At least the newspaper was faithful enough to carry a follow-up story noting that:

Commuters are disputing a report that says buses here run on time and are not overcrowded.

Two weeks ago, the Public Transport Council (PTC) noted that between last December and May this year, buses were late on only 25 occasions and too crowded only 28 times. It put these lapses against the fact that 260 bus services ply the roads daily.

But commuters who have to wait 45 minutes or more for their bus are far from convinced.

Administrator Chan Whye Chuen, for instance, said he often waits between 45 minutes and an hour for SMRT service 960.

The 54-year-old, who travels from his office in Bugis Junction to his Bukit Panjang home, said: 'My bus stop is only five to six stops from the start point, so I can't imagine it would take 45 minutes to get here.'

-- Straits Times, 30 Oct 2008, Commuters
dispute upbeat transport study

To compound matters, a council member then defends their report by blaming commuters for being subjective and unscientific when criticising the council's findings.

PTC member Vincent Chua, a bus commuter himself, put it down to the 'personal experiences' of commuters: 'If you are counting on a bus to make an important appointment and it doesn't come on time, that makes a big difference to you and how you view the bus system.'

-- ibid.

Yawning Bread's interest is piqued when I see people quarrel over data, and so I made a little effort to locate the original PTC report that is the source of the dispute. However, because of the jargon used, I too am not 100% sure that my interpretation of the data is correct, but my reading of the report leads me to believe that the Straits Times' story was misleading.

Bus stop with real-time display board showing estimated arrival times of buses. 

 
Buses running late

Let's begin with the newspaper's statement that "between last December and May this year, buses were late on only 25 occasions." The ordinary meaning of such a sentence would be that there were 25 bus journeys (bus trips) when a bus ran late, causing passengers to wait inordinately long at bus stop along its route.

Did it not strike the Straits Times that there was a huge gulf between the ordinary meaning of the sentence and reality?

Consider this:

There are, as mentioned by the Straits Times, 260 bus services operating all over Singapore. On average, my estimate is that each service is expected to run every 15 minutes, over an 18-hour day -- the PTC stipulates that bus services must run for this duration daily.

That means there are about 18,720 bus trips per day. Over a six-month period, we are talking about 3.4 million bus trips, or 6.8 million trips if you consider both directions. You tell me only 25 of them ran late?

Given this definition, what does an "instance" of non-compliance mean? It does not mean a case of a single bus departing the terminal late. It means a day on which fewer than 85% of a service number's trips started off on time. (In one direction, or both directions? Not clear.)

Assuming the quality standard refers to both directions together, if a service is expected to run every 15 minutes or so, and therefore it has 144 trips (72 scheduled trips per direction x 2 directions) in an 18-hour day, the bus company would be happily in compliance if at least 123 (85%) of these trips depart on time, and "only" 21 trips depart late -– where "late" means "more than 5 minutes".

Clearly, therefore, the reported 25 instances of non-compliance does not mean 25 trips. It must mean 525 (25 x 21 = 525) trips or more.

Moreover, even a "compliant" service does not mean all its trips left the terminal on time. In fact, there must be many instances of "compliant" services during these six months, when perhaps 10-12 percent of the bus trips drove out late. So long as it doesn't trigger the 85/15 rule, it is "compliant".

The question that should interest us is whether the PTC's standards are too lax.

I also wonder about the integrity of data collection. As reported by the same Straits Times story, a bus is "on time" if the computer at the interchange tracks it as leaving the interchange within 5 minutes of the scheduled departure time. But how is data entry carried out?

Now consider commuter Chan Whye Chuen's grouse about service 960 being quite often 45 minutes late. You'd notice from the table above that 960 is not listed as having a single "instance" of non-compliance. But, as I have explained, so long as a service has 85% of its trips setting off on time, it meets the quality standard. This means that Chan's experience could well be real and still the service failure would not show up in the data.

PTC member Vincent Chua appeared to defend the bus companies when he said: "The delay on the roads caused by congestion is out of their control, so we cannot track them at bus stops."

Chan must have anticipated the excuse of road congestion. He told the Straits Times: "My bus stop is only five to six stops from the start point, so I can't imagine it would take 45 minutes to get here."

Indeed, the PTC's report does not discuss bunching along the route. As the Straits Times pointed out:

No tracking is done on the time it gets to each bus stop en route -- the only measure that counts to the commuter.

-- ibid.

 
Crowded buses

The exact wording of the PTC's standard on crowding is "Daily, each service bus loading shall not exceed 95% during weekday peak hours."

I found it impossible to unpack the significance of the report's finding -- 28 instances of crowding non-compliance -- because I could not find from the PTC's website any definition of these terms: (a) bus loading, (b) peak hours

Following an event at the Pioneer MRT station yesterday, Transport Minister Raymond Lim announced that the [Land Transport Authority] was lowering the maximum number of passengers per train from 1,700 to 1,600.

But falling in line with this is going to be moot, because train operator SMRT has said even the most crowded trains have carried only 1,400 commuters.

Mr Seah Kian Peng, a member of the Government Parliamentary Committee (GPC) for Transport, said: 'It's a bit odd for them to make the standards higher than what the operators are already working at, but maybe the ministry is trying to approach this is in steps and does not want it to be onerous to either operator.'

-- Straits Times, 31 Oct 2008, Penalty for late trains

Singaporeans know how impossibly packed some trains can be during rush hours, but apparently, our metro system has never quite breached the Land Transport Authority's (LTA's) definition of "maximum". That being the case, our trains are never -- officially -- crowded.

 
Simple improvements

The PTC can improve their reports. Leaving intact the existing section, where they report on the "instances" of non-compliance and the fines levied, they can add a new section, using plain language, that would give a more comprehensive picture of bus service performance. For example,

For each month,

1. How many hundreds of thousands of bus trips were scheduled in total? Of these,

2. What percentage of total trips departed from the terminals (a) 3 - 4 minutes late (b) 5 – 7 minutes late (c) 8 or more minutes late

3. What percentage of total trips arrived at destination terminals early or late (and giving details of variances)

Since we have equipped buses with GPS systems, we can also add new measures, for example,

4. For each bus route that is more than 10 km in length, to define three bus stops that represent the one-quarter point, mid point, and three-quarter point of the route; and for each of these points, to report monthly variances of actual arrival time against scheduled arrival time.

These will give us some measure of bus performance as experienced by the commuter. Only then will future reports not be treated as whitewash. 

© Yawning Bread