Formula for finding a % less than MAX

[tlapointe1970]

I have 10,000 numbers down a column. I have calculated the Min, Max, average, median, etc, but I need to find - not the MAX number - but the 95% max number.... meaning I want to eliminate the oddball top 5%. I can't just do a standard deviation because of the way the numbers are calculated on the low end so it wouldn't be an accurate depiction on the high end. Any ideas?

Thanks.

this is an idea of the number structure. I don't want to include the $2.00 so I can't do a stdev of that column (or can I eliminate the $2 when I am calculating STdev?) but I need an idea of a 95% confidence interval of the numbers over $2 - Thanks a ton

8.00
8.22
14.23
2.00
2.00
8.02
11.23
22.72
12.24
11.67
2.00
8.98
9.25
2.00
13.23
28.60
2.00
12.65
10.23
2.00
9.50
9.63
14.55
2.00
2.00
8.36

[darkyam]

Question: do you want to both exclude all the instances of $2 AND cut off the top 5%? If so, which is considered first? In other words, if you have 10,000 numbers, and 1,000 of them are $2, does the top 95% mean 8500 (10,000-.05*10000-1000) or 8550 (10000-1000-.05*9000)?

[tlapointe1970]

I am used to the normal bell curve distribution, which this is not .... for my purposes, I would say reduce the number by all the $2.00 - and then take off the top 5% - then give me that highest number....

make sense?

[day92]

Are you looking for something like this?