I'm reading a paper titled "A Simultaneous Bus Route Design and Frequency Setting Problem for Tin Shui Wai, Hong Kong" (W.Y. Szeto, Yongzhong Wu) which attempts to model an improved bus network in my home town. In page 8, there is an assumption listed that "the passengers go to their destinations using minimal transfers" which I doubt.
I look up the reference lists and eventually found out this saying originated from a paper titled "The allocation of buses in heavily utilized networks with overlapping routes" (Anthony F.Han, Nigel H.M. Wilson), in page 229. It mentions that:
Step 1: choose the route with minimum number of transfers (most transfers are made at major transfer points); if there is a tie go to step 2, otherwise classify as captive flow and stop.
(step 2 omitted as not relevant)
These two path choice criteria are a good approximation of expected user behavior in a congested, overlapping network such as Cairo’s. Because of the level of crowding on most buses, and because of the extra fare required on each vehicle boarded, passengers will try to avoid transfers where possible.
... Also, because there are a large number of routes each with relatively high frequencies, it is unlikely that a multi-vehicle trip will be faster than a single-vehicle trip for movements having both options available.
However, the condition that "because of the extra fare required on each vehicle boarded, passengers will try to avoid transfers where possible." is obviously false in Tin Shui Wai and it specifies that free interchange is possible, also the routes involved have frequencies up to 12 minutes headway. This may lead passengers to change buses even when there is an option without transfer to save in-vehicle time.
So, is the model flawed?