You can't have a file with a mixed encoding. If you happen to only have characters that don't conflict in the two code pages, you might be able to get away without Unicode, but it would be iffy.
What you could try is to set Unicode on and your locale to one of the two. Read the file and save it. It will be in Unicode. Then close the data, set the other locale, and read the second file. Save it. Now merge the two files. You have a Unicode file that has all the characters correct. Open Statistics, set Unicode off, and set the one locale you want to make primary, and open the merged file. If you are lucky, all the characters will work, but if there are conflicts, Unicode would be your only option.