To me reliability is relative. In physics, statistics, quality engineering, etc... reliability is totally different definition as quality. They should be treated separately, but 90% of people mix them up.

We have capacity of the component/system (~quality) vs the component/system set to operate in (~load), how long does it last vs designed life time (reliability)?

For a high performance component that designed to last 2 years, it fails after 3 years under designed load. That's good quality and good reliability
For a low performance component that designed to last 10 years, it fails after 5 years under designed load. That's poor quality and poor reliability

But people only see the 3 and 5 years when the component fails, call the 3 years not reliable.

Designed load is also a key. Car that target average consumer A -> B, barely see 3000rpm vs fun car that see redline and hard corner all the time.
Econobox can have tonnes to software to prevent drive from doing stupid things, and tune the engine/transmission to run a lower load relative to the potential capacity.
Econobox drivers also tends to less fussy about handling (I know some the King owners don't agree), mostly suspension no sound = good. Car enthusiasts, can tell the shock support is not good, or the rubber of ball joins worn out from the handling. Then they replace early.

It is not apple-to-apple comparison. If we set same conditions, put all cars at their redline the most reliable car brand is probably Porsche... LOL.

See below, they talking about the engine load at % of their capacity.

You may jump to 10:40 - M3 Taxi vs M6 GT3


This post has been edited by constant_weight: Feb 23 2022, 06:32 PM

But we are sampling from my friend, my friend's friend, your friend etc. Do we collect large enough sample size throughout the country and globally that is statistically significant, and have an empirical value?

component 1 failure rate is x% over 1 year, y% over 2 years, z% over 3 year, then plot the trend of side by side for P and H? We don't

On top of that is another big reason to skew the perspective. The Peugeot parts are generally more expensive. Same failure on H is general wear and tear, the P fail one time the owners jump and complain over and over again, that goes the perception. Cost to fix for T and H are definitely lower, with tonnes of 3rd party, recond, even recond 3rd party parts available.

Did you subscribe to the paid detail report?

If yes, I'm sorry. If no, you need to look past the headline summary figured.

I strongly encourage you to research how the data is compiled. It helps you to really understand the context of data.

Otherwise you can read for another decade and don't understand how it works.