So far, no conclusive empirical evidence exists that would help guide questionnaire designers to make informed decisions on whether rating scales should ascend from the positive pole (or highest level) to the negative pole (or lowest level) or should descend from the negative pole (or lowest level) to the positive pole (or highest level). While previous research found mean shifts and changes in proportions endorsing various response options under different scale directions, but little is known about the influence of scale direction on other properties of the resulting answers (e.g., factor structures, internal consistency of resulted factors). Also, only limited attempts have been made in trying to reveal the underlying cognitive mechanisms at play when answering response scales. In an experiment in the UK Innovation Panel administered face-to-face and over the Web, respondents were randomly assigned to answer two sets of multi-item batteries (5 items on life satisfaction and 20 items from the Balanced Inventory of Desirable Responding) on a descending scale or an ascending scale. Preliminary analyses indicate that the scale direction significantly affects resultant survey responses by pushing answers to the beginning of the scale. We found that the mode of data collection does not interact with scale direction – scale direction effect is shown in both the CAPI and the Web mode. Further analysis will focus on the influence of scale direction on factors structure and internal consistency of the resulted factors and the role of satisficing in explaining response behavior to rating
questions.