Census data should leave no room for doubts

The Census and Statistics Department seems to be fighting an uphill battle to clear its name over the scandal involving data fabrication. A two-month probe led by the department's chief showed that although there were inconsistencies in some field data, there was no proof of frontline staff systematically faking answers. The taskforce therefore concluded that there was no evidence to query the overall authenticity of its statistics.

Statisticians may have their own ways of determining whether the inconsistencies are serious enough to compromise data. But the conclusion is hardly convincing to the layman. A private company commissioned to verify the General Household Survey found that 2.5 per cent of the results were inconsistent. But the discrepancy figure for whether respondents were ready to work in the next seven days - a question crucial to determining the jobless rate - was 11.5 per cent. This sits oddly with claims that the figures are trustworthy.

When explaining the inconsistencies, the census chief argued that it did not necessarily mean the data was wrong. The task force said it would not rule out individual field workers not adhering to guidelines.

There is no evidence to suggest the department has reached its conclusion before investigation. But if the department's logic is any reference, such possibility cannot be ruled out, either. The public can be excused for having the impression that the department is trying to play down the problem and shirk responsibility. Frontline staff are also justifiably upset that the taskforce failed to ascertain to what extent the management is responsible for the problem, which is said to stem from staff pressure to meet performance targets.

For meaningful research and sound public policy decisions, statistical data has to be accurate and reliable. The scandal has raised doubt over the integrity of our figures. Better efforts are needed to restore local and overseas confidence in the data-collection system.