Polls said Trump was popular, but Herman Cain was also popular. Difference was changed dynamics. Smart "pundits" caught that, models cannot.
-
-
Replying to @zeynep
I'm a firm believer in empirically-grounded analyses but the real art & science is knowing what kind of data supports what kind of argument.
3 replies 10 retweets 10 likes -
Replying to @zeynep
We needed more smart, empirical pundits to make sense of 2016. Not horse-race coverage, through data models or "source close to" narratives.
2 replies 5 retweets 10 likes -
Replying to @zeynep
Once again, proof how it can be done.
@NormOrnstein from Aug. 2015, explaining clearly why the models may not work. http://www.theatlantic.com/politics/archive/2015/08/maybe-this-time-really-is-different/401900/ …1 reply 3 retweets 4 likes -
Replying to @zeynep
For low frequency events like presidential elex, "models" are another version of conventional wisdom, condemned to miss changing dynamics.
3 replies 10 retweets 10 likes -
Replying to @zeynep
I had explained why
@NateSilver538 got#Bra vs#Ger so wrong. Same reasons are why models got 2016 nomination wrong.https://medium.com/message/learning-from-natesilver538s-omg-wrong-bra-vs-ger-prediction-9d47fa0611e#.mve8k2903 …2 replies 4 retweets 11 likes -
Replying to @zeynep
Best empirical method is triangulation—requires analyzing strengths of different data & continuity of dynamics, i.e data AND area expertise.
2 replies 5 retweets 11 likes -
Replying to @zeynep
The map is not the territory but they're connected. But how? Therein lies the mastery which will evade those who dismiss polls or "pundits".
1 reply 4 retweets 5 likes -
Replying to @zeynep
Here's what I wrote after
@NateSilver538 missed#BravsGre. *Exact* issue in 2016 nomination. https://medium.com/message/learning-from-natesilver538s-omg-wrong-bra-vs-ger-prediction-9d47fa0611e …pic.twitter.com/r5BtjPQL5K
2 replies 4 retweets 10 likes -
Replying to @zeynep
More from me two years ago on 538. Same in 2016. Field changed while data looked backwards. https://medium.com/message/learning-from-natesilver538s-omg-wrong-bra-vs-ger-prediction-9d47fa0611e …pic.twitter.com/XEGWg6zsBp
1 reply 3 retweets 5 likes
Finally: this is NOT about avoiding data analytics. Question is how to do it better, through multi-method, multi-data triangulation. /end
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.