0
$\begingroup$

I have a ranking of books that is based on number of sales and I want to improve it to include how many countries the book was sold in and number of libraries that carries it

Book rating 1 = # sales 

Book rating 2 = # sales + # countries + # libraries

Can I say that Book rating 2 is "better" than book rating 1 ? What do "better" mean here? more comprehensive and more than one dimension?

If I have a list of books that were ranked using these 2 equations , can I compare the 2 ratings to see if there a difference between them using t-test for example? or there is different test to use ?

Example - equation1 - equation2
Book 1  - 5000 -  5050
Book 2  - 300  - 320 
Book 3  - 90   - 99

Finally, forming a new equation and comparing it to existing ones is called modeling ?

Many thanks

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

I don't think that the second rating is very meaningful. Sales are measured in thousands to millions (probably), countries, in dozens, and libraries, in thousands as well. So countries won't contribute much.

What I would do here is to produce the percentile at which a given book is sitting, and add these. So if "Collected Writings" by Mao Zedong are sold in 1bn copies and is the top selling book in terms of sales and libraries, but all copies are sold in only one country (bottom 1st percentile), then its score would be 100+1+100=201, and it would lose to "Harry Potter" that would have been sold in say 100 mln copies (ranking at 98th percentile in sales) to 120 countries (also 98th percentile) to 20,000 libraries (60th percentile), for a total score of 98+98+60=256. The books would be directly comparable to one another on a scale like this; t-test is applicable to the groups of books (as in, "are military memoirs more popular than fantasy?")

I would not call this modeling... which would involve accounting for the year of publication, original language, genre, and a bunch of other factors. For a check of validity, I would try to correlate the above scores with Amazon rating. Who knows what that rating is actually measuring, but more popular books would be ranked higher.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thanks Statsk! So After using the % can I compare the 2 ranking? if yes, then using which test? And when using the other factors you mentioned, would is be called modeling? To check for validity with Amazon , what would be the appropriate test to use? $\endgroup$
    – tnaser
    Commented Jan 7, 2013 at 17:48
  • $\begingroup$ By comparing with Amazon, if there is correlation then what would be the benefit of the equation? if there is no correlation then does that mean it is not valid or just another way of ranking? Many thanks $\endgroup$
    – tnaser
    Commented Jan 7, 2013 at 18:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.