Which test to use to understand if higher priced version of the beer is worthy of higher rating?

#1
Hello,

I am confused to know what test to be used in scenarios such as below. Below dataset can be checked on this link also (http://users.stat.ufl.edu/~winner/data/pricequal.dat)

I want to check if the higher price of the beer is justified as per the ratings provided (number of observations)
I thought of using spearman's rank correlation first, but then realised that it is not applicable as it will not be making use of observations column.

price, rating , observations
3 0 4
2 0 1
1 0 4
3 1 8
2 1 21
1 1 20
3 2 26
2 2 22
1 2 23
3 3 15
2 3 12
1 3 9
3 4 7
2 4 4
1 4 4

Price (1 to 3: 3 is Highest)
rating (0 to 4 : 4 is best rating)
Obs (number of people rated the combination -- e.g. first row -- 4 people gave 0 rating to the beer of highest price tag (price:3)
 

noetsi

Fortran must die
#4
I am not sure how one defines justified here. I don't think statistics can answer a value laden question like "is this justified."
 
#5
I am not sure how one defines justified here. I don't think statistics can answer a value laden question like "is this justified."
with some confidence intervals , you get answers from stats.. I am looking something of that sort. Hope, it makes sense.. :)
 
Last edited:
#6
If you weight each combination by frequency of observations then you can calculate Spearman's rho.

With kind regards

Karabiner
I converted price and rating on weighted scale, ranked them and found correlation which come to be 0.75. looks fine, not sure if p value can be calculated with that and if that is worth the calculation considering that p values should be calculated if sample size is higher?