Reliability and Confidence and Excel

  • Thread starter Thread starter Guest
  • Start date Start date
G

Guest

Can excel calculate one / two-sided tolerance limit tables so I can enter a
reliability and confidence level and sample size and get the associated K
factor?
 
Gerigto;

Absolutely. Excel can calculate anything you want it to. Unfortunately I
have no clue what you are describing. Can you break out your data and give
us an idea as to what it is you are trying to calculate?

God Bless

Frank Pytel
 
Hi Frank,

Gerigto actually posted my question for me. I'm trying to see if Excel has
a function that will yield k factors for one-tail and two-tail tolerance
limits for normal distributions for any Reliability and Confidence levels I
input. Statistical books publish these tables, but only for certain R/C
combinations. I'd have to scan and send a copy of one of these tables to you
if you needed to see them.
 
Wiseguy91;

Hey. I hope you had a pleasant labor day. To the point.

I am an idiot. The research that I did into your question confirms that I am
aboslutely an idiot when it comes to statistics, which is where most of the
references sent me to when I Googled your question. There may in fact be a
function. When I searched normal distribution in the excel function search
dialog I came up with

normsdist
normzdist
norminv
normsinv
kurt
ztest

=ztest() seems to be the closest to your question. But it returns P not K
(whatever that is).

There are many very smart people here that may know exactly what your
speaking of and can guide you there. If it does not exist, you might try
reframing your question. I developed an application that can be found here,
labeled Shareware One Point.xls.

http://groups.google.com/group/excel-applications-and-spreadsheet-programming?lnk=iggc

I would take a guess at most folks not knowing what a Compaction Proctor
Value is. I did not until recently. But I do know what a curve is and I
understand how to plot this on a chart. I further know how to get feedback
from the chart that is created so I can get the values I want.

My point. Try posting the request in normal, for me the key word is idiot,
english. This will make it easier for more people to direct you to what you
want, and will give me a chance to learn something too.

God Bless

Frank Pytel
 
Wiseguy;

An after thought. You might try posting to the programming or charting
sections. These areas are more likely to be viewed by folks that are a whole
lot smarter than me.

God Bless

Frank Pytel
 
For what distribution? (Distribution-free tolerance limits do not use a
k-factor.)

One-sided tolerance factors for the normal distribution are calculated using
the inverse of the noncentral t distribution
http://www1.fpl.fs.fed.us/toltheory.pdf
Excel does not natively calculate the noncentral t distribution, but Ian
Smith has provided a library of VBA routines that do
http://members.aol.com/iandjmsmith/Examples.xls

For 2-sided tolerance bounds, I am not aware of publically available
software to do the exact calculations in Excel, but it could be programmed.
You first need to decide what kind of 2-sided tolerance limits you want. Do
you just want to assure coverage of a specified proportion of the population
(control the central coverage of the interval to be at least the specified
percentage), or do you want the larger k-factor that assures that you have
covered the central specified percentage of the population (control each tail
to be no more than half of 1-specfied percentage). The first type may be
more common, but the second is more appropriate if you are concerned about
assuring a minimum process capability index.

Odeh and Owen's book "Tables for Normal Tolerance Limits, Sampling Plans,
and Screening" gives an algorithm for calculating the second type.
http://www.itl.nist.gov/div898/handbook/prc/section2/prc263.htm
gives an approximation for the first type, that may be adequate for your
purposes. StatCalc is a standalone calculator that will calculation
essentially exact values for all three kinds of k-factors, but is limited to
the special case where df=n-1 for the second type of 2-sided k-factors
http://www.ucs.louisiana.edu/~kxk4695/StatCalc.htm

Jerry
 
Back
Top