3%, engine replacement 10.3%. SMALL NUMBERS ARE SMALL: GM 18, Toyota group 10, Stellantis 10. Those rows are published because readers search for them, and the article labels them as too small to carry a share. Raw counts across makes reflect very different sales volumes and nothing here ranks reliability. QUERY 5 — 100 complaints read and coded (p0301-100-complaints-read.csv) A random sample of 100 of the 315 (Python random.sample, seed 20260919). Every narrative was read and given ONE primary reported outcome — what the report says a shop found or did, not what the owner suspects. Categories and counts: coolant intrusion, head gasket or cylinder head .................. 29 no diagnosis reached, not repaired, or could not be found ........ 19 other (sensor, software, recall, converter, turbo, or an engine replaced with no stated cause) ................................ 15 piston rings or oil-fouled plugs ................................. 14 fuel injector .................................................... 8 valvetrain: lifter, camshaft, valve, rocker or timing chain ...... 5 low or no compression found on the cylinder ...................... 5 ignition parts fitted (plugs or coil) as the reported action ..... 5 Borderline rule: a narrative that reports ignition parts fitted AND a later, deeper finding is coded to the deeper finding, because the question is what the fault turned out to be. One reader, one pass, no second coder; a 100-row sample puts roughly nine percentage points of uncertainty around a 29 percent share. QUERY 6 — manufacturer communications (p0301-bulletins-by-group.csv) tsbs.summary matched on \bP0301\b -> 480 distinct NHTSA document IDs from 14 groups: Stellantis 172, Honda/Acura 117, Ford/Lincoln 67, VW Group 23, Toyota/Lexus 22, Mazda 21, JLR 18, GM 11, Kia 9, Subaru 6, Hyundai 6, Nissan 6, and two heavy/off-road makers. THIS IS A DIFFERENT NUMBER FROM THE PULSCAR CHECK ENGINE CODE INDEX, WHICH PUBLISHES 276 BULLETINS FROM 12 GROUPS, AND BOTH ARE CORRECT. The index excludes helpline "analysis lists", Honda and Acura dealer messages and parts-collection requests, documents whose ID contains no digits, and heavy-truck, bus and motorcycle makers, and it folds every revision of a bulletin into one. The article cites 276 for bulletin counts and uses 480 only where the point is the raw document trail. Index method: https://www.pulscar.io/research/check-engine-code-index/method.txt WHAT WE DID NOT RECOMPUTE The recall picture for misfire codes (27 campaigns in sixteen years, three of them light passenger vehicles) was measured for the Pulscar P0300 guide and is cited, not recomputed. The 120-complaint outcome reading, the P0300 keyword counts and the Ford coolant-intrusion bulletin family are likewise the P0300 page's work and are linked rather than repeated. LIMITATIONS - Complaints are self-reported, unverified, and are not failure rates. - Counting them measures what reaches a federal form. Recall and settlement publicity drives that: 38.4% of this set already names a recall, extension or settlement. - Model year is a blunt age proxy; warranty clocks run from the in-service date. - 315 reports is a small file. Every per-make figure below about 20 reports is reported as a count, not as a share.