- #1
ohwilleke
Gold Member
- 2,366
- 1,358
It isn't every day that a musician and political blogger with no mathematical experience develops a new statistical technique to analyze error in exit poll data. But, Daily Kos blogger Feeble has done just that.
Here work is described here: http://www.mysterypollster.com/main/2005/04/the_liddle_mode.html#more
and has become the basis of an academic journal article in the field.
Feeble's Fancy Function is used to remove the noise from precinct by precinct comparisons of exit poll data and actual results, because it turns out that facts like the actual partisan makeup of a precinct can impact the amount of error that is expected in a precinct. Thus, when precincts vary greatly in partisan leaning, random error can look a lot like partisan bias without the proper analysis.
In this case 2004 election results which a first glance seemed to show massive fraud, actually could quite possible have been the result of random sampling error, although no one will ever know for sure.
Here work is described here: http://www.mysterypollster.com/main/2005/04/the_liddle_mode.html#more
and has become the basis of an academic journal article in the field.
Feeble's Fancy Function is used to remove the noise from precinct by precinct comparisons of exit poll data and actual results, because it turns out that facts like the actual partisan makeup of a precinct can impact the amount of error that is expected in a precinct. Thus, when precincts vary greatly in partisan leaning, random error can look a lot like partisan bias without the proper analysis.
In this case 2004 election results which a first glance seemed to show massive fraud, actually could quite possible have been the result of random sampling error, although no one will ever know for sure.