there are several method of solving missing values problems. You can read Hair et al., 2006 and Hair et al., 2010 books named ''multivariate analysis'' and SMART PLS guide. they have discussed in detail how to solve the issue.
You may be looking for Multiple Imputations. Not all versions of SPSS have this option but essentially you create a number of new datasets with estimations for each missing value. Then you take the mean of each estimated value for the cell and use that as your score in your stat analyses.