"The plural of anecdote is not data."

I don't think individual posters' experiences help overall very much, interesting though they are. The difficulty is where to find the data. The JD Power survey is all very well but it suffers from sampling bias, as you need to volunteer to take part. Same applies to owners' reviews here or elsewhere.

Looking for problems on websites leads to a different source of bias - Fiestas are the best-selling car in the UK and if they were as reliable as, say, the Civic, you'd expect a higher total number of Fiesta faults as there are more of them to go wrong.

So my conclusion is that it's very hard to be sure of the facts.

What I do think, however, based on anecdote and observation, is that cars of this era are significantly more reliable than in the past. I remember the sounds of cars trying to start on my street of a winter's morning - Fords seemed to have a characteristic noise from their starter motors, for example. It was also quite common to return a new car for several faults to be fixed that were present on delivery; much less common now I think.

Because cars are (a) highly complex and (b) highly reliable, to a great extent it's a matter of luck if your particular vehicle has problems or not.

Because cars are (a) highly complex and (b) highly reliable, to a great extent it's a matter of luck if your particular vehicle has problems or not.

Err no.Put simply, if a car is composed of say 5,000 parts which could fail, the risk of any failure is based on the most at risk part. The chances of zero failures of any part is (100% - % chance of that product). Usually time dependent - say over 5 years.. see also Mean Time Between Failure.

The chance of zero failures for the car is the result for each product. all multiplied together..

So unless the chances for failure are verry small for all products, the chance of one failing can be quite high - like >70%...

The Japanese do their engineering to minimise failures. Some companies design right on the borderline of failure - or in Ford's case - over it (CVT gearboxes as deliberate case in point).

Approximately right. The issue is complicated by the number of potential failures (and also by their severity). Our Civic had its interior mirror fall of at about 500 miles (trivial fault, fixed by glue, only mildly irritating). But a fault that makes a CVT gearbox fail is much more likely to cause time off the road and to colour the owner's opinion. But if the Civic had had a number of trivial faults, our opinion of its reliability might be as low as the Ford CVT owner's opinion of their car.

If you know the mean time between failures you can then calculate the likelihood of several failures using the Poisson distribution. It's multiple failures as well as major failures that affect perceptions, I think.

I seem to remember a story - may well be apocryphal - that Henry Ford kept a record of Model T failures and deliberately made parts weaker if that particular part virtually never failed. The argument was that if the part never failed, it was over-engineered; if it failed often it needed redesign; but if it failed very rarely it had the right balance between customer satisfaction and production costs. Maybe someone can verify this?

What I think happens, in my experience in Quality Management of different things, is overly zealous cost-cutting. The business decides it needs to trim costs and briefs the Procurement group to go looking for cheaper components. Trials of the cheaper component take place but in the rush to save cost, the results are often ignored. I have seen this a number of times in my industries! I believe a certain German manufacturer is very good at this, according to these boards!

Like VAG ignition coils...

You would NEVER get me buying an Aston Martin. Their reliability is staggeringly bad.

With the Italisns you expect a bit of Friday afternoon car syndrome but in reality they aren't that bad. Except from the habit of catching fire.

Go back to the 80's though and they were utterly hideous.