Ranking crowdsourced data with curves(cederman.com)
cederman.com
Ranking crowdsourced data with curves
http://www.cederman.com/?p=116
2 comments
As someone who works (sort of) for Tripadvisor, I found this interesting. I'm a little suspect that "Since I changed tactics, I have been having great hotel experiences." is confirmation bias, though.
Maybe. I had a string of pretty crappy hotels, so started paying more attention to the reviews and noticed that although the hotels I had been staying at had higher averages, they had a lot less 5 star ratings. Next few hotels were much better, and when the option to get a hotel with a better curve wasn't available, I was appropriately disappointed. While I would like to think that I don't suffer from confirmation bias, this certainly isn't a double blind trial.
I have also seen a similar phenomenon with restaurants on Yelp though, which has been slightly easier to test in that I can try two similar restaurants with the different shaped curves.
I have also seen a similar phenomenon with restaurants on Yelp though, which has been slightly easier to test in that I can try two similar restaurants with the different shaped curves.
I tried a similar approach with the netflix prize, although in that case I fitted truncated normal distributions to the data.