This paper investigates biases in the assessment of von Neumann-Morgenstern
(NM) utility functions. Based on experimental data of 73 subjects, certain
ty and lottery equivalence NM utility functions are derived and analyzed. W
e observe violation of procedure invariance in the sense of dependence on t
he probability used to estimate the functions and in the sense of the prese
nce of a response mode bias. We observe also a response mode bias for risk
attitudes. Considering the intensity of subjects' risk attitudes allows int
eresting observations of dominance properties of NM utility functions deriv
ed using different probabilities and/or response modes for their generation
.