Prev: tab names are wrong when printing custom footers (using [&tab] )
Next: average if - between range
From: Meenie on 2 Jun 2010 09:09 I have an audit and am looking for a specific item in the audit. I'm asking a specific question and the answer can be either Yes, No or N/A. To figure out the compliance to the question as a percentage, I'm taking the number of "Yes + N/A" answers and dividing it by the total number of items audited. Someone else thinks we should first subtract the number of n/a answers from the total number of items - this number she calls the "applicable total" then she divides the number of "Yes" answers by the "Applicable total" and gets a compliance. Example on one question there are 485 audits, 48 are "no", 69 are "Yes" and 366 are "n/a". By subtracting out the n/a answers then dividing by that number into the yes answers she gets a 58% compliance. By just taking the total of 485 and dividing that into the number of yes + n/a, you get 90%. I feel she is making the n/a answers count against her, and she shouldn't. I think you're looking through 485 items and only finding 48 that were "bad". Which is the correct way?? thanks, Meenie :) --
From: Steve Dunn on 2 Jun 2010 09:37 Only you can answer that, Grasshopper... Seriously though, what you and your colleagues need to ask yourselves is "what actually defines compliance with the question asked?". As an outsider, it would seem that the only answers which comply with the question are the ones who answered Yes, and therefore your compliance percentage would be based on Yes/Total, but it may be that you want to ignore those who answered "n/a", Yes/(Total-n/a), that really depends on the question. Personally, I can't imagine a case where (Yes+n/a)/Total would be correct, but again you need to first ignore the figures and define "compliance with the question" in words. <rant mode> This is the problem with most statistics gathering, the stat's are gathered, then the math used suits the desired outcome! </rant mode> "Meenie" <Meenie(a)discussions.microsoft.com> wrote in message news:FAD1C4EA-4965-42D2-BF58-ED17773785D3(a)microsoft.com... >I have an audit and am looking for a specific item in the audit. > I'm asking a specific question and the answer can be either Yes, No or > N/A. > To figure out the compliance to the question as a percentage, I'm taking > the > number of "Yes + N/A" answers and dividing it by the total number of items > audited. > Someone else thinks we should first subtract the number of n/a answers > from > the total number of items - this number she calls the "applicable total" > then > she divides the number of "Yes" answers by the "Applicable total" and gets > a > compliance. > > Example on one question there are 485 audits, 48 are "no", 69 are "Yes" > and > 366 are "n/a". By subtracting out the n/a answers then dividing by that > number into the yes answers she gets a 58% compliance. > > By just taking the total of 485 and dividing that into the number of yes + > n/a, you get 90%. > > I feel she is making the n/a answers count against her, and she shouldn't. > I think you're looking through 485 items and only finding 48 that were > "bad". > > Which is the correct way?? > > thanks, Meenie :) > > -- >
From: Meenie on 2 Jun 2010 09:46 "<rant mode> This is the problem with most statistics gathering, the stat's are gathered, then the math used suits the desired outcome! </rant mode> " I agree with that, and dislike stats intensely, lol, but ya gotta give 'em what they ask for, lol. -- Meenie "Meenie" wrote: > I have an audit and am looking for a specific item in the audit. > I'm asking a specific question and the answer can be either Yes, No or N/A. > To figure out the compliance to the question as a percentage, I'm taking the > number of "Yes + N/A" answers and dividing it by the total number of items > audited. > Someone else thinks we should first subtract the number of n/a answers from > the total number of items - this number she calls the "applicable total" then > she divides the number of "Yes" answers by the "Applicable total" and gets a > compliance. > > Example on one question there are 485 audits, 48 are "no", 69 are "Yes" and > 366 are "n/a". By subtracting out the n/a answers then dividing by that > number into the yes answers she gets a 58% compliance. > > By just taking the total of 485 and dividing that into the number of yes + > n/a, you get 90%. > > I feel she is making the n/a answers count against her, and she shouldn't. > I think you're looking through 485 items and only finding 48 that were "bad". > > Which is the correct way?? > > thanks, Meenie :) > > -- >
|
Pages: 1 Prev: tab names are wrong when printing custom footers (using [&tab] ) Next: average if - between range |