I am using the GHS over a number of years. The 1983
GHS (as well as for other years) prior to 1998 is divided in a number of small
datasets (modules) (i.e. person data, household data, income data, etc.).
I have been searching the ESDS and the UK Data Archive homepages to look for
instructions about merging the different modules. However, I haven't found
such information. The only two common variables which allow merging the different
data files are "hserno" and "persno". By applying the
command "merge hserno persno" one can uniquely identify observations
in the master data. However, there is only one common variable "hserno"
between the household data module (household.dta) and the person data module
(person.dta). So, if I do "merge hserno" the command I get from
Stata is that "variable hserno does not uniquely identify observations
in the master data" whereas my master data is the individual data "person.dta".
Also, shall one keep observations that occur in both datasets (i.e. when the
variable "_merge" is equal to 3)? If this is the case, then too
many observations are being thrown away. Furthermore, how can one derive individual
identification numbers through the household serial number variables and other
variables? Are there some instructions on this?
