Title Matrix versus paging designs in a brand attribution task
Workshop Workshop 2014
Year 2014

Purpose of the study: Typical paging designs (one item per page) are visually simpler than matrix designs but leave respondents at sea concerning the number of remaining decisions. We compare data quality in an online brand attribution task in matrix and paging designs when the latter provides progress feedback.
Design/methodology/approach: 2110 opt-in respondents indicated which of several brands of sweets were described by a series of attributes (e.g., “Is worth what they cost”) in one of nine conditions: 6, 14 or 28 attributes in a matrix or one of two graphical paging designs. The paging designs displayed “tiles” to represent the number of remaining attributes. The text on the tiles was either visible or not.
Findings: The paging designs led to better data quality than the matrix design by three measures: fewer breakoffs, shorter completion times, and less straightlining. The paging designs reduced breakoffs more when the number of attributes was smaller. Respondents took less time to complete the attributions with the paging design especially when the number of attributes was large. Straightlining was defined as not selecting any attributes (all of which were positive) for a preferred brand or indicating that all attributes described a non-preferred brand. These patterns of response were reduced more by paging for smaller numbers of attributes, a somewhat counter-intuitive finding.
Originality/value: This project is the first attempt, to our knowledge, to improve the quality of brand attribution by adapting design ideas from conventional web surveys. The graphical paging designs with progress information have not been previously used – they serve the function of progress indicators but do not have the now well-known effect of increasing breakoffs that conventional progress indicators do, perhaps because the feedback is inherent in the display.
Research limitations/implications: We don’t test a paging without progress feedback condition.
Practical implications: This research makes a compelling case for using paging designs that provide abstract, graphical feedback about the number of remaining judgments in a brand attribution task. Other online evaluations can similarly benefit from the results.