Flavour vs Score
It is often said that there is no “better” flavour class of whisky than any other. If you prefer delicate Lowland whiskies, they are just as “good” as heavily-peated Islay malts, or “sherry bomb” aged Speysides, etc.
Again, so it is frequently said (and so I truly believe). But if this is true, then it is a truth that is more honored in the breach than in the observance among most whisky reviewers, as Shakespeare might say. 😉
To understand what I mean by that, let us consider for a moment how most of the variance in the flavour clusters could be accounted for in the PCA by just two scales – winey-smoky and delicate-rich. These are an excellent starting point to see if we can spot common biases in reviewer scores.
Key Flavour Scale: Winey vs Smokey
I will start with the winey and smoky scales separately, since there was an obvious second-order pattern in that data (i.e., the two arms of the V-pattern). Do reviewers scores correlate with “winey-ness” or with “smoky-ness”?
The smoke story is not entirely surprising, since it is quickly apparent from reading reviews that reviewers who like smokey flavours (thanks mainly to the use of peat in the malting) typically prefer more rather than less in a given whisky. This is likely why heavily-peated Islay malts tend to score higher than their somewhat less peaty Highland counterparts.
The lack of correlation for winey features suggest that there is less obvious bias in this direction. So much for all that sherry cask finishing. 😉
Delicate to Rich Scale
But what about the continuous delicate-to-rich scale?
Now that seems pretty clear to me – reviewers definitely seem to give higher scores preferentially to more “complex” whiskies (i.e., those with “richer” flavours). This pattern is not surprising – as you develop a taste for whisky, it stands to reason that you would seek out (and prefer) those with the greatest range and diversity of flavour over time.
What this means is that we have to consider whisky scores in relation to where the whiskies fall on the delicate-rich gradient (and to a lesser extent, on the neutral-to-smoky gradient). The easiest way to do this would be to simply rank whiskies within each flavour cluster. That way, it wouldn’t matter what the arbitrary range of scores were – only how similar whiskies relate to each other. That said, many of the clusters span a pretty good range of the delicate-rich continuum, so there is still going to be a confound here.
But there’s another reason why we can’t simply consider rank order. We also need to consider what the degree of variation between reviewers represents – as discussed on my next page.