The Tanzania 2012 census has two weights - person and household - but they are the exact same except in 775 cases. In those cases the household weight is missing.
The Tanzania 2012 census weights (i.e. only the person weights) do not add up to the total popluation in Tanzania as release by the government - it is 3% off - and if one looks at the figures at the sub-country level the numbers are off by 20% in some cases (the official government reports and numbers are in this PDF: http://www.tzdpg.or.tz/fileadmin/docu…)
The IPUMS PDF says that the 2012 census used complex sampling and that the weights for both household and person must be used but that is not possible.
Are the household weights for the 2012 census wrong or is there a problem with the extract for person and household weights?