In the spirit of Nonaka and Takeuchi's Socialization, Externalization, Combination and Internalization (SECI) model, this paper explores how to integrate the heterogeneity of managers' attitudes, as a behavioral dimension of tacit knowledge, into risk scoring for software offshoring. We propose a linear mixed model approach to make individual heterogeneity explicit within a conjoint analysis experimental framework. The model is then estimated using a data set on Japanese software outsourcing. Based on heterogeneity elicitation, an optimist-pessimist categorizing of managerial attitudes towards risk reveals the drivers of such heterogeneity. We find that experience and the criteria used for selecting vendors are the key drivers. Our behavioral risk scoring, incorporating the attitude heterogeneity, outperforms the ordinary least square (OLS) estimate in risk prediction and yields real benefits in terms of risk reduction, forgone benefits saving and profits improving.