Pearson Correlation problem

Question

Let's assume that we have 2 people and 5 objects with the following ratings (scale 1-5):

John has rated obj1:5, obj2:1 , obj3:(unknown value), obj4:2 , obj5:2 and Jack has rated obj1:1, obj2:5 , obj3:2, obj4:5 , obj5:5. Do i need obj3:2 when calculating the mean of Jack's ratings or just hypothetically delete it and use the other objects that have been rated by both of them? I want to calculate Pearson correlation between John and Jack. Also, i know that the obj3 will not be used in Pearson correlation function.

Thank you for your time.

Pere · Accepted Answer · 2017-09-29T15:51:58.187

1

Yes, we do need to know all of Jack's ratings to calculate their mean.

And we need John's rating for object 3 to calculate correlation between John and Jack's ratings.

However, if Jack's rating for object 3 were equal to the mean of all of other Jack's ratings (which it is not), then we could calculate Pearson's correlation without knowing John's rating for object 3. You can look at the formula for covariance to see the reason.

edited Sep 29 '17 at 15:51

answered Sep 22 '17 at 16:30

Pere

6,583

1

Yes, because the subtraction will be zero. Thank you very much! – George Gkekas Sep 22 '17 at 18:32

Pearson Correlation problem

1 Answers1