I have 3 scales in a survey form.
Each scales ranges from 1 to 10 which measure the severity, occurrence and identification ratings. This is for conducting Failure Mode effect Analysis by calculating Risk Priority Number.
Experts select one value for each of the three scales.
Basically : RPN = severity * occurrence * identification ratings
example: RPN = 6*7*2 = 84 and analysis is made.
As of now I have only 5 experts from the company who have filled the survey. I have seen some research papers taking 20 inputs for the same kind of questions and taking the average for each 3 categories and rounding off to nearest whole number.
Does the same technique apply to 5 responses scenario?
Or should I do that approach and obtain the final value and ask any different expert from the same company to make a final validation on the final values?
Hello Vishnu,
If what you are trying to do is capture a consensus rating from a set of five judges, you'd likely be better served using a median rating rather than a mean rating.
Would more judges be better? Quite possibly, especially if there was a lot of heterogeneity in the judgments given for a specific scale. Of course, a number of rater/judge sources of bias and error have been identified over the years, so that, too, is generally an argument for more rather than fewer. If there is little or no variation in judgments among your chosen judges, then you'd likely be more confident in the resultant answer.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey