It isn't just "the Left" that indulges in the logical fallacy that "correlation = cause."And it isn't a problem that's confined to sociological topics.
What I'm noticing is that the era of Big Data is overawing people- including professional researchers who should know better. It's so easy to do data correlations these days, often generat…
It isn't just "the Left" that indulges in the logical fallacy that "correlation = cause."And it isn't a problem that's confined to sociological topics.
What I'm noticing is that the era of Big Data is overawing people- including professional researchers who should know better. It's so easy to do data correlations these days, often generated from huge data sets. This is done in some fields of science and medicine as much as in any academic social discipline or humanities field. One of the newest data fads is a heightened focus on "regression analysis and causative inference"- generating conclusions based on counterfactual hypotheses. The tool doesn't entirely lack value, but it's inherently limited. It can provide useful guidelines to follow a fruitful avenue of research, but it isn't probative. Yet it's increasingly being treated as if it was- or, at any rate, its findings are often assigned much more value than they possess.
There's also an increasing amount of swooning over "data forwarding"- drawing conclusions about future trends projected from assorted multivariate vectors that already exist in data sets. (Numbers! Big ones!) I'm noticing a lot of this in regard to projections of economic behavior, and also the "public mind" on other questions, like political questions. The researchers and readers of those findings need to increase their wariness, not their credulity. . Correlations drawn from a whole lot of numbers and a whole lot of factors crunched together do not add up to Prescience, much less Omniscience.
It isn't just "the Left" that indulges in the logical fallacy that "correlation = cause."And it isn't a problem that's confined to sociological topics.
What I'm noticing is that the era of Big Data is overawing people- including professional researchers who should know better. It's so easy to do data correlations these days, often generated from huge data sets. This is done in some fields of science and medicine as much as in any academic social discipline or humanities field. One of the newest data fads is a heightened focus on "regression analysis and causative inference"- generating conclusions based on counterfactual hypotheses. The tool doesn't entirely lack value, but it's inherently limited. It can provide useful guidelines to follow a fruitful avenue of research, but it isn't probative. Yet it's increasingly being treated as if it was- or, at any rate, its findings are often assigned much more value than they possess.
There's also an increasing amount of swooning over "data forwarding"- drawing conclusions about future trends projected from assorted multivariate vectors that already exist in data sets. (Numbers! Big ones!) I'm noticing a lot of this in regard to projections of economic behavior, and also the "public mind" on other questions, like political questions. The researchers and readers of those findings need to increase their wariness, not their credulity. . Correlations drawn from a whole lot of numbers and a whole lot of factors crunched together do not add up to Prescience, much less Omniscience.