Be careful of the trap of statistics

I agree with senior transport correspondent Christopher Tan that the statistics on rail network reliability are ultimately unconvincing ("Measuring rail reliability"; April 7).

Many statisticians will readily admit that quantitative data has considerable limitations, both practical and intrinsic.

First, the sheer number of possible variables in play could preclude a clear-cut conclusion.

In some instances, these contributory factors are not taken into account to adjust the data. This is, of course, assuming we are even aware of these factors to begin with.

In the case of the Land Transport Authority's (LTA) figures, an increased distance between incidents could, in theory, be attributed to a larger train fleet being run at higher frequency, thereby accruing greater overall mileage, rather than any significant improvement in mechanical reliability.

Second, it is problematic to draw a conclusion as sweeping as "improved overall reliability" from a single dimension of data.

Here, we can draw a comparison to economic indicators.

When studying the development of nations, gross domestic product merely describes a state's economic output; whereas the Human Development Index takes other aspects of development, such as healthcare and education, into account.

Likewise, the LTA's claim is based on the singular measurement of distance travelled between disruptions, without regard for, say, the time between events, the duration of said events, the number of passengers affected overall, and so on ("Rail reliability statistics out of touch with the ground" by Mr Sum Siew Kee; April 9).

Third, and perhaps most importantly, statistics fail to account for the qualitative experience of using mass transit, which is virtually impossible to measure and quantify.

Human beings are not naturally attuned to keep time or make precise measurements. We, instead, rely heavily on our own preferences, expectations and subjective experiences.

Thus, a commuter is unlikely to be impressed by the technical details of 200,000km between breakdowns, especially if he is facing a massive delay in getting to work.

It is, therefore, more meaningful to work towards constant improvement without self-congratulation.

Paul Chan Poh Hoi