27.12.2012 Views

The 1995/1996 Household Income, Expenditure - (PDF, 101 mb ...

The 1995/1996 Household Income, Expenditure - (PDF, 101 mb ...

The 1995/1996 Household Income, Expenditure - (PDF, 101 mb ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

in the speed of processing of any survey at CAPMAS.<br />

LD.10 Keyer Productivity<br />

Table 1.5 is an annualized summary of keyer productivity measures for<br />

the data entry operators. "Production" is the nu<strong>mb</strong>er of questionnaires for<br />

which operators were ultimately responsible. "Speed" is the average nu<strong>mb</strong>er<br />

of keystrokes per hour, timed by CENTRY. "Error" is the average error rate<br />

- the nu<strong>mb</strong>er of times they made an invalid keystroke per 1000 keystrokes.<br />

<strong>The</strong> intention of this table is to show the evaluation detail that good data<br />

entry software provides, as well as the heterogeneous nature of skills, but<br />

overall the high degree of professional capability in the core of the<br />

processing team.<br />

<strong>The</strong>re were 9 full-time operators, and 6 part-timers, overseen by three<br />

managers intimately connected with the allocation of work flow. Data entry<br />

must be considered the bright spot in the <strong>1995</strong>/<strong>1996</strong> HIECS, not simply<br />

because speed rates were above the international standard of 8000 keystrokes<br />

per hour. <strong>The</strong> job was a very big one, often tedious, and all of the data was<br />

double-keyed (verified - entered twice for accuracy). Yet turnover was not a<br />

serious problem. In short the success of the Egyptian HIECS owes much to<br />

the computer operators who entered the data.<br />

LD.ll Edit Routine<br />

<strong>The</strong> CONCOR edit routine produced error reports about lack of<br />

consistency between items from verified data. <strong>The</strong>se reports were then used<br />

to correct the raw data. It is possible to summarize annually what the<br />

common sources of error were, given the monthly batch samples. Usually<br />

approximately 82 PSUs were represented per month, so the edit manager<br />

selected one random questionnaire from each of the 82 monthly PSU batches,<br />

and annualized this, to produce the table of common frequency errors for the<br />

year. Assuming that this sample was representative, two-thirds of all CONCOR<br />

errors were typical of the ones in Table 1.7.<br />

<strong>The</strong> main sources of inconsistency in the verified (and prevalidated)<br />

HIECS data remained row and column summary errors. <strong>The</strong> computer was<br />

much more efficient at catching enumerator summation errors than were the<br />

editors and coders in the field, or the workers in the CET department.<br />

LD.12 LAN<br />

L26<br />

Fifteen computers comprised the HIECS local area network (LAN), two<br />

primary servers, three analysis computers, and ten computers for data entry.<br />

<strong>The</strong> LAN setup initially permitted the linking of only the ten data entry<br />

terminals due to incompatible components. <strong>The</strong> two primary servers were<br />

rotated and used for backup on a scheduled basis, but this sequence, too, was<br />

derailed a nu<strong>mb</strong>er of times by difficulties restoring the system. Despite<br />

having two uninterruptible power supply (UPS) units, electricity supply was<br />

problematic. CAPMAS would not permit electricity to remain on after hours,<br />

necessitating booting up the system every day. When power outages<br />

occurred, approximately three days per month, the raw batch files were<br />

corrupted by CENTRY screen characters, requiring later correction by editing.<br />

In October of <strong>1995</strong>, management spent a good deal of downtime

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!