PROVIDE Project Technical Paper 2005:1
February 2005
P0303Q0.. |
P0303Q02 |
P0303Q03 |
P0303Q04 |
P0~050101 |
P0~050102 |
P030~0502 |
P030~0503 | |
103. |
24 |
0 |
0 |
0 |
5280 |
10788 |
0 |
72 |
25875. |
12 |
0 |
0 |
0 |
0 |
4800 |
0 |
0 |
25895. |
12 |
0 |
0 |
0 |
1029.6 |
842.4 |
0 |
0 |
25905. |
12 |
0 |
0 |
0 |
5874 |
4806 |
0 |
0 |
25982. |
24 |
0 |
0 |
0 |
240 |
264 |
0 |
0 |
26015. |
6000 |
0 |
720 |
0 |
0 |
0 |
0 |
720 |
26031. |
840 |
0 |
0 |
0 |
0 |
0 |
0 |
900 |
26078. |
1236 |
0 |
0 |
0 |
679.8 |
556.2 |
7500 |
0 |
26102. |
1188 |
0 |
0 |
0 |
1201.2 |
982.8 |
0 |
0 |
26153. |
3600 |
0 |
0 |
0 |
3102 |
2538 |
2040 |
0 |
26180. |
39600 |
0 |
0 |
0 |
2983.2 |
2440.8 |
0 |
0 |
26189. |
1200 |
0 |
0 |
0 |
3828 |
3132 |
0 |
0 |
26198. |
11760 |
0 |
0 |
0 |
2724 |
1980 |
0 |
0 |
26202. |
12 |
0 |
0 |
0 |
6600 |
5400 |
0 |
0 |
26208. |
6000 |
0 |
0 |
0 |
2400 |
12000 |
0 |
0 |
26220. |
2400 |
0 |
0 |
0 |
9622.801 |
7873.2 |
0 |
0 |
26221. |
1200 |
0 |
0 |
0 |
9622.801 |
7873.2 |
0 |
0 |
26222. |
12 |
0 |
0 |
0 |
0 |
0 |
17496 |
0 |
26229. |
10800 |
0 |
0 |
0 |
8400 |
0 |
0 |
11244 |
26250. |
30600 |
0 |
0 |
0 |
16500 |
13500 |
0 |
600 |
26254. |
6000 |
0 |
0 |
0 |
19800 |
16200 |
0 |
0 |
The remainder of do-file cleanup.do looks at some of the other problems in the database.
The housing section (Part 3) of the questionnaire contains various problems. In section 3.3,
question 1.1 should, by definition, be the total of questions 1.1.1 and 1.1.2. For 4525 out of
the 26265 households interviewed this is not the case. Poswell (2003: 2) ascribes this to the
“wording problem” in question 1.1.1. Fortunately when the expenditure categories used by
PROVIDE are calculated, only the total monthly rent, i.e. the reported value in question 1.1,
is taken into account. Hence this specific problem does not affect the work.
More problematic is question 5.1 of section 3.3. In this question homeowners with bonds
are asked to provide a breakdown of their monthly instalments. The capital and interest parts
should add up to the total monthly payment, but this is not the case for 1213 of the
respondents. Since these individual components are used separately in the calculation of
various expenditure categories, it is important to attempt to identify the problem. These 1213
observations are classified into four different ‘types’ of errors, as shown in Table 9.
Table 9: Four error types in housing section (monthly instalment on bond)
Erro |
No of |
Problem description |
Action taken |
1 |
23 |
No total reported, only breakdown |
Breakdown provided is nonsensical; hence |
2 |
149 |
Assumed calculation errors (+/- R100) |
Total monthly instalment is recalculated |
3 |
846 |
Only total reported, no breakdown |
Capital-interest breakdown of 55/45 assumed |
4 |
195 |
Nonsensical reporting, incorrect capital_______ |
Same action as for error type 3._______________ |
TOT |
1213 |
The 23 observations of error type 1 are listed below. From the list it is quite clear that the
reported capital component is often nonsensical. It appears as if respondents reported the
42
© PROVIDE Project