Correlation

[cgendron]

So I have two columns of numbers, about 20 in column a and 20 in column b. Their correlation is .06 so very weak almost no relationship. Does anyone know a formula saying as my numbers in column A increase, what percent chance do I have that the numbers in column B increase? I know it would be an extremely low percentage.

 

[BiocideJ]

I don't know of any formulas, however, if the correlation is very low, I would suspect the probability of it moving in the same direction would trend towards 50% (not 0%). Basically the same as flipping a coin.

 

[cgendron]

Thanks. Im trying to see if the length of a biopsy determines the cancer positivity rate among a group of physicians. With my correlation of .06, basically there is no relationship with the data i'm using, so its saying, if the length gets longer there is almost no relationship that the positivity rate will increase. I was hoping I could find something that would use this correlation and say there is less than 1% chance(example) that the positivity would increase with the length increasing.