The use of information in the sequential evaluation of audit evidence
This study develops and tests a model of sequential information use that considers the interactive effects of commitment to a prior decision, task importance and information sequence on the data attended to in an audit judgment task. Auditors' conservatism, the tendency to weight negative cues more heavily than positive cues, serves as a central component of the model. In addition, the model is used to show that both primary and recency in data use can occur due to the interaction of these factors.
To test the model audit professionals evaluated the level of control risk present in the inventory cycle of a hypothetical company. Subjects either received no information on which to form an early belief, or were induced to form a commitment to a positive or negative initial belief. Task importance was manipulated by varying the level of inherent risk. The information sequence was manipulated by changing the order in which positive and negative evidence was presented. In addition to assessing control risk, each subject listed the information that was relevant to forming the assessment. One hundred twenty auditors participated from six large international accounting firms.
In general, the model's predictions were supported by the findings. As predicted, auditors exhibited a strong general bias towards negative information. Risk assessments were most negative, and negative evidence was attended to most, when auditors were committed to a negative initial belief, task importance was high, and the initial control test results were negative. Both primacy and recency were observed, with the model correctly predicting such information use in ten of the twelve cases investigated. The strongest recency tendencies occurred with no prior commitment and low task importance.
This study demonstrates the complex nature of the belief-updating process by illustrating the effects of various task characteristics on the process. It also identities factors that may account for conflicting findings of primacy and recency in prior studies. The results also emphasize the importance of incorporating the effects of task and environmental characteristics into a model of audit judgment.