journey times are very consistent per rider, and far more independent of traffic conditions than [for] larger vehicles such as cars. Cyclists do not travel in the main traffic stream and therefore avoid the main barriers (barring signals) that other road users face
Then there's a survey done by the Chartered Management Institute in 2004, which doesn't appear to be available itself any more but is referenced in a few places, notably this Haringey Council document:
According to a Chartered Management Institute survey (2004) of 4,000 of their members, cyclists are more likely to arrive at work on time, and are more productive and less prone to stress than their counterparts arriving by car or public transport. The Institute found that 58% of cyclists say they are never disrupted by traffic, compared to only four per cent of drivers. Nine per cent of cyclists say they are stressed by their journey to work, compared to nearly 40% of drivers; almost a quarter of motorists feel their productivity is affected by the stress of their commute, compared to zero percent of the cyclists.
The upshot is that cycling journey times are very reliable, much more reliable than car journeys, because cyclists can usually filter through stationary traffic. This implies that if more people switched from cars to bikes then overall journey time reliability would go up, all else equal. But if your main measure of journey time reliability excludes cyclists and focuses only on motor vehicle traffic, as TfL's does, then the risk is that your attempts to improve reliability for motor vehicles will come at the expense of a mode which will always be much more reliable.