the difficulty with this type of prediction is that you have to know the
distribution.
for a true gaussian distribution the 90 % confidence interval for the mean
would be +- sigma/ square root of the number in the sample. I have worked
with very few distributions which are sufficiently "Normal" to allow this to
be a good predictor.

You can run a monte carlo to make a prediction but some of the issue is what
level of confidence to you want and what is your definition of "waver".

"IC Data" wrote:

The blanks are part of the data, just incomplete. I want to be able to
assume a confidence level that the mean of the dataset would not waver if the
blanks were indeed complete.

"bj" wrote:

I do not understand your question,
how are you calculating margin of error and confidence level if you ignore
the blanks?
how do you want the blanks treated?


"IC Data" wrote:

Hello,
I have file with a total size of 1438. However, 919 of the total are blank.
I am doing an analysis on the 519 good items, but I also want to include a
margin of error and/or confidence level that includes the blanks.

How would I accomplish this?