|The Mothership - 2017-01-27 |
The short version is, Trump won because you filled out the "which Star Wars character am I?" survey on Facebook and liked certain things, and then you got targeted advertising on Facebook.
Terrifying if you buy it; I am not yet convinced, as I am fundamentally skeptical of personality tests. Healthy debate here would be encouraged.
Lesson whether you buy it or not: stop using FB.
If anything, Trump won because CLINTON was relying heavily on that kind of data and it doesn't work for shit of course.
|somedongus - 2017-01-27 |
Fuck this guy, fuck Ted Cruz, and above all else, fuck the future.
|gmol - 2017-01-27 |
I'm not sure I buy the relative contributions of sophisticated statistics methods played to Trump's win vs. being a good 'ol white guy. I don't think anyone really really knows.
We conveniently attribute the same reasons to whoever won. Romney was a "data guy", Obama won because of "analytics", Clinton supposedly had google "geniuses" on her side this election. Dems were confident about their polling numbers on election night.
Only 10% of statistics is useful, the trick is figuring which 10%.
But there is a lot of money being made from the other 90%.
|GravidWithHate - 2017-01-27 |
Here's a website from the researcher who developed the model Trump is supposed to have used. I've tried pasting in some of my comments. It's always got my gender right, it's not great with my age, and I can't judge it's performance on my psychological profile.
What Trump is supposed to have done with this is really no different than anything anyone has ever done since people first noticed that different groups of people voted different ways. His campaign was just able to slice the pie finer and target their efforts more precisely.
I also think focusing on a narrow technical innovation of the Trump campaign is misguided. Because after Brexit, Trump, Matteo Renzi, with Marine Le Pen currently leading for the first round of voting, and AfD looking to head into parliament, if you're trying to explain Trump's victory by focusing on how Trump's campaign was planing ad buys, you're very much not focusing on the fact that there seems to be a massive crisis of confidence sweeping western democracies.
|Raggamuffin - 2017-01-27 |
All that data and all that work and it wasn't worth a dog's fart. Ted Cruz lost and so did Hillary. These people are trying to take statistics and apply it to individual people, and statistics just don't work that way. Think about how silly it is to develop a psychological profile of you based on your web browsing habits.
recommended reading: Weapons of Math Destruction by Cathy O'Neil
A major point being that many of these statistical models are actually just confirming our existing biases, but making them look like science.
I haven't read it yet (I ordered it at work a couple months ago but I haven't had the time to take it home yet), but the interview she did with Sam Seder is great:
Even as someone with a lot of skepticism about statistics in the social sciences (and the social sciences in general, to be quite frank) I was pretty surprised by the recidivism and crime rate stuff.
As somebody who worked in a university research group and crunched statistics, I can sadly confirm that way too much science is just people trying to prove the conclusions they've already arrived at. That doesn't mean science is bad--that's what peer review is for.
| Register or login To Post a Comment|