A Brief Guide for Matthew Yglesias, Perplexed as He Is About the Metaphysical Status of 538's "Forecasts"
Live from Data Journalism: Matthew Yglesias is perplexed:
Was anyone on the @FiveThirtyEight staff *really* prepared to place an even-money bet on Trump last weekend when the polls narrowed?— Matthew Yglesias (@mattyglesias) August 3, 2016
I don’t really understand the metaphysical status of a forecast that has this much volatility. pic.twitter.com/XK71ts9YwU— Matthew Yglesias (@mattyglesias) August 3, 2016
Matt is correct to be skeptical. The http://fivethirtyeight.com "polls-only" "forecast" is not a forecast. It:
- Takes recent past polls to estimate a current state of the race...
- Estimates an uncertainty of our knowledge of the current state...
- Projects that that current state of the race will drift away from its current state in some Brownian-motion like process...
- Calculates the chance that that process--starting from the estimated present--would produce 270 electoral votes for one candidate or the other.
It has no metaphysical status. Or, rather, it's metaphysical status is that it is just a number, rather than a "probability" of anything--other than that this stochastic process (which does not describe anything in the real world) starting at this point with this uncertainty with these parameters evolves in this way.
The real 538 forecast is the "polls-plus" forecast, which knows, say, that the candidates usually get convention bounces and that it is the week between the Republican and the Democratic conventions. It is much more sensible:
Nate and company might be willing to make a market (with appropriate spreads) on their "polls-plus" forecast.
But probably not. Even the 538 "polls-plus" forecast does not "know" that Trump is an intrinsically disordered personality, narcissistic and quite possibly moderately demented...
Cf: Sam Wang: http://election.princeton.edu/history-of-meta-analysis/