Tradeoffs between different protection policies and policies on developing natural spaces require knowledge of the preferences of their users. However, measuring willingness to pay can pose a problem when individual preferences do not respect some of the usual axioms of rationality. One of these, the axiom of continuity, states that it is always possible to compensate for a drop in the quantity of one good or attribute consumed, by a larger quantity of another good or another attribute. When the axiom is contravened, then we talk about "lexicographic preferences". This article is based on a survey using the multi-attribute choice experiment. We define a lexicography as "declared" when the individuals declare that they favour certain attributes when formulating their choices between the different alternatives, and we distinguish an "apparent" lexicography, as revealed by the choices actually made. The two notions overlap only partially or may do so without being consistent if the declarations of those surveyed are not in agreement with their choices. This results in the definition of a third type of lexicography, called "crossed". We examined the frequency and the determinants of these different lexicographies in a specific context, that of visiting a coastal forest. In this context, the lexicography, in whatever form, appears to be very extensive. The declared lexicography concerned 41% of individuals, and the apparent lexicography 72.5%. However 31% of those surveyed presented these two forms of lexicography simultaneously, and only 60% were "consistent" between their lexicographic choices and their actual declarations. The econometric results show that a lesser knowledge of the item being assessed and a lesser ability to reply to the questionnaire appeared to be the two main factors to account for the crossed lexicography.