The covariance is a statistical technique that is widely used to measure the dispersion between two sets of elements. This work develops new covariance measures by using the ordered weighted average (OWA) operator and Bonferroni means. Thus, this work presents the Bonferroni covariance OWA operator. The main advantage of this approach is that the decision maker can underestimate or overestimate the covariance according to his or her attitudes. The article further generalizes this formulation by using generalized and quasi-arithmetic means to obtain a wide range of particular types of covariances, including the quadratic Bonferroni covariance and the cubic Bonferroni covariance. The paper also considers some other extensions by using induced aggregation operators in order to use complex reordering processes in the analysis. The work ends by studying the applicability of these new techniques to real-world problems and presents an illustrative example of a research and development (R&D) investment problem.