A study was conducted to determine how well the display format effects described by Tullis (1983, 1984) and the resulting prediction equations could be generalized to other display situations. Task complexity and visual monitoring load were identified as task variables which could potentially moderate the format effects and, thus, were varied factorially. The current study also sought to extend Tullis's findings to tasks where the use of several pieces of information from predictable display locations is required. In general, the data indicate the need to study Tullis's format dimensions more fully before using his regression equations to evaluate display designs for use outside the task situation in which the equations were developed. Also, subjects were unable to evaluate their performance accurately under alternative display designs. Their evaluations seemed to be determined mostly by the perceived ease with which information was extracted from the display. This outcome should serve as a warning to system designers. That is, empirical human performance research should be conducted when performance is the paramount design criterion and a validated prediction system, such as the one developed by Tullis for search, is not available.
