Graphical Methods, Inductive Causal Inference, and Econometrics: A Literature Review

Abstract

Recent work with graphical methods for inductive causal inference with observational econometric data is reviewed and compared with earlier work. Two alternative algorithms are described. Caveats on applications are discussed.

Foreign portfolio investment – economic growth nexus: Directed acyclic graphs analysis

“…Ancak bununla birlikte, teori çoğu zaman kesin bir nedensel yapı ortaya koymak adına oldukça heterojendir veya altta yatan nedensel yapıyı tanımlamak için yeterli bilgiyi sağlamamaktadır. Üstelik bu tür modeller, teoride nedensel yapı genellikle verilerin istatistiksel özellikleri tarafından belirlendiği için gözlemsel verilere dayanan nedensel yapıyı belirleyememekte ve yanlış nedensel çıkarımlar ortaya koymaktadır (Kwon ve Bessler, 2011). "Tümevarımsal nedensellik" gözlemsel verilere dayanır ve değişkenler arasındaki koşullu bağımsızlıklardan nedensel bir graf ortaya koyar; oysa "tümdengelimsel nedensellik" ya doğuştan idelerden (düşünce) ya da varsayılan davranışların matematiğinden ortaya çıkar (Li, Woodard ve Leatham, 2013: 620).…”

Signal Fusion and Semantic Similarity Evaluation for Social Media Based Adverse Drug Event Detection

“…In this study we propose a new methodology to fuse signals from three different social media channels: Twitter, discussion forums and FDA Adverse Event Reporting System (FAERS) based on a causality model for ADE surveillance. Many studies have exhibited the usefulness of causality models in solving similar identification problems in economics [15]. We used graphical causal models to discover potentially hidden connections between data channels and use such associations to generate signals for adverse drug events.…”

“…For causality modeling, we used graphical causal models [15], [29], [36]- [38] to represent causal relations, and then used the Granger causality test [28], [39]- [42] to detect potential causal relations. Causality between two variables, say A and B is determined by checking their relationship with a third variable, say C, in particular their informational (in)dependence with C. For example, Graph (19) in Figure 3.1 (each rectangle contains equivalent structures.…”

“…Causality between two variables, say A and B is determined by checking their relationship with a third variable, say C, in particular their informational (in)dependence with C. For example, Graph (19) in Figure 3.1 (each rectangle contains equivalent structures. Figure adapted from [15]). An arrow indicates dependence between nodes.…”

“…Since C has two parents B and D, both are involved in its representation. By computing all the possible joint probabilities for a given network triple, Kwon and Bessler [15] identified 11 possible classes of observationally equivalent causal structures for a given network triple. Figure 3.1 shows these classes for variables A, B and C. Each block denotes an equivalent class.…”