I think many of these 'reliability surveys' are a waste of time, partly for the reasons RobJP says, but also because:

a) It and others only looks at a specific period in the lifespan of the car - year 1 for the JD Power survey, and during the period that perople take out an aftermarket warranty (the 'Reliability Index' survey from Warranty Direct). The first often capture manufacturing 'faults', the second one, whilst some measure of longer term reliability, doesn't start off on an equal footing - a KIA would be more than 7 years old, whereas a Ford (for example) would only be 3 years old, and neither would the record of in-warranty replacements and official recalls before that period would be counted.

For example, a 3.5 year old Ford Focus that had had 3 recalls and gone through (replaced under warranty) two PS gearboxes would have newer parts in it than a 7.5 year old KIA Ceed which had no recalls or in-warranty replacements. If Ford had (say permanently fixed the reliability woes of the PS box (not yet), then the car would have a gearbox etc up to four years younger with less wear on it than the KIA car, but RI treats them as 'year 1 equal'. This, to me, is patently unfair to makes that are reliable over their warranty period and who give longer warranties anyway.

b) I wonder how objective they are, given that people have different expectations of what they buy, how it drives and how they will use it. An OAP who (like my parents) uses their car as a shopping car (with an occasional longer journey every few months, mileage around 3k a year) will mean they don't use all the toys (to me, it doesn't matter if a car has more or less, as they SHOULD all work fine for at least the length of the warranty period) such as the radio or CD player, and of course, they do far less mileage than someone using a similar car from another make for commuting, even if that's around or just below the annual service interval mileage.

For those doing 25k miles a year for commuting or as a company car on business (especially if used commercially like taxis), or even those cars used a lot by parents with kids, then 'reliability' can surely easily be mixed up with 'wear and tear'. I'm not sure this sort of thing is really taken into account, as surely it could make a HUGE difference if not.

- - - - - - - - -

I think what is needed is an ongoing survey over the life of cars, say a 10 year (or longer) period, and also should take into account all the above I've stated and the cost-benefit of, for example, pricier but more reliable parts against cheaper and less reliable ones. This would especially benefit those makes that offer longer warranties, especially on parts that other deem 'wear and tear' on their shorter warranty periods, so they currently 'get away' with not having to replace parts for free, even if the EU rules sort-of state that many should last 6 (?) years without fault if used/maintained correctly.

The same goes for servicing - if a firm says their cars need servicing (especially oil changes) less often than another, then it would be only deemed more reliable if those parts actually lasted the interval period and didn't (as long oil changes seem to do) contribute to other faults. I suspect that customer service does have some bearing on the JD Power survey, but I'm unsure as to how much, as car firms who are 'less than willing' to honour geneuine claims should be marked down than those who do, especially when they are serviced at a main dealer even outside the warranty period (even when that's a part 'goodwill' contribution').

Reliability, or at least how good a make is, should include the length of time a car is off the road (owner inconvenienced or how often they have to revisit the dealer with the same problem that wasn't fixed the first X times) and how many times it took for it to be fixed.

Sorry to go rather off topic here.

This doesn't take into consideration the number of bmws sold compared to volvos

Since the figures are "per hundred cars", it does take that into account.

No because it could per 100 out of 1000 volvos or per 100 out of 10000 audi's

I think you'll find that per hundred cars is the ONLY way to compare fairly.

Absolute garbage

If you understand statistical comparisons, then please explain how they should be made

If 10 Volvos fail out of 1,000 sold, that's 1 per 100 - if 100 out of 10,000 Audi's sold fail, that's also 1 per 100 - what other way is there to compare when volume sold is different.

Absolute garbage

If you understand statistical comparisons, then please explain how they should be made

If 10 Volvos fail out of 1,000 sold, that's 1 per 100 - if 100 out of 10,000 Audi's sold fail, that's also 1 per 100 - what other way is there to compare when volume sold is different.

There should be a mileage basis as well - a car that only does 3 miles a year wears differently to one that does 10k or 25k miles, as one that does short urban vs longer motorway trips. Numbers alone cannot be accurate.

Just to chime in again on this rumbling argument - Sorry Nathan, but I'm not agreeing with your side when I put forward my 'argument' about adding a mileage component to the surveys. RT is dead right in their logic about the percentages stats (if taken separately from any other), i.e.

I failure in 100 - 1% failure rate

100 failures in 10,000 - also 1% failure rate

What WOULD be significant is when the sample size is deemed too small. In the past, some car manufacturers have been excluded from customer satisfation surveys precisely because they sell too few cars to be able to get an accurate result. Lexus isn't (these days at least) one of them, Tesla may well be, at least in the UK. What that minimum level of sample size is would be down to a mathematician or statistician, whom I'm sure knows their stuff.