How well may the direct linguistic declarations substitute AHP in defining accurate criteria weights?

This paper investigates the problem of defining the weights in multiple criteria decision‐making analysis. In particular, we analyze if weights determined by decision makers (DMs) using the classic analytic hierarchy process (AHP) approach can be reproduced using more straightforward direct linguistic weights declarations (DLWD). Using the dataset from an online decision‐making experiment, we analyze the concordance of systems of weights obtained by DMs from AHP and DLWD. Applying optimization models, we check if the discrepancies between the AHP and DLWD weights result from (a) inefficiently declared mixes of linguistic labels in the symmetrically distributed scale or (b) using the unbalanced linguistic term set by DMs. The main findings are that the systems of weights obtained by DMs in the experiment from AHP and DLWD differ significantly. While using DLWD, most DMs defined criteria ranks contradictory to those resulting from AHP. They avoided extensive use of extreme labels from the 7‐point scale. The optimization results show that DLWD based on the symmetric scale might produce the most similar weights to AHP if a full range of linguistic terms was used, not only the intermediate ones. The differences in DMs’ cognitive profiles seem not to affect the results.