The easiest thing to do is to contact your local library, and ask what kinds of census (for 1990) resources they have. Some may have pre-written crosstabs. Try to find a crosstab of Ancstry1 (first ancestry) vs. any one of the following:

• Pob (Place of Birth) (you will find that the vast majority of "American Indians" born outside of the USA were born in India)
• Lang2 (Language) (you will find that a large number of "American Indians" speak Asian Indian languages, the most popular of which is Hindi)
• Race (you will find that many "American Indians" are in the "Asian Indian" race)

Another option would be to contact the census bureau, and ask for an appropriate two dimensional crosstab, created from the 1990 long form data.

If you have the PUMS (long form) data available to you, another option would be to look at it yourself. Use the following command (in UNIX):

grep ^P filename | cut -b18-21,44-46,53-55 | grep 210920$ | cut -b1-4 > newfile

• The first 'grep' gets all person records out of the file and hands them to 'cut'
• The first 'cut' cuts out the weighted person count, place of birth, and primary ancestry, and hands them to the second 'grep'
• The second 'grep' extracts only those records which indicate people born in India who claim an ancestry of "American Indian", and hands them off to the second 'cut'
• The second 'cut' saves only the person count, and stores it in 'newfile'

To actually duplicate our results (performing multi-dimensional analyses, household level categorizations, error corrections, etc., instead of just doing one of the simple things above) would likely require a great deal of custom software development, which would consume a substantial amount of time and money.