In the econometric-statistical field, to quantify a cause and effect relationship it is required a detailed preliminary work of knowledge of the context in which the studied phenomenon occurs, in order to provide useful arguments to attribute a causal interpretation to a correlation found empirically: ”Correlation does not imply causation”. The causal diagrams, more precisely the ’Directed Acyclic Graphs’ (DAGs), are an effective tool to synthesize and communicate the system of causal relationships that occur in the context in which the causal inference analysis takes place and, therefore, to set up the research work. This paper aims to explain, with simple words and examples, why causal inference requires a preliminary knowledge of the context and, then, how to use the DAGs to set your own research in order to find the searched cause-effect relationship. The paper is thought for who approaches to this discipline with minimal statistical bases: simple and multiple regression; graphs.
In ambito econometrico-statistico, per quantificare una relazione causa effetto è richiesto un dettagliato lavoro preliminare di conoscenza del contesto in cui si manifesta il fenomeno studiato, al fine di fornire argomenti utili ad attribuire una interpretazione causale ad una correlazione riscontrata empiricamente: “Correlazione non implica causalità”. I diagrammi causali, più precisamente i ’Directed Acyclic Graphs’ (DAGs), sono uno strumento efficace per sintetizzare e comunicare il sistema di relazioni causali che si presentano nel contesto in cui si svolge l’analisi di inferenza causale e, di conseguenza, per impostare il lavoro di ricerca. Questo elaborato mira a spiegare, con parole semplici ed esempi, perchè l’inferenza causale richiede una conoscenza preliminare del contesto e, poi, come utilizzare il DAGs per impostare la propria ricerca al fine di trovare la relazione causa-effetto ricercata. L’elaborato è pensato per chi si approccia a questa disciplina con minime basi statistiche: regressione semplice e multipla; grafici.
Diagrammi Causali per l'Inferenza Causale: Un'Introduzione
BERTIN, NICOLA
2021/2022
Abstract
In the econometric-statistical field, to quantify a cause and effect relationship it is required a detailed preliminary work of knowledge of the context in which the studied phenomenon occurs, in order to provide useful arguments to attribute a causal interpretation to a correlation found empirically: ”Correlation does not imply causation”. The causal diagrams, more precisely the ’Directed Acyclic Graphs’ (DAGs), are an effective tool to synthesize and communicate the system of causal relationships that occur in the context in which the causal inference analysis takes place and, therefore, to set up the research work. This paper aims to explain, with simple words and examples, why causal inference requires a preliminary knowledge of the context and, then, how to use the DAGs to set your own research in order to find the searched cause-effect relationship. The paper is thought for who approaches to this discipline with minimal statistical bases: simple and multiple regression; graphs.File | Dimensione | Formato | |
---|---|---|---|
Bertin_Nicola.pdf
accesso aperto
Dimensione
595.5 kB
Formato
Adobe PDF
|
595.5 kB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/35272