To examine a change in unified efficiency of provincial power industry, this study proposes an approach which combines Data Envelopment Analysis-Discriminant Analysis (DEA-DA), DEA environmental assessment and a rank sum test. The proposed approach is designed to overcome the following difficulties: (a) how to classify various decision making units into different groups, (b) how to identify the existence of group heterogeneity, (c) how to measure unified efficiencies of power industry in different regions, (d) how to separate among various unified efficiencies, and (e) how to unify them into a single measure which expresses total efficiency. To document the practicality, this study applies the proposed approach to examine unified efficiency measures of Chinese provincial power industry from 2009 to 2015. We obtain three empirical findings. First, the unified efficiency measurement identifies an existence of heterogeneity between two groups of provinces. Second, profound differences were confirmed in unified efficiency at a provincial level. Special attention should be given to the provinces with poor performance under both natural and managerial disposability. Finally, under both DEA and DEA-DA frameworks, large differences were confirmed between natural and managerial disposability. These two concepts may assist in developing well-designed environmental policy.