In the context of the digitalization of big data information, to solve the problems of timeliness and accuracy of traditional online opinion dissemination models. This paper firstly proposes a vector autoregressive model in the research problem of online opinion dissemination model and constructs a vector autoregressive model based on multivariate time series and unit root test. Then the emotional information entropy is determined based on the statistical characteristics of the emotional index, and the evaluation index is mainly composed of the current mainstream online public opinion dissemination modes. Finally, data analysis of online opinion dissemination modes based on the vector autoregressive model is conducted, and the dissemination efficiency of different modes is compared based on the analysis results. The results show that in the microblog dissemination mode, online public opinion’s dissemination efficiency remains in the range of 38.02% to 44.81%, and its average dissemination efficiency is 41.51%. In the online short video dissemination mode: the online public opinion dissemination efficiency remains between 32.15% to 35.74%, and its average dissemination efficiency is 33.97%. In the consultation mode, online public opinion’s dissemination efficiency remains within the range of 22.06% to 28.99%, and its average dissemination efficiency is 25.61%. In terms of the average dissemination efficiency, the microblog dissemination mode performs better than the other two. Through a comprehensive and objective study of online public opinion dissemination modes, this study has a guiding reference value for improving online public opinion dissemination to promote the development of online public opinion dissemination in China.