各位大佬，请教一下模型因变量是来来自cnki报纸数据库中三年正面报道的计数数据，自变量是来自调查问卷的，这种情况下该如何做回归分析？
我看到有一篇文献提到“Because our measures of negative journalist reports are count variables with extra-Poisson variation (i.e., the number of negative journalist statements about the firm in various categories), and these variables could be correlated, we used seemingly unrelated negative binomial regression to estimate journalist reports(Wooldridge 2003). ”但我没看明白该如何操作。
也有文献是把每年的正面报道和负面报道聚类成一个系数，然后用tobit模型处理，但好像没法和调查问卷的自变量做回归。我对统计分析这块还是各小白，还请大家指点迷津，谢谢。