Don't be a follower -- an analysis of internet shopping and crowd bias

Can you trust the hordes when it comes to product voting and rankings on websites such as Amazon?

Would you indeed trust a book review from some nonce on Imdb or BookCrossings (BC)? Are they all biased?

Vassilis Kostakos, of the Department of Mathematics & Engineering, University of Madeira
and Human Computer Interaction Institute at Carnegie Mellon University certainly thinks so in the paper Is the crowd’s wisdom biased? A quantitative assessment of three online communities

Forgetting the obvious argument that everything in this world is biased so of course voting would in turn should also be considered such, Vassilis sets out to discover whether you will get bang for your buck if you follow the “wisdom of the crowd”.

Vassilis' research aims to identify and remove said "wisdom" by designing a system that would give a more solid identification of whether a product is good or bad.

"This study shows that despite the large community size of each website, there exist significant biases in users’ voting and rating behavior," he writes.

He finds:

"That more than 50% of users on Amazon and BC cast only one vote. Similarly large portions of items receive a single vote, with the exception of Imdb where only 7% of items are rated only once.

"Furthermore, the experts (top 5% of voters) cast as few as 7 votes on Amazon, while on BC they cast 30, suggesting that BC consists of proportionately more expert users within the community.

"Even so, the top 5% of popular items receive as little as 10 votes on both sites.

"Conversely, Imdb’s popular items receive at least 662 votes."

Imdb, he explains, is simple -- any person can vote without leaving anything more than a tick or a cross. As long as you are a member you have the right to vote.

Conversely not only do you have to be a member of Amazon and BC but you MUST leave a review as well as a star rating.

That means you have to think about why you like/dislike the product to tell people.

Vassilis finds that in particular, BC expert voters "vote for mostly obscure titles with few votes" and "that many obscure items of small popularity are likely to be of interest to a few select experts, while the crowds -- made up of people who buy only few times -- are likely to be interested in “popular” items."

Check out the full report but be warned, basing your ideas on the thoughts of others will generally lead to disappointment.