cor576.wp5
                                                May 3, 1995

Data collection for 324 and 324d was inconsistent with the data
collected for the selected child in the children's roster.  For
those cases where the children's roster indicated the selected
child lived with R, but an address was given in 324, a comparison
of R's address and selected child's address was done.  See
WLSPG112.LIS.  For almost all cases, these addresses were the same.
An address should only be given in 324 if the child does not live
with R.   So variable  RD008SK from the children's roster section
was used to determine if the selected child lived with R for the
codes assigned to RJ002SK.  Those selected children coded as living
with R in 324 but coded as not living with R for RD008SK  were
coded as not ascertained for RJ002SK.  198 cases were coded 999 due
to this inconsistency.

A check of date of interview revealed that these inconsistencies
occurred during the entire course of the interview.  See
WLPG113.LIS. A check of inum  was also done.   For those
interviewers who had an exceptionally high rate of inconsistencies
about place of residence, they actually had a higher percentage of
inconsistent interviews on these questions than they had compelted
consistent interviews for these questions.  See WLSPG113.LIS and 
WLSPG114.LIS.  Thus the conclusion that this data was collected
incorrectly consistently by a few interviewers and only
occassionaly incorrectly by other interviewers.