PAE is intended to reflect that great, unmeasurable component of Survivor: a finalist's social game. We can only use it to evaluate finalists, because it measures the degree to which the jury votes a finalist receives reflects their physical (challenges) and strategic (voting people out) games. If the finalist receives significantly more jury votes than their prior performance would predict, then it seems likely they had a good social game. If they receive fewer (or no) jury votes, then their social game was probably poor. Straightforward, right?
Caveats: (1) Everything is context-dependent: one person's decent social game might look pitiful against a social master, yet brilliant against a belligerent, anti-social troll. (2) Also, if one of the finalists completely dominated every aspect of the game, AND was brilliant socially (Kim Spradlin), they will probably be undervalued here. But it's the best we can do (feel free to debate this assertion in the comments). Here's how we came up with the formula:
PAE uses the "No jury" score (ChW + wTCR). First, we calculate "expected" jury score (ExJ), which is the number of jurors (TotJ) times the ratio of the player's No jury score (NoJ) divided by the sum of all the NoJ scores of the finalists (SNoJ). Basically, in a jury of the size the player saw, they should receive ExJ votes. ExJ = TotJ * (NoJ/SNoJ).
Then, PAE is simply the actual number of votes received (JVF), minus ExJ, then divided by the number of jurors, or:
PAE = (JVF - ExJ)/ TotJ.
The final number reflects the percent of the jury votes received that did not reflect prior challenge or tribal council performance. Or "social game." For example, in Fiji, Earl, Cassandra, and Dreamz had ExJ scores of 3.2, 2.5, and 3.3, respectively (Earl didn't win any challenges). This projects as a virtual tie between Earl and Dreamz. Yet in reality, Earl swept the jury vote, 9-0-0, for a PAE score of 0.65, or 65%. A testament to Earl's ability to convince the jurors to vote for him.