The paper investigates the effect of time of day at an Automated Teller Machine (ATM) interface. Two dependent variables, time taken and error rate were measured. 11 male students participated in the experiment which involved a set task at an ATM simulated on a PC under controlled environment. Subjects were asked to perform the task at three different times of the day, 9.00am, 1.00 pm and 9.00 pm. The order of the sequence inn which the subjects performed the task was randomized by Latin square design to minimize order effect. One-Way Analysis of Variance (ANOVA) on transaction time and error rate indicated significant difference (p=0.001 for both). Student Newman Keuls test for time indicated least time taken in the morning and there was no significant difference in time in the afternoon or evening. The same for error rate indicated least error rate in the morning with no significant difference between afternoon and evening.

Thus designers when designing complex interfaces for 24 hours use like ATM, mobile, or even complex interfaces like control display unit of a hospital ICU should consider the fact that such interfaces are vulnerable to errors and slow responding, when tasks are carried out at different times of the day.