Living Standards Measurement Survey 2004 (Wave 3 Panel)

w3_hh_all
The file has information at household level, summing up data drawn from section cover page (metadata), dwelling, numdurables, agriculturea1, agriculturea2, agricultureb1, agricultureb2, filters, remittancesa, remittancesb, socialassistance, subjectivea, subjectiveb, household interview outcome where information has been asked the household head or the most knowledgeable person. As in the individual level case, sometimes a reshaping has been applied to the original files to insure that each record has been assigned to a single household.

w3_hh_basic
The file contains data referred to sections: cover page (metadata), household interview outcome, dwelling, filters, social assistance, numdurables, subjectivea, subjectiveb. It is a sub-set of the w3_hh_all.

w3_hh_farm
The file has information drawn from the following modules: cover page (metadata), household interview outcome, agriculturea1, agriculturea2, agricultureb1, agricultureb2. As w3_hh_basic, it is a sub-set of w3_hh_all.

w3_ind_allx
This is an individual-level file for all survey members including children under 15 years old, which collect information from sections: control form (hhroster), cover page (metadata), panelmembers, educationorig, educationnew, health, fertility, labor and migration. The file has been organized to have an individual record for each row; hence sometimes a reshaping of the original individual files was needed. It occurs because for some sections data were collected by item instead of by individual.
Wherever possible, variables have retained same name and label of wave 1 and 2. A detailed description of them is available in the documentation file "Codebook_w3_ind_all".

w3_roster_all
The file has information from the household control form (hhroster) at individual level. Attention should be paid in using the w3_roster_all (and w3_ind_all) file, as it contains 168 duplicated records of individuals. Out of 168, 167 have moved away and 1 is "not present". The reason for this apparent anomaly is that migrants have been followed, whenever possible, in their new family, and they appear twice in the dataset, once in their old (classified as "not member", because moved) and again in their new household (considered as valid members, instead).
[Note: They have an equal m1_q01 personal identification code (pid), but a different m0_q00 household identification code (and, hence, chid). Depending on the users' objective, these duplicated records can be dropped so that national representativeness is guaranteed. A solution to leave them aside in the APS 2004 is to select individuals at their final location residence, through the variable cfinloc. The APS 2003 shares the same feature in w2_ind_all and w2_roster_all files: in that case the relevant filter variable is bfinloc.]
The file contains a sub-set of variables present in w3_ind_all and includes all enumerated household members

Sampling weights