The present study investigates the impact of probability levels on the response mode bias in utility elicitation and its stability across a wide range of conditions. Experiments are performed with subjects from two different cultures, Austria and Pakistan, and a variety of methods to measure risk attitudes of subjects on a cardinal scale are used. Results indicate robust influences of probability levels on both the Certainty Equivalent method and the Probability Equivalent method of utility elicitation. Both methods are affected by probability changes in the opposite direction, creating the characteristic “tailwhip” pattern observed in previous research. Our study shows that this effect remains stable across different cultural backgrounds, measurement methods, and problem parameters, and is thus not an artefact but a reproducible phenomenon.