When it comes to mountains of data like the NHTSA vehicle complaint data base, one man’s junk is another man’s treasure(trove).  Yes, there are limitations. But the remarkable degree of similarity of the UA rate between most badge engineered cars, and GM’s consistently low UA ratings speak of statistical relevance to some of us. One of those is David Lapidus, (TTAC handle: David42), whose data diving and Excel skills vastly overshadow mine.

In the last data dive, I used numbers from Edmunds to come up with a UA complaint rate for 117 cars from MY 2005-2010, excluding the flurry of complaints filed with NHTSA after the Toyota mat recall 0f 9/29/09. But those are cumulative, over the five years. David has taken it to the next level, to the individual model year. We may come back to the whole list another time, but initially we decided to focus on the Toyota Camry, since its large volume of sales and UA complaints would inherently increase the statistical accuracy of the numbers. And although this data dive may not solve the UA puzzle, it does raise serious questions about the most recent recalls and the Congressional barbecue of Toyota. David sifted and examined complaints going back to 1995 (and cut off after 9/29/09) for all 1995 and newer MY Camrys for all UA incidents (vehicle speed control including the following areas: accelerator pedals, cables, cruise control, linkages, springs).

This following chart lets us read the annual complaint history for each model year:  for each model year, it shows how many complaints were filed per calendar year.  For example, in the upper-right corner you can see a bubble labeled with the number “11.”  (It’s all the way to the right, second from the top).  This means that for model year 2007 Camrys, people filed 11 complaints in calendar year 2009.

Some observations:

  • Among recalled cars (MY 2007-2010), only MY 2007 looks problematic.  That is the first year of the CTS pedal, as well as a new physical pedal design that could cause mat entrapment.  Yet model years 2008 and 2009 have barely any complaints at all.  Specifically, MY 2007 Camrys have 89 complaints (19 + 70) in the first two calendar years.  MY 2008 has only 7 complaints (3 + 4).  And though there’s only one year of data for MY 2009, the ONE complaint for that MY hardly suggests that the accelerator pedals or floor mat issue was a common UA issue with actual users.
  • There is/was obviously a serious problem with the previous-generation Camry (2002-2006).  MY 2002 was the year the first e-pedal was introduced. But its design was completely different from the CTS or Denso units used after 2007. And its physical pedal shape made mat entrapment unlikely, if not impossible. (I have taken detailed pictures of these various vintage pedals). We don’t know the specific nature or how serious the problem in the complaints is. (Complaints can be filed for causing anything from a minor inconvenience to death.)  We DO know that the 2002-2006 models caused an average of 2.2 to 3.9 injuries per calendar year.  That’s far more than the supposedly-dangerous 2007 model year cars, which caused an average of 1 injury per calendar year.
  • For most (but not all) model years, the second year is the worst.  Perhaps this tells us something about when the problematic component is most likely to fail.

Now here’s a simpler chart that shows only the year the complaint was filed—all model years are lumped together.  So it’s easy to see that the number of Camry complaints spiked in 2004, and stayed pretty high until 2007 (especially compared to the Honda Accord):

Now here’s the strange thing:  In 2008 and 2009, complaints dropped back down to the pre-2002 level.  It’s as if all the Camrys were suddenly cured, just before the recall and untold billions chopped from Toyota’s market capitalization. And what exactly was going with those MY ’02-’03 Camrys? That’s no statistical fluke. Whatever was wrong with them, how come they stopped generating complaints in 2008 and 2009? Can anyone help us make more sense of these numbers? We promise to share that one million dollar prize.

Get the latest TTAC e-Newsletter!


26 Comments on “NHTSA Data Dive 4: Did Toyota Recall The Wrong Cars?...”

  • avatar

    The year the complaint was filed vs. the model year of the car can throw things off.

    Dave’s earlier chart with the Camry, Accord, and Torus, which I believe was by model year, sealed it for me that in 2002 Toyota lost it and still has not gotten it under control. The latest chart, showing the complaints filed by year, does not make that point so clear,.

  • avatar
    crash sled

    “Whatever was wrong with them, how come they stopped generating complaints in 2008 and 2009? Can anyone help us make more sense of these numbers? We promise to share that one million dollar prize.”


    The exorcism finally took, and the demon was expelled.

    I want my 1/2 of the money in cash. ;-)

  • avatar

    The first chart (bubble chart) implies to me that in 2004, something caused a bunch of 2002-2004 Camry owners to file SUA complaints on the NHTSA website. Then, in 2007, something caused a bunch of 2007 Camry owners to file complaints. The 2007 activity could correspond to a product defect, that got corrected in 2008. It would be interesting to map this to when the CTS pedal designs were modified.

    The 2004 activity seems odd since it implies the 2002 cars took 2 years to fail, the 2003 cars took one year to fail, and the 2004 cars were defective upon delivery. Since this seems unlikely, I interpret this to mean there was a news event in 2004 about e-throttles causing SUA, and a bunch of people with the e-throttle Camries decided to file reports either because they had experienced SUA, erroneously thought they had, or were opportunistic.

    • 0 avatar

      Exactly! This is where one has to look at the actual complaints. I’ve read some where the writer states that the event happened at some time in the past, sometimes a year or two earlier. Report date is by no means the same as incident date.

    • 0 avatar

      This is an excellent question, and while there might be something like this going on, it accounts for a minority of complaints. Let’s look at what happened for a single model year:

      MY 2002
      -Failures that occurred in 2002: 8 filed in 2002, 4 filed in 2003, and 9 filed in 2004. So something happened in 2004 that caused reports to be filed for old failures.
      -Failures that occurred in 2003: 12 filed in 2003, 9 filed in 2004. So there was NO spike in 2004 for failures that occurred in 2003.
      -Failures that occurred in 2004: 47 filed in 2004.

      So, among the complaints filed in 2004, 27% had occurred in 2002-03 (9 + 9), while 73% occurred in that year (47).

      For MY 2003, the story is similar:
      -Failures that occurred in 2003: 10 were filed in 2003, and 16 were filed in 2004.
      -Failures that occurred in 2004: 46

      So again, there was a slight delay (26% of those filed in 2004 occurred in the preceding year), but a large majority of the complaints filed in 2004 are for events that occurred in that year.

      Of course, it’s possible that something occurred in 2004 (a news story?) that made people more likely to think their Camrys were suddenly accelerating. So we can’t discount that some sort of artificial inflation occurred in 2004, but these data don’t indicate it.

  • avatar
    crash sled

    “(vehicle speed control including the following areas: accelerator pedals, cables, cruise control, linkages, springs).”


    I hate spinning you 2 guys around, and you’re doing a marvelous job mining through this muck, and thank you so much for the diligent efforts, but if you can do one more machination, and somehow segregate cruise control issues? Anecdotally, I’ve heard talk about Toyota cc issues, and this might open up the analysis a bit for the B&B here, and even us slobs.

    I think you guys have done a remarkable job in potentially surfacing 2 discrete “event cycles” here. Well done.

    • 0 avatar

      Sorry, the data are hard to analyze for this question. Here is how the complaints break down:


      The problem is that the first category is by far the largest, which means that most of the complaints are not subclassified into the specific component which failed. And that category contains over 500 complaints, which alas I’m not up to reading through!

  • avatar

    How about this?


    http://www.theautochannel.com/news/2004/07/27/206132.html (scroll down to find the text)


  • avatar
    Telegraph Road

    Paul and David: Excellent! Are you able to tell which of the MY 2002-2003 Camry incidents have the 4-cyl 1MZ-FE engine, which was the subject of two Technical Service Bulletins (TSB EG008-003 and TSB EG017-02) for “engine surging”? See http://www.safetyresearch.net/toyota-sudden-unintended-acceleration/toyota-sudden-acceleration-timeline/

    • 0 avatar

      Sorry, most of the NHTSA records don’t include the number of cylinders in the vehicle. There is a field for cylinders in the database, but it’s empty for the vast majority of records. (And for the Camry complaints, they’re ALL empty.)

  • avatar

    Looking just at Camry MY2007 and VEHICLE SPEED CONTROL, VEHICLE SPEED CONTROL:ACCELERATOR PEDAL & VEHICLE SPEED CONTROL:CRUISE CONTROL, I found 11 claims, 8 of which were hesitation & 3 being SUA by looking at the description of the complaint. The graphs above change significantly if the Camry numbers are reduced by 2/3.

    Can anyone confirm if I’m looking at this correctly? I’m concerned that the ePedal hesitation is getting lumped in with the SUA and that this can’t be filtered out without reading through all of the descriptions (instead of the category).

    • 0 avatar

      It’s true, there may be some hesitation problems lumped in here. However, I just looked at a random sample of 9 complaints about MY 2007 Camrys, and only two of them involved hesitation.

      As you suggested, I don’t think we can get a more solid answer without reading and re-classifying all of the Camry complaints.

  • avatar

    When I’ve looked at the NHTSA complaints for various cars in the past, on thing I’ve noticed is the sheer number of obvious duplicate reports filed by the same person. The verbiage is either the same or parts of it are the same, indicating the same person posted it twice. Do the data in the charts posted here account for duplicate entries?

    • 0 avatar

      There was no attempt to remove duplicates, but I think that the problem is not severe enough to throw off the conclusions. Among the 618 complaints where the failure date was provided, 84% of them occurred on a day where no other failures were reported, so we can be pretty sure that these records are not duplicates. (If a report was a duplicate, we would expect it to have the same failure date as the original record.) Another 13% had exactly one other failure on the same day. Perhaps some of these are duplicates, but probably not all of them. There are another 18 complaints (3%) that occurred on a day with 2+ other failures. These are more likely to be duplicates, but they’re obviously a tiny fraction of the total.

  • avatar
    Gardiner Westbound

    We had a ’95 Toyota Camry, 4-cylinder, automatic. It was an excellent car in every respect. The cruise control occasionally acted up. Touching the brake pedal immediately corrected the problem with no ill effect. It never occurred to us to issue a complaint.

  • avatar

    “And what exactly was going with those MY ‘02-’03 Camrys? That’s no statistical fluke. Whatever was wrong with them, how come they stopped generating complaints in 2008 and 2009?”

    Simple. Within 5-6 years, the majority had been unintentionally accelerated out of existence.

  • avatar

    Has anyone heard if the recall will affect car insurance rates for Toyota owners whose cars are affected?

  • avatar

    2002 was the most resent “high” peak year for sun spots and 2007 was the most resent “low” peak year…


  • avatar

    There is something funny about this mess, did I read that ~40% of the UA incidents are cars with manual Trans? How is that possible? I would like to see correlation between weather and complaint date, were these people wearing winter boots? (Lots of people not used to winter driving did a lot of it this year), that is the only way I have ever came close to a UA event (7 Tacoma’s all manual Trans., 21 years, 1.9 million kilometers). Now I read that ABC is faking the videos, sounding a lot like the Audi fiasco (operator pedal misapplication)20 yrs ago.

    • 0 avatar
      prince valiant

      You can’t just look at raw complaint numbers. The complaints have to be normalized to a common metric such as complaints/100K vehicles or complaints/10K vehicles. To normalize the complaints, the US vehicle volumes for each model by model year are required. This is a more valid method of analyzing the compliant data for any trends or patterns.

  • avatar

    In late 2004, the NHTSA began investigating electronic pedals for their role in unintended acceleration.

    Here is a news article about it: http://www.usatoday.com/money/autos/2004-04-13-unintended-acceleration_x.htm

    And here the google news archive search where I found it: http://news.google.ca/archivesearch?as_q=unintended+acceleration&num=10&hl=en&btnG=Search+Archives&as_epq=&as_oq=&as_eq=&ned=ca&as_user_ldate=2004&as_user_hdate=2004&lr=&as_src=&as_price=p0&as_scoring=a

    You can play with the dates and see the relative frequency of stories for any time period.

  • avatar

    Could the flash update for the 2007 model year heasitation problem have also fixed the 2007 model year UA problem?

  • avatar

    Toyota and others knew they were having issues and attempted to hide it. All Car Companies should have came forward with a full disclosures of what car were dangerous. Instead of waiting for a huge media blitz and tons of public pressure. I never seen so many car companies GM – NISSAN – TOYOTA – HYUNDAI having recalls all at the same time. I had no idea my car was affected until I looked on http://www.carpedalrecall.com and found I had a bad Anti Lock control unit on my 2008 Pontiac G8 , my co workers Ford Truck had a recall also. So be careful

  • avatar
    George B

    I blame the 2004 movie Harold and Kumer Go to White Castle. The appearance of a Camry in that movie attracts a small number of much, much less safe drivers to the car. Becomes less socially unacceptable to drive a Camry while totally wasted. Herbally enhanced pedal misapplication?


  • avatar

    I’m in the money management business and deal with statistical data every day. A possible explanation of the high bar in 2004 is that it is not out of the ordinary and indicates nothing other than random variation. We can’t know if it’s statistically significant because we don’t have nearly enough years of data to meet statistical criteria, we’re not comparing the data to UA problems in other vehicles, etc. If UA incidents distribute themselves in a normal curve we could determine the significance of the 2004 numbers. I’ve repeatedly wondered about this Toyata media slaughter. Did significantly more consumers die or get injured in Toyotas in 2004 than died or got injured in other makes and similar sized models per 10k vehicles? I sorta kinda doubt it.

  • avatar

    Toyota in general and the Camry in particular** have suffered first-year teething problems, enough to see CR revoke their post-redesign automatic recommendation. What you might be seeing is the result of those troubles.

    I asked this question once before: when did the NHTSA make their complaint system accessible via the web?

    ** as well as the 4Runner, Avalon and Sienna

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • threeer: I’m conflicted on this one. I spent over a decade living just outside of Charleston (Summerville) and...
  • jalop1991: AV club…audiovisual club…
  • Scoutdude: The other advantage the GM twins have, at least for those fleets that already own them is the ability to...
  • DeadWeight: Fans of Tesla & Scientology have much in common!
  • DeadWeight: Honestly, this looks very similar in exterior design, with the exception of the front clip, to the Tesla...

New Car Research

Get a Free Dealer Quote


  • Contributors

  • Matthew Guy, Canada
  • Seth Parks, United States
  • Ronnie Schreiber, United States
  • Bozi Tatarevic, United States
  • Chris Tonn, United States
  • Corey Lewis, United States
  • Moderators

  • Adam Tonge, United States
  • Kyree Williams, United States