Creating a 2000 IES-LFS Database in Stata

PROVIDE Project Technical Paper 2005:1

Poswell (2003) has no explanation for the discrepancy, apart from reporting or calculation
errors. The figure for total housing expenditure will not be used in further calculations, hence
no attempt is made to try and find the error.

Variable |

-------------+

P0303total |

P0303total~K |
diff |

Section 3.4 of Part 3 covers annual housing costs. These are costs that are typically made
only once (if at all) in a given calendar year. Expense items in this category include
alterations, improvements, and transfer costs. As seen in the Stata output table below, there is
an average difference of R16.72 between the calculated and reported total. These differences
occur in 171 of the observations. Closer inspection suggests that most of the differences are
simply calculation errors. For some observations the total annual housing expenditure was
reported as zero despite having reported expenses that supposedly make up the total. This
explains many of the large differences. The total annual housing expenditure figure will not
be used further, and hence no attempt is made to correct the discrepancies that exist.

Finally, section 3.5 of Part 3 covers expenditure on holiday accommodation. As shown in
the Stata output tables below the calculated total differs only slightly from the reported total.
The difference cannot be explained properly, but only occurs in 180 of the observations. At
this point nothing is done about the problem.

Part 5 covers expenditure on food items. All totals were recalculated and compared to the
reported totals. As shown in the Stata output table below, no substantial errors were found.
The largest difference occurred in question 5.3 (P0503tot*). The error is small enough to
safely ignore and is likely to be a calculation error.

A more serious problem in the food expenditure section is that about 351 observations
(1.3% of the dataset) report zero food expenditure. Given that respondents should also report

45

© PROVIDE Project

<<   47   48   49   50   51   52   53   >>

More intriguing information