The cooling effectiveness of a turbine blade is conventionally validated through scaled model experiments which involve the challenge of thermal similarity. To ensure accurate prediction of cooling performance in a real engine, it is necessary to investigate the dimension scaling effect and establish a reliable correction method for the cooling effectiveness obtained from the experiment. This paper presents a novel method derived from a one-dimensional heat transfer model to correct the scaling effects on thermal similarity. A group of experiments with a dummy blade is conducted to verify the feasibility and accuracy of the correction method. Results show the dissimilarity between scaled articles due to geometry dimension, material thermal properties, and alignment of the thermal barrier coating. A down-scaled model has higher cooling effectiveness than its full-length counterpart because of the enhanced internal convective heat transfer and reduced external surface heat transfer. In addition, dissimilarity due to mismatched heat conductivity of blade material and surface convective heat transfer coefficient can be corrected using Biot number replacement. Misalignment of thermal barrier coating can be replaced by a pseudo-blade metal of equivalent heat resistance and wall thickness including the metal and coating layer. Experimental result from a scaled model is corrected to improve the prediction accuracy of its full-length performance. Corrected cooling effectiveness matches its prototype within an uncertainty of 10 %. The present method is applicable to evaluate gas turbine components by means of scaled model experiments.