Because existing sources of auto reliability information left much to be desired, in late 2005 I started conducting my own reliability research at TrueDelta.com. TrueDelta reports absolute stats like "trips to the shop" that make the differences between cars much clearer. Results are updated four times a year, so it's possible to provide reliability information on new models quickly. Also, any significant changes in a model's reliability become apparent quickly.

I’ve started collecting data on many competitors to the Accent, including the Honda Fit, Nissan Versa, and Toyota Yaris. But I need more Accent owners signed up to help out before I can start collecting data on the small Hyundai.

Participants simply report repairs the month after they occur on a one-page survey. When there are no repairs, they simply report an approximate odometer reading four times a year, at the end of each quarter.

To encourage participation, panel members will receive full access to the results for free.

For the details, and to sign up to help out:

Auto reliability research

This post has been edited by mkaresh: Nov 14 2007, 11:17 AM

If you quit after you sign up, this guy will ask you, "why did you sign up in the first place?"

If you simply don't have time to reply to his e-mails, you shouldn't sign up for the survey.


It's kind of weird to have reliability research with less than 20 samples sizes.

Less than 20? With this month's results I require 25 to even post official results, and for some models I have two to three times that.

And of course I send an email to everyone who quits asking why they quit. It's important to find out what I need to improve.

Should Hyundai try to speak with everyone who trades a Hyundai in on a different kind of car to find out why they didn't buy another Hyundai? Absolutely.

As for not signing up to participate in the surveys unless you actually plan to participate in the surveys...I fail to see how this is a revelation. They do only take about ten minutes a year.

Ultimately it's a matter of priorities. You did find the time to make over 500 posts to this forum. Sorry my research wasn't for you, I try to please everyone, but it's simply not possible.

This post has been edited by mkaresh: May 12 2007, 11:21 PM

only 25 required?

it can never be trusted.

Why not? It's not ideal, but it's a start, and does provide a rough indication of reliability.

Let's say you asked 25 Sonata owners how reliable their cars have been. Would you consider the information they gave you worthless, because you only spoke to 25 people?

Honestly, going in, I wasn't sure whether or not 25 would be enough. But I'm now about to issue my third quarter of results, and in nearly all cases the numbers just don't change much from quarter to quarter.

For example, repair trips per year for the 2006 Sonata...
Year through 9/31/2006: 0.27 (26 cars)
Year through 12/3l/2006: 0.36 (30 cars)
Year through 3/31/20007: 0.24 (33 cars)

It's not perfectly stable, but I'm ready to conclude from this that the Sonata isn't in the shop much, since the average is around 0.6 and many models are in the shop about once a year.

The exceptions have been some new designs where the repair rate has improved each quarter. But is this because I'm not surveying enough people, or because the manufacturer had some early glitches and fixed them, such that they only affected the earliest cars? I suspect the latter.

If everyone who didn't participate because they feel the sample is too small participated, then the sample wouldn't be too small.

This post has been edited by mkaresh: May 12 2007, 11:46 PM

I've posted the latest results:

TrueDelta Vehicle Reliability Survey results

Would like to include the Accent in the future, but need more cars signed up first.

Still hoping to get started soon with the Accent. The latest results included a number of Hyundai models, but not the Accent.

For the details, and to sign up to help out:

Vehicle reliability research

Getting only survey results from less than 100 people isn't valid information. it's welll known that 60% of your responding surveyers will be people that are unhappy with there items in some way, What you need to do is do a survey of 100+ people and than find the median of the min, med, and max of the results and that's it what you post. There is a lot to survey's to get acturate data. But please don't get me wrong what youare doing is great and is helpful, It's just that a person reading it has to know that how you got your information to know that it isn't gossiple but just some thought that they might want to ask when dealing with the dealerships.

i gotta agree with komptek.....

I'm sorry, but this simply isn't true. While studying for my Ph.D. at the University of Chicago, I worked for five years at the National Opinion Research Center, one of the top places in the country for survey research. I know what I'm doing here, and the results so far bear me out.

Consider just this: in any given month, about 10% of respondents report a repair trip, and some of these repair trips are follow-ups on earlier trips when a part was ordered, etc. For many models, the great majority of people never report a repair.

One explanation might be the research design. People can only report problems that happened the previous month, and participation is continous. So even if people sign up because they've had trouble with their car in the past, this cannot bias the results.

With other surveys, participation is not continous, and people can report of repairs that happened up to a year earlier. With these, what you say can be very true.

With my methods, I'm seeing surprisingly solid results with sample sizes as low as 25, and somewhat useful results with sample sizes as low as 15.

Would I like larger sample sizes? Sure, and they're getting there. This month there have been 120 responses so far for the 2007 Nissan Versa, and the sample sizes are around 100 for a few others.

Finally, please consider that "I'm not participating because your sample size is too small" is self-fulfilling.

The latest results were recently posted. For those of you fixated on sample size, that for the 2007 Nissan Versa was 123, and a few others were around 100. So they're growing.

Getting closer to including the 2007 Accent in this research. A half dozen more owners would enable data collection to start.

Auto reliability research -- need more Accents!

like what i see here, legitimizes the whole deal enough for me not to crack open a statistics text book & research sample sizes and so forth.

*signs up* with my 1 month old canadian model 2007 hyundai accent sport.

Thank you for signing up. I generally don't say anything about the Ph.D., since the results will speak for themselves.

The Accent is still not included in the main reliability survey, but it's getting close. Need just six more owners of the 2006.

Auto reliability research -- need more Accents!

The 2007 is now just a few owners short of the minimum needed to get started, so hopefully soon. Other model years also getting closer.

Hi mkaresh...I've been giving repair & MPG data to your website for my 2007 Dodge Caliber. Thank you for your passionate work to give people accurate automobile information untainted by opinions.

I have just placed my wife's 2008 Accent into your website & hope other people here will know the value of adding their information to your website.

Thank you for the encouragement, it helps.

There are now 19 2007s signed up - only need six more to start.

For the details, and links to the enrollment form, go here:

Vehicle reliability research

This post has been edited by mkaresh: Mar 4 2008, 02:29 PM

Now only need four more 2007s signed up to include that year in the survey. Other years also getting closer, but not this close.

For the details, and to help get your model year included:

Vehicle reliability research

we must have enough, as i've been receiving emails from truedelta with results of other models etc for a few months but just now received my first email requesting data from me on my accent. completed.

for those that haven't signed up, it took me less than 4mins to fill it out - including reading everything. next month's survey will likely be able to be completed in less than 2mins since i won't have to read everything over again. very quick process for building some valuable info for ourselves & others, imo.