What's the best method to see if there's a strong correlation between two types of ratings on a product level (product $1, 2, \dots, n$) where each product has ratings of type '$a$' and '$b$'?

For the paired averages of ratings (of type $a$ and $b$) on a product level, would it be biasing my results if I filter out products with small number of ratings per type of rating, or low standard error in either average of ratings, or a combination of both?

Which would be best: Pearson's or Spearman's?

What's the most valid way to observe a potential existence of a relationship?

Also, would said method allow for more explanatory factor variables, like product type?

The overall goal is to see if there is a correlated (or non-random) relationship between the two averages on a product level (since they're what's surfaced on the marketplace).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.