According to classical rationality presumptions the preferences should not be dependent on the response type and on the description of the problem in gain and loss terms. However, the data of empirical research often negate the invariance presumptions of both the procedure and the description. The examples of such negalion could be preference reversal and the framing effect. The main purpose of the research was determining whether the framing effect manifests itself after presenting not only choise but also numerical estimation task. The research subjects were given a task, which is very similar to Asian disease problem according to its logical structure, the only difference that it was economic to its content. It was mentioned in the task there was a danger of losing equipment and work places.
There was a hypothesis presented that the framing effect will be determined only when the research subjects are given the choice but not the numerical estimation task. The hypothesis was confirmed by the research data. When the estimation task was presented, most research subjects appreciated more the risky alternative, irrespective how the task was formulated. It is very likely that the reason for such results could not have been accidental guessing. According to the research data, the guessing hypothesis could be rejected with marginal significance level (p = 0,053). Since the estimations of the research subjects were not dependent on the task formulation characteristics, there is no basis to assume that the preferences of the research subjects were determined by high referente point. It is most likely that the necessity to estimame the attractiveness of the alternatives in quantitative indicators made the research subjects do a more detailed analysis of the tasks, and compare the magnitude and the likelihood of gain and loss. Under these circumstances the emphasis of the gain or loss already did not have any influence on the responses of the research subjects. After evaluating all the consequences, the research subjects gave higher score to the risky strategy because it gave a chance to avoid potential losies.
There should be the demand for the detailed analysis of problems. When the choice task is presented, differently from the numerical estimation task, the research subjects do not have such a need. One cannot postulate universal rationality standards that are not dependant on the structures of the task.