Does it make sense to calculate relative ratios of items from different scales when I need to normalize the result?

19 Views Asked by At

Let's say I have two different scales for size, where the first (A) is 0-10 and the other (B) is 0-1000. Then I will use each to measure (relative) sizes of products for which I know the number of flaws/defects. Now I get two sets of results:

  • A (sizes and product defects): 5 (2 defects), 8 (3 defects), 2 (1 defect)
  • B (sizes and product defects): 480 (2 defects), 600 (2 defects), 300 (1 defect)

I would like to express the average defect rate but because the scales are different, A would look terribly defective while B would look almost perfect if both were compared as No of defects / size unit.

I was wondering, would working with averages be a possible solution here? I mean:

  • For A, average size is 5. The ratio of individual sizes to the average would be 1, 0.63 and 2.5
  • For B, average size is 460, and ratios would be: 0.95, 0.76 and 1.53

Now, if I add the relative sizes, I get 4.1 for A and 3.3 for B. And finally, calculating the defect ratios:

  • 0.68 for A
  • 0.66 for B

This sounds right because they are virtually the same.

Is this a better approach than simply calculating Defect/ number of products? (so 6/3 and 5/3 in this example)? Thanks a lot