Joint prevention and control of atmospheric pollution (JPCAP) policies play a vital role in alleviating regional pollution. Based on Latent Dirichlet Allocation (LDA) model, we construct two policy strength measures of effectiveness and number, and investigate the effects of policy strength on air pollutant emissions for four types of JPCAP policies. The results show that the effects of economic incentive policy tools and supporting policy tools on emission reduction deviate significantly from policy preferences. Economic incentive policy tools are the most effective in promoting emission reductions in SO₂, NOx and dust, but their effectiveness are the lowest in reality. Supporting policy tools, with the highest strength, have little effect on emission reduction. Command-control policies and persuasion policies are both relatively high in quantity and effectiveness. In addition, policy strength plays a more important role in reducing air pollutants in key regions than in non-key regions. JPCAP policies have gradually changed from a single policy tool to multiple policy tools, and the government shifted its attention to improving the legal effectiveness of policies after 2015. Finally, we propose some policy implications to optimize JPCAP policies and address regional air pollution problem.