Film cooling has been studied profusely in subsonic flows, but in transonic flow, the determination of adiabatic cooling effectiveness faces a dilemma in the state of the art. Specifically, derivation of reference temperature, or local recovery temperature, in adiabatic effectiveness is disputable. Some researchers designate it to be the adiabatic wall temperature for the uncooled model (linear regression method, LRM), but others calculate it from an iterative procedure based on a pair of cooling tests which only differs in coolant temperature (dual linear regression technique, DLRT). As the first of the kind effort to explore this dilemma, this paper carried out transient thermal measurements by infrared thermography, for transonic flow over an idealized blade tip model. Heat transfer experiments were conducted for the uncooled and cooled cases, at two mainstream temperatures (340K and 325K), and two coolant temperatures (276K and 287K). Data from these six experiments were processed by LRM and DLRT respectively. It is found that heat transfer coefficient is insensitive to temperature boundary conditions and data reduction methods, as expected. However, for adiabatic effectiveness, LRM results are sensitive to the 11K decrease of coolant temperature in areas confined to the upstream of cooling injection, and much less so to the 15K rise in mainstream temperature. DLRT result reduces globally and conspicuously with 15K increase in mainstream temperature. Furthermore, adiabatic effectiveness obtained by LRM is qualitatively different to that by DLRT, which is attributed to the large discrepancy of reference temperature between the two methods.