While most of the dimensionless parameters that govern energy transport are readily apparent, there is some controversy as to whether the collisionality should be represented by the collision frequency normalized to the bounce frequency (ν*) or the density normalized to the Greenwald density limit (n/nlimit). To help resolve this question, experiments on the DIII-D and JET tokamaks have compared the normalized energy transport in ELMing H-mode plasmas with matched dimensionless parameters. When ν* was kept fixed, the normalized energy transport on JET and DIII-D was in good agreement, but when n/nlimit was kept fixed, the normalized energy confinement time on DIII-D was 20% lower than on JET (for all cases, n/nlimit < 1). Therefore, ν* appears to be the correct form of the normalized density, and scaling transport properties from present day tokamaks to ITER at fixed n/nlimit can result in incorrect predictions.