Bugs the shit out of me that 538 still have the 10th place on their odds, while doing so much work to make the odds human brain understandable (favored to win, 67 out of 100, etc). Showing the whole 0 to 100 on their chart and some damn error bars wouldn't hurt either.
Yes, I know, it doesn't matter, but still bugs me.
Explicitly-reported confidence intervals, standard errors, and p-values for shit like Trump's approval rating or probability of winning would be extremely useful.
It wouldn't, but it would make me feel better, as a nerd and someone that took statistics in college. But the whole 538 business model depends on giving a false sense of accuracy. For an event with an n of one, for the layman, is there any real difference between 60%, 75% or 85%? Maybe as a campaign director it would help you to spread out resources and shit, but for a person checking the news, its like reporting the temperature to the 100th place (i.e. 68.34 degrees).
Bugs the shit out of me that 538 still have the 10th place on their odds, while doing so much work to make the odds human brain understandable (favored to win, 67 out of 100, etc). Showing the whole 0 to 100 on their chart and some damn error bars wouldn't hurt either.
Yes, I know, it doesn't matter, but still bugs me.
Explicitly-reported confidence intervals, standard errors, and p-values for shit like Trump's approval rating or probability of winning would be extremely useful.
It wouldn't, but it would make me feel better, as a nerd and someone that took statistics in college. But the whole 538 business model depends on giving a false sense of accuracy. For an event with an n of one, for the layman, is there any real difference between 60%, 75% or 85%? Maybe as a campaign director it would help you to spread out resources and shit, but for a person checking the news, its like reporting the temperature to the 100th place (i.e. 68.34 degrees).