Creating a 2000 IES-LFS Database in Stata

PROVIDE Project Technical Paper 2005:1

Next comparisons between person- and household-level data for province, age, gender,
location and race are made. The province variable was a perfect match for all observations
and needs no further investigation. As far as age is concerned, we find that 1,214 observations
report different ages. A further 2,091 observations are missing in either the LFS or the IES
data, and hence cannot be compared. The remaining 102,359 of the observations
(approximately 97% of the sample) report the same age. Furthermore, in 278 cases age only
differs by one year, which could be accepted as an actual birthday taking place between the
surveys or a minor reporting error. This suggests that the merge is fairly accurate on account
of the age variable.

In Table 3 gender, location and race are cross-tabulated. The accuracy rates are very high
for all these variables, and differs very little between the person- and household-level
variables. This suggests that the mismatched person numbers is, after all, not such a big factor
(see footnote 14). The location variable is, however, a bit worrying, with a substantial number
of households (1,415) reporting rural in the LFS and urban in the IES, giving an accuracy rate
of only 91.1%. This perhaps points at a definitional difference of urban and rural between the
two surveys.

15

© PROVIDE Project

<<   17   18   19   20   21   22   23   >>

More intriguing information