How to determine the stability?

E

Eric

There are three lists of numbers under column A,B,C
Is there any build-in function to determine the stability within change?
For example,
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
On above list, there is no change at all, which is considered very stable.

Does anyone have any suggestions on following lists?
Thanks in advance for any suggestions
Eric

Under column A:
1064 1099 1116 1154 1105 1118 1092 1089 1087 1055 1057 1058 967 923 874 845
864 950 1024 971
Under column B:
845 840 887 863 865 886 851 843 825 805 818 792 711 687 677 642 642 640 681
702
Under column C:
1064 1065 1100 1129 1098 1120 1086 1084 1070 1022 1034 1003 921 883 862 843
843 882 923 905
 
J

Joel

You can use Stdev - Standard Deviation as a simple method.

A better method is to use NORMDIST which uses the Standard Deviation to
calculate the width of the Normalized curve. The width of the normalized
curve is oftern refer to by the Term sigma. 3 sigma would indicate 95
percent of the results where inside th enormalized curve. 6 sigma is
something like 98%.
 
E

Eric

Does anyone have any suggestions on how to select the most stability among 3
lists?
If I use STDEV(), I cannot compare which one is the most stability with
different scaling.
Does anyone have any suggestions?
Thank everyone for any suggestions
Eric
 
E

Eric

Thank everyone very much for suggestions

After using the function confidence (95%, 2 S.D. 100 data samples)
there are the result, 0.133, 0.155, 0.137.
Does it mean that data under column A will be the most stable?
Do they compare like with like? since they are different set of numbers?
Does anyone have any suggestions?
Thank everyone very much for any suggestions
Eric
 
J

Joel

For 95% you should be entering .05 like the VBA help indicates.

What confidience means for 0.133 is that 95% of your Normalized Value will
lie on the x-axis over a span of 0.133. the smaller the number the more
stable your results.

3 sigma is normally refered to as 95% which means that 95% of your measured
values will be within the center of a normal curve. 6 Sigma results which is
98% is better results. 6 Sigma will have a smaller confidence number.

What we are doing is looking only at 95% of the data and throwing the rest
away and then seeing how the 95% of the data compares between the three
differrent sets of data. You may want to run the data at 90% and compare the
90% results to the 95% results.
 
D

David Biddulph

Yours is perhaps a different sigma from that usually used. The percentage
between the -3 sigma and +3 sigma points is about 99.73%
Try =NORMSDIST(3)-NORMSDIST(-3)
 
E

Eric

For NORMDIST(x,mean,standard_dev,cumulative), could you please tell me where
to locate the sample of data for measurement?
Thank everyone very much for any suggestions
Eric
 
E

Eric

Could you please tell me where to locate the data of sample for
=CONFIDENCE(0.05,2,30)? It is valid statement to insert a number in
CONFIDENCE(0.05,2,30), but it is not valid statement to insert a list of
sample data in CONFIDENCE(0.05,2,range)?
Does anyone have any suggestions?
Thank everyone very much for any suggestions
Eric
 
D

David Biddulph

If this is similar to your other question about the CONFIDENCE function, you
could use =NORMDIST(x,AVERAGE(range),STDEV(range),cumulative), but you need
to remember that NORMDIST is telling you the shape of the distribution if
you know that the population has a Normal distribution with a particular
mean and standard deviation, but if you have a sample of data it may or may
not have a Normal distribution. If, for example, your data is produced from
the RAND() function, which gives a uniform distribution, the NORMDIST
function isn't going to tell you anything very meaningful about that
population. It is wise to understand the statistical theory you are using
before you start trying to use the Excel functions.
 
E

Eric

Thank everyone very much for suggestions
When I use following code, error - #NUM occurs when all data 1 > numbers >= 0
Does anyone have any suggestions on how to fix it?

=CONFIDENCE(0.05,STDEV(range),COUNT(range))

Thank everyone very much for any suggestions
Eric
 
D

David Biddulph

I suggest that you recheck your formula and your data. Excel help for the
CONFIDENCE function will tell you in which situations you can get the #NUM!
error, and none of them should occur for the formula given, unless all the
sample values are the same (which would, of course, give a STDEV of zero,
and the confidence range has a trivial answer, being of width zero for any
confidence level). The fact that your numbers fall between 0 and 1 should
not be a problem.

What values does your sample give you for STDEV(range) and COUNT(range)?
 
E

Eric

There is a list of numbers

Conference(0.05,2,data of samples) = #NUM!
STDEV = 0.284907799
COUNT = 9

Do you have any suggestions on solving conference problem?
Thank everyone very much for any suggestions
Eric

0.404304233
0.791429142
0.588616716
0.114627819
0.816068776
0.448024832
0.375591582
0.982228653
0.869244707
 
D

David Biddulph

You are obviously struggling to understand what you have been told about the
syntax of the CONFIDENCE function. Please read again the description and
example in Excel help, and read that in conjunction with what I and others
have told you here.

In your case the STDEV is 0.284907799 and the sample size is 9, so
=CONFIDENCE(0.05,0.284907799,9) gives a result of 0.186136.
Hence you have a 95% confidence that the population mean is in a range
between +/- 0.186136 from your sample mean (hence between 0.4128 and
0.7850).
You get the same answer if you use the formula
=CONFIDENCE(0.05,STDEV(A1:A9),COUNT(A15:A9)) as I suggested earlier.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Similar Threads

Help with this challenging VBA coding 3
Outlook connector crashes outlook 0
Outlook crashes with Outlook connector 2002 0

Top