Jump to content
270soft Forum

Nate Silver guarantees there is a 10% chance Trump wins!!!


PoliticalPundit
 Share

Recommended Posts

Sure, they're different, but they're not radically different (they're all elections, not radically different types of events being modeled) and my understanding is that the models are very similar.

And yes, each state is it's own election and can (and *should*, given the EC) be treated as separate.

Link to comment
Share on other sites

4 minutes ago, Defiant said:

Sure, they're different, but they're not radically different (they're all elections, not radically different types of events being modeled) and my understanding is that the models are very similar.

And yes, each state is it's own election and can (and *should*, given the EC) be treated as separate.

Do you really believe that developing a House elections model will translate to a Presidential election model?

Your 2nd point is a bit different, because Silver's election forecast depends on those 51 'elections' (not sure that's the right term, but let's go with it). According to this, we have 51 trials for his 2016 Presidential election model. Was the model shown to be correct or incorrect, given those 51 trials? 

Link to comment
Share on other sites

33 minutes ago, admin_270 said:

Do you really believe that developing a House elections model will translate to a Presidential election model?

Your 2nd point is a bit different, because Silver's election forecast depends on those 51 'elections' (not sure that's the right term, but let's go with it). According to this, we have 51 trials for his 2016 Presidential election model. Was the model shown to be correct or incorrect, given those 51 trials? 

Directly, no.  But fairly similarly?  Yeah. It's election vs election.  It's not election vs stock market or election vs weather or election vs die rolling.  And, as far as I can tell, the models are very similar.

 

First off, it's an estimate - as such, I wouldn't rate it as a Boolean, but see just how good it is.  Similarly, my 66% from earlier was just an approximation.   It's not an exact, but it's a very good approximation.

 

Second, I don't think that 51 is sufficient by itself  (I would include the house and senate races and also run the model on prior elections, too).  But from a basic look of what is far too few die throws,  while there's clearly room for improvement, but it didn't perform that horribly, either. Of the 10 states they gave high 70s/low 80s probabilities to,  7 times they were right, and 3 times, the underdog won.

(I'd also say that the model relies on polling, and the polling clearly didn't capture the shift from the Comey letter - the model is unlikely to capture something that's not captured by the polling.)

Link to comment
Share on other sites

3 minutes ago, Defiant said:

but it didn't perform that badly, either. Of the 10 states they gave high 70s/low 80s probabilities to,  7 times they were right, and 3 times, the underdog won.

Thanks for this - I might actually look state by state tomorrow and see how they did in 2016.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...