Display:
That graph doesn't tell you anything.  You could set it up in any election anywhere.

I believe Nate or one of the other stats gurus of the blogosphere debunked it last night using the 2008 American election (releasing results in six waves, alphabetically).

Conservatives want live babies so they can raise them to be dead soldiers. - George Carlin

by Drew J Jones (myfriends@thisispancakes.com) on Sun Jun 14th, 2009 at 08:55:43 AM EST
[ Parent ]
That would be this
by det on Sun Jun 14th, 2009 at 10:37:45 AM EST
[ Parent ]
See my chart in a parallel comment.

The brainless should not be in banking. — Willem Buiter
by Migeru (migeru at eurotrib dot com) on Sun Jun 14th, 2009 at 01:50:32 PM EST
[ Parent ]
The debunking has been debated, because while it would make sense to get very close estimates if you got random samples added up at different times, in this case they brought in data from very differet nregions in lumps (and there are, or should have been, significant regional variations in the voting)

And there are th screen captures discussed here

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (etg@eurotrib.com) on Sun Jun 14th, 2009 at 12:55:23 PM EST
[ Parent ]
But what was plotted were not the lumps, but the cumulative totals. You cannot run a linear regression on the cumulative totals, but only on the lumps...

The brainless should not be in banking. — Willem Buiter
by Migeru (migeru at eurotrib dot com) on Sun Jun 14th, 2009 at 01:06:02 PM EST
[ Parent ]
If you plot the lumps, you get this:

Which is much more noisy than the cumulative totals.

It is well known you cannot apply linear regression to a time series, which is what this is, but only to the successive differences (the "lumps").

The brainless should not be in banking. — Willem Buiter

by Migeru (migeru at eurotrib dot com) on Sun Jun 14th, 2009 at 01:42:21 PM EST
[ Parent ]

Display:
Login
. Make a new account
. Reset password
Occasional Series