SPC Excel Confidence Intervals

Hi,

The confidence intervals illustrated with SPC excel are not the same as those I caulcate myself and I'm not sure why so if someone could verify my logic I would really appreciate it. I have run 5 simulations and am calculating the confidence interval range as a result.

I calculated the following:

1. Batch means (mean of the means from the 5 sim runs): 2.976674
2. The standard deviation of the means from the 5 sim runs: 0.009827
3. The sample size is 5 (5 simulation runs)
4. The confidence coefficient is 0.95 so the level of significance or alpha value is 0.05 (1 – 0.95).
5. I calculate the margin of error using the Excel CONFIDENCE function passing in the level of significance (step 4), the standard deviation (step 2) and the sample size (step 3)
6. The lower confidence interval is thus the batch means (step 1) minus the margin of error (step 5)
7. The upper confidence interval is the batch means (step 1) plus the margin of error (step 5).

This gives me a confidence interval of 2.968061 – 2.985287

I verified this with SPC Excel (I entered in the batch means (step 1), the std dev (step 2), the sample size (step 3) and the confidence coefficient (step 4) and it gives me back a confidence interval of 2.964472168 – 2.988875832. I have read all the EXCEL SPC documentation and can’t understand why.

Thanks in advance.

Thanks for the response. I contacted them and this was there reply just in case anyone ever runs into the same problem:

"The limits you calculated using Excel’s CONFIDENCE are based on an assumption of a normal distribution. Our software uses this if the sample size is greater than 30 (although you can select the large sample case at any time). If it is less than 30, the software uses the t distribution. Limits are average +/- t(SD)/sqrt(n) where t is the t value for alpha = 0.05 and n-1 degrees of freedom. If you select the large sample case when you run the data, you will get the results you got with the CONFIDENCE function."