Missing data ** New **

XGobi can now handle missing values.  In the data file, the missing
values are represented by the character '.' or by the words na or NA.
(Sounds like Sesame Street, no?)  To be quite clear, here are three sample
rows:

467 585 na   43   NA
 .  580 86   na   12
425   . 77   38    4

______________________ Imputation _________________________

Rescale when imputing
  A toggle button that controls whether the view should be rescaled
  when a new imputation is performed or read.

Perform random imputation
  Use 'group' variable

  Instruct XGobi to sample from the present values for each variable
  to populate the missing values.  If you have done some brushing to
  partition the cases, you could turn on the button "Use 'group' variable",
  and then the sampling will only be done using cases brushed with the
  same color and glyph.

Assign fixed value
  10% above max
  10% below min
  Specify:

  An alternative to imputation:  assign a fixed value which is either
  some fixed percentage above (below) the maximum for each variable, or
  specify some fixed value for the missings in the entire data set.

Imputation menu

  If you have calculated one or more sets of imputed values that
  you would like to view and explore using XGobi, you must construct
  two files, datafile.imp (containing the imputed data) and datafile.impnames
  (a list of imputation names, one to a line, to be used in the menu).

  The structure of the data in datafile.imp is as follows:
  Each column is an entire set of imputed data, ordered such that
  they will be used to fill in the missing data in column-wise fashion.

  Let's use the example above, but start by numbering the missing values:

  467 585 na_1  43  na_3
  463 580  86  na_2  12

  Each column of datafile.imp will contain 

  imputed_value_1
  imputed_value_2
  imputed_value_3