QoS Standards and their measurements

The news came yesterday (or is it today?) that the Public Transport Council is 'satisfied' with public transport operators

"The Public Transport Council (PTC) said yesterday it was satisfied with the service standards of public transport operators over a six-month period ending in May.

The PTC said SBS Transit maintained its performance and complied with all the standards.

SMRT Buses met the standards on service provision, bus breakdown, scheduled trips operated, loading and safety. But in three instances, it did not adhere to scheduled timings for buses leaving from interchanges and must pay a penalty of $300."

Source: Today Online

I was intrigued and decided to look around online to see if I can find out how and what they measure. Now the PTC website does provide a nice and dandy breakdown of what the Quality of Service (QoS) standards cover. But it doesn't tell me how they measure those standards. For example, "bus service should adhere to not more than 5 minutes of its scheduled headway (frequency) upon departure at the bus interchanges and terminals not less than 85% daily." This is the measures that the public transport operators failed in 3 instances according to the news article. Seems easy to measure on paper, but it can be quite problematic to operationalize. Do you send independent auditor to all the bus interchanges and terminals to keep track of all the buses departure? I doubt. I can't find any information online on how they managed to get such data, but one possible way that comes to my mind is that they get the data from the public transport operator (I see bus drivers tapping in and out of the interchanges/ terminals), which raises the question of credibility.

This reminds of me the time when I used IRIS (Intelligent Route Information System) to check on the next bus arrival timing at a bus stop that is 2 stops away from the interchange, which minimizes the possibility of traffic jam. At first, the IRIS showed me that the next bus will arrive in 5 minutes. 10 minutes later, I checked it again and it said that the bus will arrive in 15 minutes. In the end, I waited for slightly less than 30 minutes in total. It raises the possibility that buses are scheduled to depart (with the drivers tapping out) but didn't. I'll give them the benefit of doubts though, as this is just my conspiracy theory.

Another way they could have measured that is to do random sampling, but that would make their statement that only 3 instances were found to have failed the standard an over stretching of the findings; probabilistic sampling gives you probabilistic data, so 3 instances of n sample population will mean that 3/n percent failed to meet the standard. How then do you justify the penalty since the penalty framework seems to deal with discrete measures rather than ratios.

So how exactly do they get/ measure their data?