Search This Blog

Tuesday, September 4, 2012

Predicting Election Results

Please, someone tell me what is wrong with my nonexpert analysis of predictions of election results based on probability:

Probability cannot foretell the outcome of a specific event—say, any particular flip of a coin—but can only be assigned in percentages to a range of possibilities. Weather is more predictable than coin flips, human actions more complicated than horse races, but for any prediction on a specific event’s outcome based on probability, no outcome will or can show the prediction to have been "wrong." The person having made the prediction need not even acknowledge having left out relevant factors. Picture the careless shrug and casual statement to the effect that a “less likely outcome” prevailed. What I’m saying is that anyone can criticize a prediction for not taking everything relevant into account but that no one can ever say, regardless of outcome, that the prediction was “wrong.” 

Am I right that such a prediction cannot be wrong? If so, tell me again why we should give a rip what anyone predicts? If not, please explain.