Comparing Subjective and Objective Measurements of Contract Complexity in Influencing Construction Project Performance: Survey versus Machine Learning

Abstract

Contracts are the main governance mechanism regulating construction participants’ behaviors and safeguarding project performance. Besides traditional subjective measurement via survey, machine learning has been utilized as a more objective method to measure contract complexity in the field of construction management. However, little is known about whether misalignment between these two measurements exists and whether their impact on project performance differs. To answer these questions, we collected 202 pairs of construction contracts and questionnaires. Construction contracts were analyzed by machine learning to obtain the objective measurement of contract complexity, whereas questionnaires provided the subjective measurement. Through a multifunctional perspective of contracts, we found that the two measures are positively correlated with each other for contractual coordination and adaptation but nonsignificant for control. Regarding the ongoing debate on the relationship between contract complexity and trust, we selected trust as another antecedent of construction project performance. The results showed that trust is only positively related to the subjective measurement of contract complexity, which has a direct impact on project performance. In contrast, the objective measurement strengthens the positive effect of trust on project performance. Theoretically, our study contributes to construction contract research by highlighting that different measures cannot be used interchangeably and that scholars should be aware of the measurement issue when conducting and assessing relevant research. Practically, construction project managers are provided with guidance on performance improvement through perspectives from both objective design and subjective perceptions of contracts.