Thanks for your reply - unfortunately I do not have access to someone to
guide me through the statistical literature. Advanced statistcal
techniques isn't my strongest point by a long way - I was hoping that
the solution would be a lot simpler then the suggested multiple
imputation... ideally more along the lines of fitting a linear trend to
the existing data in order to estimate (via extrapolation) any gaps at
the beginning and/or end of a sequence.

Would anyone be able to design a VB macro tailored to the example
described in my original post?

Thanks again,
Steve


Jerry W. Lewis Wrote:
A single estimate ("imputation") of each missing value may simplify the
analysis, if all you are interested in is the equation coefficients.
However, your approach will invalidate any tests of the model or
estimates of
its precision, since the analysis will be carried out under the
assumption
that you have more data (and more critical data at the ends) than you
really
do.

Better approaches would include multiple imputation
http://www.stat.psu.edu/~jls/mifaq.html
(or any of a number of sites suggested by a Google search) or some kind
of
maximum likelihood analysis that takes account of the missing values.

Do you have access to a statistician to help guide you through this
literature?

Jerry


--
smurray444
------------------------------------------------------------------------
smurray444's Profile: http://www.excelforum.com/member.php...o&userid=28956
View this thread: http://www.excelforum.com/showthread...hreadid=506070