Posted By:
GlennPresumably the percentage of "near mint" cards that are near mint is inversely proportional to the age. So if it's 20% of cards from the 50s, it might be 50% of the cards from the 80s and 5% of the cards from the 20s.
If there are any other statisticians here, what do you think of this? We can collect data from everyone on the forum for a few months and run a multiple regression with buyer's impression of 1) the card's grade from photo(s) only, 2) the age of the card, 3) the condition claimed by the seller, and 4) the seller's percent positive feedback as predictor variables, and with 5) the eventual grade as judged by the buyer upon receiving the card as the criterion. I'm guessing the four predictors would account for at least 90% of the variance.
I'd be happy to run the analyses, but I don't buy enough myself to get a decent sample size anytime soon. And in any case it wouldn't be a representative sample. If anyone wants to send me the data (i.e., the numerical values for each of the five variables mentioned above) for any cards they buy on ebay in the future, I can report the results here once I have a sufficiently large dataset.