隨機實驗通常被認為是識別變量之間因果關系的最值得信賴的方法。但是在社會科學,尤其是在新聞傳播學的科研活動中,大量有價值的研究還不得不依賴非實驗數據展開。使用觀測數據進行因果關系推斷的一個主要困難就在于數據的內生性問題。論文以新聞傳播學中的典型變量關系為案例,對什么是內生性問題、相關關系與因果關系在定量研究中的區別、內生性問題對因果推斷產生的影響,以及形成內生性問題的遺漏變量、選擇偏差、循環因果、測量誤差等原因進行了深入探討。通過回顧近年來在新聞傳播學主流期刊上發表的采用量化研究方法的論文后,發現處理內生性問題可能會成為未來傳播學定量研究的重要內容之一并對具體的方法提供了初步建議。

Randomized experiments are generally regarded as the most reliable approach to identify a causal relationship between two variables. However,in the social sciences,especially in the scientific research of communication,a lot of important studies have to be conducted with non-experimental data. One of the major concerns applying observational data for causal inference is the endogeneity problem. Taking typical variables in communication as examples,this paper discusses such significant questions as what is endogeneity,what is the relationship between correlation and causation in the quantitative study,howendogeneity threatens causal inference in communication studies,and specifically the formation of the endogenous concerns including omitted variables,selection bias,simultaneity,and measurement error.After carefully reviewing the papers published in mainstream journals of journalism and communication in recent years using quantitative research methods,this paper finds that dealing with endogeneity may become one of the main directions of quantitative research in communication in the future and provides some preliminary suggestions for the field.