He should write a script to harvest a subsection of Amazon.
Then he should parse the results and see how closely correlated the Vine-to-Bad-Review there is.
.
A couple of data points here or there are not very convincing, but if he sampled say 10,000 products he could say:
* How many have reviews from Vine?
* Of those Vine reviews, what is the distribution of votes?
.
He should write a script to harvest a subsection of Amazon. Then he should parse the results and see how closely correlated the Vine-to-Bad-Review there is.
.
A couple of data points here or there are not very convincing, but if he sampled say 10,000 products he could say:
* How many have reviews from Vine?
* Of those Vine reviews, what is the distribution of votes?
* Of non Vine reviews, what is the distribution?