1

Is there a statistical test I can apply to compare multiple rates and get which one is the biggest and which one is the smallest?

Example:

1 person converted out of 2 for page A. conversion 1/2
10 person converted out of 1000 for page B. conversion 10/1000
600 person converted out of 800 for page B. conversion 600/800
1000 person converted out of 3000 for page B. conversion 1000/3000

From does N pages I want to know, which one is statistically bigger and which one is statistically smaller.

mdewey
  • 16,541
  • 22
  • 30
  • 57
Nessim Btesh
  • 11
  • 1

2 Answers2

1

I suggest you plot confidence intervals. Your values represent binomial proportions or, equivalently, your results are from Bernoulli trials where the independent outcomes are either a success or a failure. Thus the confidence interval type that you are after is the confidence interval for binomial proportions.

There are many ways to calculate the relevant confidence intervals because none of them is completely correct. See the following questions and answers from this site for details.

Confidence interval for Bernoulli sampling

Confidence interval around binomial estimate of 0 or 1

Community
  • 1
Michael Lew
  • 10,995
  • 2
  • 29
  • 47
0

It sounds like you have two variables and one outcome, and these three are observed 1000 times. In that case, regress your outcome (pages?) on the two variables and see which number is biggest. You can use a standard test procedure to test one coefficient bigger than the other.

Superpronker
  • 722
  • 5
  • 6