Maybe you should write a book for non-scientists on how to navigate criticism and debate. It's clearly easy for skeptically-minded people to uncritically embrace bad critiques.
Psudeocritics make the jobs of the TRUE heroes (people who make fun of professionals for sharing horrible graphs on Twitter) so much harder. How am I supposed to make fun of the dude who made a market concentration v inflation graph where there were so many industries at 100% that the 100% mark was just a contiguous blue line when other people are posting dumb critiques? Think of the good-faith trolls you monsters!
One interesting exercise is to figure out how many datapoints we can add to / remove from the chart for the weak correlation between variables to be zero or in the opposite direction.
A small correlation might be very meaningful in a situation where we expect year-over-year stability. People have fewer children now than they used to. If you did a scatterplot of the fifty states over a ten year period it wouldn't look like much change at all. Yet we know that over time - say, over fifty years - it's a very big deal indeed. When even small changes have large consequences, it adds up.
the funny thing is that the change in cooperation from start to finish was large enough relative to the standard error that you can break it into two non-overlapping periods in which you can say with over 95% confidence that an increase occurred within each period.
"it was quote-tweeted by one of the biggest pseudocritics of them all, N.N. Taleb"
I have very little knowledge of statistics and have been convinced by Taleb's tweets in the past. Could you explain (or maybe even write a new post!) what you disagree with Taleb about?
Maybe you should write a book for non-scientists on how to navigate criticism and debate. It's clearly easy for skeptically-minded people to uncritically embrace bad critiques.
Psudeocritics make the jobs of the TRUE heroes (people who make fun of professionals for sharing horrible graphs on Twitter) so much harder. How am I supposed to make fun of the dude who made a market concentration v inflation graph where there were so many industries at 100% that the 100% mark was just a contiguous blue line when other people are posting dumb critiques? Think of the good-faith trolls you monsters!
One interesting exercise is to figure out how many datapoints we can add to / remove from the chart for the weak correlation between variables to be zero or in the opposite direction.
A small correlation might be very meaningful in a situation where we expect year-over-year stability. People have fewer children now than they used to. If you did a scatterplot of the fifty states over a ten year period it wouldn't look like much change at all. Yet we know that over time - say, over fifty years - it's a very big deal indeed. When even small changes have large consequences, it adds up.
the funny thing is that the change in cooperation from start to finish was large enough relative to the standard error that you can break it into two non-overlapping periods in which you can say with over 95% confidence that an increase occurred within each period.
"it was quote-tweeted by one of the biggest pseudocritics of them all, N.N. Taleb"
I have very little knowledge of statistics and have been convinced by Taleb's tweets in the past. Could you explain (or maybe even write a new post!) what you disagree with Taleb about?