Based on the panel data samples of 30 provinces(due to the lack of data, the statistical data mentioned here do not include the Tibet Autonomous Region, the Hong Kong Special Administrative Region, the Macao Special Administrative Region and Taiwan Province) in China from 2012 to 2021,the three-stage DEA was used to measure S&T innovation efficiency,explored its spatial autocorrelation and the driving mechanism of S&T innovation efficiency. The results show that China’s S&T innovation efficiency is in an unbalanced and insufficient state, and on the whole, the S&T innovation efficiency in the eastern region is higher than that in the central region, which is higher than that in the western region. It is found that environmental factors and random noise have a significant impact and interference on the measurement of S&T innovation efficiency. From the perspective of spatial correlation, regions with high S&T innovation efficiency have not driven the overall situation of regions with low S&T innovation efficiency. Further analysis shows that the level of economic development, professional quality S&T personnel, innovation ability of scientific research institutions and technology export have a positive impact on S&T innovation efficiency. It is also found that expenditure, enterprise innovation ability and technology absorption have a negative impact on S&T innovation efficiency. Based on this, countermeasures and suggestions are proposed.

S&T innovation efficiency  /  the three-stage DEA  /  spatial autocorrelation  /  driving mechanism