originally posted by Jayson Cohen:
originally posted by VLM:
Over time, individuals provide their own anchors when using scales thus making their individual use of them consistent.
If you're worried about the multivariate nature of wine, consider my score the first principal component. There are always levels of analysis and generalization. We use them in natural language and the language of thought. To pretend that they are useless misses the point, I think.
Nathan, not giving you a hard time just for the sake of it: I have a hard time understanding how these two thoughts are internally consistent. The first is a purely subjective viewpoint about the value of your points. The second attempts to impose some sort of objectivity to your points in isolation. I think the usefulness to others lies elsewhere (if at all), whether we attempt to couch that usefulness in statistical terms or linguistic terms.
For the first, it has been empirically shown that people will create their own anchors and range and use a scale consistently. That's really all.
The first principal component is the vector that describes the most variance in the data, by definition. The analogy was to point out that the multivariate nature of the question has been taken into account in the score, and one can think of that score as the first principal component if it helps. Of course, this will miss small nuances that are idiosyncratic to one dimension or the other; thus, it is a level of generalization. The last point was that we use and accept levels of generalization all the time, e.g. we use them in natural language, language of thought (heuristics), empirical science, etc. We do this because people are rarely able to think in three dimensions let alone k.
I find the complaint that art/wine/literature are complex, multivariate and thus irreducible to be intellectually lazy. Of course there is error and nothing explains anything perfectly, we don't really know the exact nature of protection of some vaccines, does that matter at the level of public health outcome? The consistency, accuracy and precision of the generalizations make them more or less useful.
As to objectivity, I am of the belief that real things that exist in the universe have a causal relationship to my perception of them. While we have different perceptions of things here or there, we all have very similar (a tight distribution) of physical tools and share a common language for communicating those things. Thus, I think that my communication of those things to you (or just thinking to myself) carries some objective truth encapsulated within it. I'm just a simple follower of Hume's empiricism.
Where I would go further, is that if you put those of us who have been drinking wines together for the last 20 years and we drank and rated a set of wines together, we could use the scores to come up with some objective idea of the relative and absolute quality of the wines. There are some caveats there, but I don't understand why this idea is so outrageous to people.
Principal components are standardized and can be transformed to have any mean and SD that you want. Theoretically, there is no upper or lower bound.
