"...the odds it will be a cherry."

What about the odds it will be copied by the Chinese and sold as a Chery? That'd be a neat statistic.

;-)

Reply

↓↑report

I've long thought there is a need for better reliability information for cars, and I hope that TrueDelta is succesfull in adding another point of data to the mix. The problem with sites like this, as well as sources like Consumer Reports and JD Power, is that they only draw their data from a small segment of the total auto population. Normally a small segment representing the total population is how statistics are derived for everything from political polls to the effectiveness of medications. The issue with TrueDelta/Consumer Reports/JD Power data is that it is from surveys that people respond to voluntarily, which means that the data comes from only those people that are willing to take the time to respond, and that are part of the self-selected group (i.e., readers of Consumer Reports, etc.), who probably don't accurately represent a real cross-section of a particular group. A prime example is when Consumer Reports provides data on products that are sold in smaller numbers (i.e., the Passat W8); Because the sample is so small, any issue that shows up will have a larger impact on the data. Or how about JD Power's initial quality surveys that gave low marks to the Mini or Hummer H1? They were skewed by a large number of responses for lack of cup holders in the Mini, and poor gas mileage in the Hummer.

Now it is true that the above mentioned sources have been able to provide a good general sense of the reliability and quality of autos, but wouldn't it be great if we had access to the manufacturers data? Then we could evaluate the quality of a car or brand based on ALL the data! Hmm, how to get the manufacturers to turn over that data. Guess that will have to wait until the Donkey party is running the show in DC.....

Reply

↓↑report

I'm the founder of TrueDelta. Thanks, Jeff, for writing about my site.

The issue of self-selection has come up a few times before. My research will actually suffer less from this source of bias, as people will only report trips to the shop that occur after they join the panel. People also will not be able to sporadically participate and remain members, as this harms the quality of the data.

In contrast, when people decide whether or not to participate in the CR survey they already know what they will report and they very well could participate some years and not others. It is quite likely that people who have had problems are more likely to participate, as they might have a grievance to voice. It has long been known that people dissatisfied with their cars are far more likely to tell others about their experience. The same might apply here.

As for J.D. Power, as you note it is very unclear what they are measuring. Real issues, or just sources of dissatisfaction? I will be focus on real issues, things that actually had to be repaired. This will minimize the impact of subjectivity, further reducing the effect of various sources of bias.

Reply

↓↑report

My apologies, John. It seems your first and last names recombined in my mind. There's no way to edit a comment, is there? I'm slowly realizing the purpose of the "preview" button...

Reply

↓↑report