Aiming at the defects of the existing researches on the correlation coefficients of probabilistic hesitation fuzzy sets(PHFSs), such as not considering the number of membership degrees and having counterintuitive phenomena, a new hybrid correlation coefficient is proposed. The hybrid correlation coefficient can comprehensively reflect the individual and global correlation among the PHFSs, which is more comprehensive and reasonable than the existings. Firstly, three factors of integrity, distribution and length of elements in PHFEs are considered comprehensively, and three basic correlation coefficients of mean, variance and length rate are defined respectively. On this basis, the hybrid correlation coefficient is obtained, which is proved to meet the axiomatic definition criterion of correlation coefficient. The results of case analysis show that the hybrid correlation coefficient overcomes the defects of the existing PHFS correlation coefficient. Based on the hybrid correlation coefficient, the multi-attribute decision making method in the PHFS environment is further designed. Finally, the validity and rationality of the correlation coefficient proposed in this paper is verified by a specific case analysis.