Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 0 down vote favorite
1

I have data counting the number of words that appear in text reviews, each review rated 1 to 5 stars. So for every word, I have the rating breakdown of reviews that contain that word.

I would like to perform a beta-binomial analysis on each word by rating (n=4) but I also want the rating categories to be evenly weighted (currently 5-star reviews are 59% of total reviews). If I proportionally weight the numbers (e.g. multiply by total number of reviews and divide by total number of reviews with that star rating), can I still say meaningful things about the data?

share|improve this question
It may help if you can clarify what you are trying to accomplish. I also have a few questions: Why is n=4 when you have a 5 rating scale? I am not sure how your weighting scheme works. What are numbers are you referring to when you say " If I proportionally weight the numbers.."? Also your weighting scheme seems a bit odd. Do you mean that you want to compute the weighted avg of the no of occurrences of a word with the weights being the number of 1, 2,.. star reviews? – user28 Aug 1 '10 at 15:30
Also (in addition to SV), can you please clarify why do you want to use beta-binomial analysis? The way I understand your question, you probably want to compare 1 .. 5 categories (right?) – suncoolsu Sep 21 '10 at 8:32
Flagged this to be closed - don't think this question can be rescued since it is not even clear what analysis the OP wants to do – Corone Feb 12 at 14:44

closed as not a real question by gung, Russell S. Pierce, cardinal, whuber♦ Feb 12 at 17:32

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, see the FAQ.

Browse other questions tagged or ask your own question.