3 ms·

It's like those heatmaps that geographically display prevalence of human-contagious diseases, and they exactly maatch heatmaps of population density.

WoahNoun 3y ago
As always, XKCD is on the case: https://xkcd.com/1138/ https://xkcd.com/1138/
- tiffanyg 3y ago
  Yes, I too have been waiting for the day when we have not just "Martha Stewart Living", but "Martha Stewart Yiffing". I always thought she'd make a good 'Tanuki'. ;)
tiffanyg 3y ago
I.e. "base-rate neglect" or "fallacy", etc. The concept, and examples, of "base rate" are pretty intuitive, IMO. And yet, somehow, considering how INCREDIBLY CRUCIAL even the most basic understanding of "base rate" is to "critical thinking", they seem completely neglected in the vast majority of education. I'd argue, now, that we'd be far better off skipping crap like "integral rules for the cotangent function" (or whatever the current equivalent is) than skipping serious discussion and repeated allusions to the ubiquitous theme of "base rates" (and even "probability" and "statistics" in general - just having SOME feel for these) in interpreting, well, just about any ... putative "information". Further, I suspect that if we introduced kids to this kind of stuff when they were even 5, or 6, they'd pick it up far more easily than most adults, who have to wedge this often "unwelcome news" into their already rather rigid / solidified worldviews ... possibly confronting how 'wrong' of a perspective they've brought to many chains of reasoning etc... leading, often enough, to rejection of the unwelcome news / much more difficulty understanding it at all (often enough probably driven just by the 'subconscious' interfering with quite "getting it" ... but not in a way almost anyone can realize ... [i.e., why the hell they can't "get it" ... why some concept that seems simple enough is actually so slippery and like 'quicksand' in their minds ... something that might look clear enough 'on paper']).
tiffanyg 3y ago
This was still a bit on my mind, so came back to amend / improve on prior comment (though, seems not much interest in it, still wanted to better fill it out): There are two "classic" issues in these sorts of cases, in terms of statistics. One is the "base rate" I mentioned before, the other is "conditional probabilities". I.e., Bayes' rule (or theorem) etc. These are intimately intertwined, generally. The issue is: if you're trying to make a (statistical) inference, it, much like sentences in writing (among other things), needs to be appropriately "contextualized". So, for example, you might be able to "find something" ... "interesting" ... (or infer / "say" something) in the XKCD example mentioned by another commenter, if you were looking at ratios - for example, (number of "Martha Stewart Living" subscribers) / (number of people). Then, at least, you have something other than just a sample straight from the "base heatmap" (population). Now, if you want to look at those subscribers, and consumers of "furry porn" (or whatever - just following the XKCD set up, here), then, you want to look at things like P(a | b) where a = "subscriber to Martha Stewart Living" and b = "consumer of furry porn" (and, P(a | b) means "the usual" - probability of "subscriber to Martha Stewart Living" GIVEN "consumer of furry porn"). Bayes' rule, and other "transformations" available (i.e., that are valid - that are "Theorems"), help you to find ways to compute such statistics (importantly, under CERTAIN assumptions [specified in the statements of the theorems, underlying propositions, etc. - and generally explained in stats textbooks]). Finally, and this probably prompted returning to write this comment more even than what I just wrote - you don't HAVE to be "formally trained" at all to 'get' this kind of stuff, of course (though most people end up with some kind of 'muddle' / fuzzy thinking without). One of my favorite examples of someone more clever than me (IMO - who seems to have gotten it without "formal" training / consideration / exploration / ...) is Bill Burr: https://youtu.be/ylHisHlMWQQ https://youtu.be/ylHisHlMWQQ

WoahNoun 3y ago

As always, XKCD is on the case: https://xkcd.com/1138/ https://xkcd.com/1138/

tiffanyg 3y ago
Yes, I too have been waiting for the day when we have not just "Martha Stewart Living", but "Martha Stewart Yiffing". I always thought she'd make a good 'Tanuki'. ;)

tiffanyg 3y ago

I.e. "base-rate neglect" or "fallacy", etc. The concept, and examples, of "base rate" are pretty intuitive, IMO. And yet, somehow, considering how INCREDIBLY CRUCIAL even the most basic understanding of "base rate" is to "critical thinking", they seem completely neglected in the vast majority of education. I'd argue, now, that we'd be far better off skipping crap like "integral rules for the cotangent function" (or whatever the current equivalent is) than skipping serious discussion and repeated allusions to the ubiquitous theme of "base rates" (and even "probability" and "statistics" in general - just having SOME feel for these) in interpreting, well, just about any ... putative "information". Further, I suspect that if we introduced kids to this kind of stuff when they were even 5, or 6, they'd pick it up far more easily than most adults, who have to wedge this often "unwelcome news" into their already rather rigid / solidified worldviews ... possibly confronting how 'wrong' of a perspective they've brought to many chains of reasoning etc... leading, often enough, to rejection of the unwelcome news / much more difficulty understanding it at all (often enough probably driven just by the 'subconscious' interfering with quite "getting it" ... but not in a way almost anyone can realize ... [i.e., why the hell they can't "get it" ... why some concept that seems simple enough is actually so slippery and like 'quicksand' in their minds ... something that might look clear enough 'on paper']).

tiffanyg 3y ago

This was still a bit on my mind, so came back to amend / improve on prior comment (though, seems not much interest in it, still wanted to better fill it out): There are two "classic" issues in these sorts of cases, in terms of statistics. One is the "base rate" I mentioned before, the other is "conditional probabilities". I.e., Bayes' rule (or theorem) etc. These are intimately intertwined, generally. The issue is: if you're trying to make a (statistical) inference, it, much like sentences in writing (among other things), needs to be appropriately "contextualized". So, for example, you might be able to "find something" ... "interesting" ... (or infer / "say" something) in the XKCD example mentioned by another commenter, if you were looking at ratios - for example, (number of "Martha Stewart Living" subscribers) / (number of people). Then, at least, you have something other than just a sample straight from the "base heatmap" (population). Now, if you want to look at those subscribers, and consumers of "furry porn" (or whatever - just following the XKCD set up, here), then, you want to look at things like P(a | b) where a = "subscriber to Martha Stewart Living" and b = "consumer of furry porn" (and, P(a | b) means "the usual" - probability of "subscriber to Martha Stewart Living" GIVEN "consumer of furry porn"). Bayes' rule, and other "transformations" available (i.e., that are valid - that are "Theorems"), help you to find ways to compute such statistics (importantly, under CERTAIN assumptions [specified in the statements of the theorems, underlying propositions, etc. - and generally explained in stats textbooks]). Finally, and this probably prompted returning to write this comment more even than what I just wrote - you don't HAVE to be "formally trained" at all to 'get' this kind of stuff, of course (though most people end up with some kind of 'muddle' / fuzzy thinking without). One of my favorite examples of someone more clever than me (IMO - who seems to have gotten it without "formal" training / consideration / exploration / ...) is Bill Burr: https://youtu.be/ylHisHlMWQQ https://youtu.be/ylHisHlMWQQ