A key theoretical prediction concerning willingness to pay is that it is positively correlated with benefit size and is assessed by testing the 'sensitivity to scale (scope)'. 'External' (between-sample) sensitivity tests are usually regarded as less powerful than 'internal' (within-subject) tests. However, the latter may suffer from 'anchoring' effects. This paper studies the statistical power of these tests by questioning the distributional assumption of empirical data. We present an integrated model to capture both internal and external variations, while controlling for sample heterogeneity, applied to data from a survey estimating the value of reducing symptom-days. Results indicate that once data is properly transformed, WTP becomes 'scale sensitive' and consistent with diminishing marginal utility theory.