Family Structure and Income

I am working with a colleague to evaluate the median family income using PUMS data (in this case 2007-2011 5-year data that combines personal and household files based on serialno) and have run into a question about family structure versus household structure versus incomes.

For example, PUMS has some observations that look like the attached file (headers use PUMS variable names). All observations have the same serialno, the sporder 1-4 makes sense, the personal incomes (pincp) seem reasonable, and the household income (hincp) makes sense.

But, npf says the family size is 3 while the family income variable says every member of the household has the same family income? I would expect one member of the household would not have a family income (i.e. sporder = 1 to 4 but nfp=3 so one member of household is not a member of the family), or maybe the household has two families so it should have 2 different family incomes and the family sizes should all say 2. Suggestions about how to interpret this type of outcome, alternative variables to use, or errors in logic?