Estimating the Amount of Eye Movement Data Required for Panel Design and Instrument Placement

A procedure is presented for estimating the amount of eye movement data required to obtain a specified statistical accuracy with a given degree of confidence. It is applicable when eye movement statistics are to be used in panel redesign and instrument placement. Application of the procedure involves taking a short length of data and dividing it into segments. Subsequently, Stein's double sampling plan is employed to determine the required number of segments to meet accuracy specifications. The assumptions associated with the procedure are examined for a specific eye movement experiment and are found to be valid. Application of the theory produces overall results having the anticipated degree of accuracy. The procedure should be applicable to a variety of human factors design problems.