Reliablity index is always flawed, because it does not account for number of reports. That way companies not present and not sold in UK for half a decade like Daihatsu or companies that registered literally 0 cars since 2014 and single numbers in years prior like Chevie (previously Daewoo) are on top of the reliability list because the single examples with extended warranty reliability report counted repairs against, didn't have as big of an impact vs for example tens of thousands VWs. Just the fact that Skoda is much more reliable than Volkswagen or Seat in the chart, while using the same components and tech, is an indicator of another flaw - is it more reliable because it fills market with more crapola base spec cars, is it because there are less claims or because less people seek extended warranty for something at bottom dollar? Or is the same flawed TSI engine with the same flawed DSG gearbox genuinely more reliable when put together by Czech vs Germans or Spanish?
I agree with a lot of the points you make here, but would argue that any reliability ranking is going to be flawed in some way(s) or other. From a statistical point of view, cars of which only a handful were included in the survey, should have been excluded. At the very least, Reliability Index should have made the absolute figures available.
But Daihatsu and Chevrolet don't invalidate the entire survey.
Once the number of cars is a couple of hundred then the averages start to be meaningful. The survey doesn't include tens of thousands of VWs, as the total sample is only fifty thousand. It doesn't need to either: such a large number of data points would be nice, but is unlikely to change the overall outcome much. If we assume that the survey contains 4,500 VWs (based on a market share of almost 9%) as against only 1,500 Renaults (based on a market share of just under 3%) then that's actually plenty for the comparative figures to be pretty robust. And VW, despite its carefully crafted image, does not come out on top.
Similarly, the averages for Honda, of which even fewer are likely to be included in the survey than Renaults (based on a market share of just above 2%) are still based on easily enough cars that their reputation for reliability seems justified.
You're right of course in suggesting that other factors come into play in deciding the ranking as well. Cars with fewer toys are indeed less likely to need repairs, all other things being equal. And since Skodas usually don't come nearly as fully loaded as Lexus, their similar position -- much higher than stablemates VW and Seat -- is probably due in part to that.
The argument about fewer people insuring cheaper cars doesn't stand, though, at least as long as enough are included in the survey for the calculation of averages to be meaningful. Also, why would people claim less for a Skoda than for a VW or an Audi, once they've taken out a warranty?
Another factor you haven't mentioned yet is the way in which these cars are driven. When is the last time a Fabia driver floored it to overtake you on a single carriageway with oncoming traffic? I suspect that the staid image of Skoda is doing them no harm in rankings such as this.
Finally, I wasn't trying to promote the Reliability Index specifically. Rather, I wanted to say that basing one's judgment on a small number of personal experiences is meaningless (
human, but meaningless). There are other surveys out there, such as the German TÜV reports summarised on
http://www.anusedcar.com/, the MOT pass rates reported on by Honest John, and so on. The point is that all these are based on large samples, and so despite their shortcomings, they are at least far more valid than a few personal observations and anecdotes.