DOCSLIB.ORG

Automating Path Analysis for Building Causal Models from Data

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Automating Path Analysis for Building Causal Models from Data From: AAAI Technical Report WS-93-02. Compilation copyright © 1993, AAAI (www.aaai.org). All rights reserved. AutomatingPath Analysis for Building Causal Models from Data: First Results and Open Problems Paul R. Cohen Lisa BaUesteros AdamCarlson Robert St.Amant Experimental KnowledgeSystems Laboratory Department of Computer Science University of Massachusetts, Amherst [email protected]@cs.umass.edu [email protected] [email protected] Abstract mathematical basis than path analysis; they rely on evidence of nonindependence, a weaker criterion than Path analysis is a generalization of multiple linear regres- path analysis, which relies on evidence of correlation. sion that builds modelswith causal interpretations. It is Thosealgorithms will be discussed in a later section. an exploratory or discovery procedure for finding causal structure in correlational data. Recently, we have applied Wedeveloped the path analysis algorithm to help us path analysis to the problem of building models of AI discover causal explanations of how a complex AI programs, which are generally complex and poorly planning system works. The system, called Phoenix understood. For example, we built by hand a path- [Cohenet al., 89], simulates forest fires and the activities analytic causal model of the behavior of the Phoenix of agents such as bulldozers and helicopters. One agent, planner. Path analysis has a huge search space, however. called the fireboss, plans howthe others, whichare semi- If one measures N parameters of a system, then one can autonomous,should fight the fire; but things inevitably go build O(2N2) causal mbdels relating these parameters. wrong,winds shift, plans fail, bulldozers run out of gas, For this reason, we have developed an algorithm that and the fireboss soon has a crisis on its hands. At first, heuristically searches the space of causal models. This this chaos was appealing and we congratulated ourselves paper describes path analysis and the algorithm, and for building such a realistic environment. However,we soon realized that we could explain very little about the presents preliminary empirical results, including what we behavior of the fireboss. Weturned to regression analysis believe is the first example of a causal model of an AI system induced from performance data by another AI to answer some questions, such as, "Which has more system. impact on the time to contain a fire: the wind speed or the numberof times the fireboss must replanT’ But although regression assumedthese factors interacted, it provided no 1. INTRODUCTION explanation of their causal relationship. For example, we knew that the wind speed could affect the incidence of This paper describes a statistical discovery procedurefor replanning and not vice versa, but this causal, explanatory finding causal structure in correlational data, called path knowledgewas not to be found in regression models. Nor analysis lasher, 83; Li, 75] and an algorithm that builds would automated regression procedures (e.g., stepwise path-analytic models automatically, given data. This multiple regression) find causal models of our planner. work has the same goals as research in function finding Path analysis, however, is a generalization of regression and other discoverytechniques, that is, to find rules, laws, analysis that produces explicitly causal models. Webuilt and mechanisms that underlie nonexperimental data one such model of Phoenix by hand, and by automating [Falkenhainer & Michalski 86; Langley et al., 87; 1 path analysis as we describe below, we have been able to Schaffer, 90; Zytkow et al., 90]. Whereas function discover other causal explanations of how the Phoenix finding algorithms produce functional abstractions of fireboss works. (presumably) causal mechanisms, our algorithm produces explicitly causal models. Our workis most similar to that Readers whoare familiar with regression analysis might of Glymouret al. [87], who built the TETRADsystem. skip to the end of the next section, where we introduce Pearl [91; 93] and Spirtes [93] have recently developed path analysis, or Section 3 where we illustrate a path causal induction algorithms with a more general analysis of Phoenix. Section 4 describes our algorithm. Section 5 discusses two experiments, an informal one in which we applied the algorithm to Phoenix data, and a 1The term "nonexperimental" is perhaps confusing, because data are factorial experiment in which the behavior of the usually collected in an experiment. Nonexperimental means that the algorithm was probedby applying it to artificial data. experiment is over and the opportunity to manipulate variables to see effects has passed. Causal hypotheses must therefore be generated and tested with the data, alone. AAAI-9$ Knowledge Discovery in Databases Workshop 1993 Page 153 beta coefficients is plausible because betas are standard- 2. BACKGROUND: REGRESSION ized partial regression coefficients; they represent the ef- fect of a predictor variable on Y whenall the other pre- Path analysis is a generalization of multiple linear regres- dictor variables are fixed. Youcan interpret 131 as what sion, so we will begin with regression. Simplelinear re- happens to Y when only XI is systematically varied. In gression finds a least-squares line relating a single this sense, beta coefficients providea statistical version of predictor variable x to a performance variable y. A the control you get in an experiment in which X2 and X3 least-squares line is one that minimizes the sum of are fixed and XI is varied; in such an experiment, the ef- squared deviations of predicted values from actual values. fect on Y is attributable to X1. (Alternatively, the effects That is, simple linear regression finds a line ~ = bx + a 2. might be due to an unmeasuredor latent variable that is that minimizesY.i(Yi - Yi ) correlated with Xl; we will not consider this case here.) Multiple linear regression finds least-squares rules (i.e., planes and hyperplanes) for more than one predictor vari- able, rules of the form ~=blxl+...+bkxk+a. The regression^equation is better represented in standardized form as Y = fliXi +f12X2+f13X3 (standardized variables are denoted with uppercase letters). The interpretation of this model is that a change in XI of one standard d^eviation, Sl, producesfll standard deviations changein Y. Thus, beta coefficients axe comparable: if fll =. 4, f12 =. 8, then a standard deviation changein X2 has twice the influence on Y as a standard deviation Figure 1: The path model that corresponds to change in XI. the multiple linear regression of Y To fred beta coefficients the following equations are on Xl, X2 and X3. derived from the regression equation and solved: = + + So far we have described how a prediction model 9x. = + + x.x. I~ = fllX1 +/~X2+ f13X3gives rise to a set of normalequa- tions, and to beta coefficients that makethe prediction = + + model a least-squares fit to our data. If this were all we The previous equations can be rewritten in terms of could do, path analysis would be identical to linear correlations: regression and not worth the effort. The power of path analysis is that we can specify virtually any prediction q,x,rrx’ = ~[~ +r ~zrx, x:x, +x, f12 + [33rx,+ ~rx3x, xl model we like, and then solve for beta coefficients that (1) ensure the modelis a least-squares fit to our data. rrx, = ~rx,x, + flzrx, x, + f13 Clearly, with these three equations we can solve for the 3. PATH ANALYSIS OF PHOENIX DATA three unknownbeta coefficients. Wehave not shownthat Let us illustrate a path model other than the regression these coefficients guarantee that Y =flaXl +/~_X2+f13X3 model with an example from the Phoenix system [Cohen is a least-squares rule, but the interested reader can find & Hart, 93; Cohenet al., 89; Hart & Cohen, 92]. Weran this demonstrationin [Li, 75] or any goodstatistics text. ...... Phoenix on 215 simulated forest fires and collected many The three equations (I) are called the normal equations, measurementsafter each trial, including: and they have an interesting interpretation, illustrated in WindSpeedThe wind speed during the trial Figure 1. Consider the first normal equation, rl’Xt =ill +fl2rx2xz +fl3rxax~. The [31 term is represented RTK The ratio of fireboss "thinking speed" to the in Figure 1 by the direct path between Xl, and Y; the rate at whichfires bum second term is represented by the indirect path from Xl, NumPlans The numberof plans tried before the fire is through X2 to Y; and the third term is represented by the contained indirect path through X3. Thus, the correlation rl’xt is given by the sumof the weightsof three paths in Figure 1, FirstPlan Thename of the first plan tried where the weight of a path is the product of the coeffi- Fireline The length of fireline dug by bulldozers cients (either correlations or betas) along the constituent links of the path. The second and third normal equations FinishTime The amount of simulated time required to have similar interpretations in Figure 1. By convention, contain the fire curved arcs without arrows represent correlations and di- We specified a prediction model and the path model rected arcs represent causes. The causal interpretation of shownin Figure 2, and solved for the path coefficients on Page 154 Knowledge Discovery in Databases Workshop I998 AAAI-93 each link. The coefficient on a path from X to Y is the correlation of X and Y if X is uncorrelated with any other 4. AUTOMATIC GENERATION OF PATH node that points to Y; otherwise it is the beta coefficient MODELS from the regression of Y on all the correlated nodes that point directly to it (including X). Withthis rule it is easy The question that motivated the current research is to solve for path coefficients by hand, using the whether models like the one in Figure 2 can be generated correlation matrix and running multiple regressions as automatically by a heuristic search algorithm. This sec- necessary. tion describes such an algorithm.
Load more

Recommended publications
  • Causal Inference with Information Fields
    Causal Inference with Information Fields Benjamin Heymann Michel De Lara Criteo AI Lab, Paris, France CERMICS, École des Ponts, Marne-la-Vallée, France [email protected] [email protected] Jean-Philippe Chancelier CERMICS, École des Ponts, Marne-la-Vallée, France [email protected] Abstract Inferring the potential consequences of an unobserved event is a fundamental scientific question. To this end, Pearl’s celebrated do-calculus provides a set of inference rules to derive an interventional probability from an observational one. In this framework, the primitive causal relations are encoded as functional dependencies in a Structural Causal Model (SCM), which maps into a Directed Acyclic Graph (DAG) in the absence of cycles. In this paper, by contrast, we capture causality without reference to graphs or functional dependencies, but with information fields. The three rules of do-calculus reduce to a unique sufficient condition for conditional independence: the topological separation, which presents some theoretical and practical advantages over the d-separation. With this unique rule, we can deal with systems that cannot be represented with DAGs, for instance systems with cycles and/or ‘spurious’ edges. We provide an example that cannot be handled – to the extent of our knowledge – with the tools of the current literature. 1 Introduction As the world shifts toward more and more data-driven decision-making, causal inference is taking more space in applied sciences, statistics and machine learning. This is because it allows for better, more robust decision-making, and provides a way to interpret the data that goes beyond correlation Pearl and Mackenzie [2018].
    [Show full text]
  • Week 2: Causal Inference
    Week 2: Causal Inference Marcelo Coca Perraillon University of Colorado Anschutz Medical Campus Health Services Research Methods I HSMP 7607 2020 These slides are part of a forthcoming book to be published by Cambridge University Press. For more information, go to perraillon.com/PLH. This material is copyrighted. Please see the entire copyright notice on the book's website. 1 Outline Correlation and causation Potential outcomes and counterfactuals Defining causal effects The fundamental problem of causal inference Solving the fundamental problem of causal inference: a) Randomization b) Statistical adjustment c) Other methods The ignorable treatment assignment assumption Stable Unit Treatment Value Assumption (SUTVA) Assignment mechanism 2 Big picture We are going to review the basic framework for understanding causal inference This is a fairly new area of research, although some of the statistical methods have been used for over a century The \new" part is the development of a mathematical notation and framework to understand and define causal effects This new framework has many advantages over the traditional way of understanding causality. For me, the biggest advantage is that we can talk about causal effects without having to use a specific statistical model (design versus estimation) In econometrics, the usual way of understanding causal inference is linked to the linear model (OLS, \general" linear model) { the zero conditional mean assumption in Wooldridge: whether the additive error term in the linear model, i , is correlated with the
    [Show full text]
  • Causal Inference and Data Fusion in Econometrics∗
    TECHNICAL REPORT R-51 First version: Dec, 2019 Last revision: March, 2021 Causal Inference and Data Fusion in Econometrics∗ Paul Hunermund¨ Elias Bareinboim Copenhagen Business School Columbia University [email protected] [email protected] Learning about cause and effect is arguably the main goal in applied economet- rics. In practice, the validity of these causal inferences is contingent on a number of critical assumptions regarding the type of data that has been collected and the substantive knowledge that is available about the phenomenon under investiga- tion. For instance, unobserved confounding factors threaten the internal validity of estimates, data availability is often limited to non-random, selection-biased sam- ples, causal effects need to be learned from surrogate experiments with imperfect compliance, and causal knowledge has to be extrapolated across structurally hetero- geneous populations. A powerful causal inference framework is required in order to tackle all of these challenges, which plague essentially any data analysis to varying degrees. Building on the structural approach to causality introduced by Haavelmo (1943) and the graph-theoretic framework proposed by Pearl (1995), the artificial intelligence (AI) literature has developed a wide array of techniques for causal learn- ing that allow to leverage information from various imperfect, heterogeneous, and biased data sources (Bareinboim and Pearl, 2016). In this paper, we discuss recent advances made in this literature that have the potential to contribute to econo- metric methodology along three broad dimensions. First, they provide a unified and comprehensive framework for causal inference, in which the above-mentioned problems can be addressed in full generality.
    [Show full text]
  • Path Analysis in Mplus: a Tutorial Using a Conceptual Model of Psychological and Behavioral Antecedents of Bulimic Symptoms in Young Adults
    ¦ 2019 Vol. 15 no. 1 Path ANALYSIS IN Mplus: A TUTORIAL USING A CONCEPTUAL MODEL OF PSYCHOLOGICAL AND BEHAVIORAL ANTECEDENTS OF BULIMIC SYMPTOMS IN YOUNG ADULTS Kheana Barbeau a, B, KaYLA Boileau A, FATIMA Sarr A & KEVIN Smith A A School OF Psychology, University OF Ottawa AbstrACT ACTING Editor Path ANALYSIS IS A WIDELY USED MULTIVARIATE TECHNIQUE TO CONSTRUCT CONCEPTUAL MODELS OF psychological, cognitive, AND BEHAVIORAL phenomena. Although CAUSATION CANNOT BE INFERRED BY Roland PfiSTER (Uni- VERSIT¨AT Wurzburg)¨ USING THIS technique, RESEARCHERS UTILIZE PATH ANALYSIS TO PORTRAY POSSIBLE CAUSAL LINKAGES BETWEEN Reviewers OBSERVABLE CONSTRUCTS IN ORDER TO BETTER UNDERSTAND THE PROCESSES AND MECHANISMS BEHIND A GIVEN phenomenon. The OBJECTIVES OF THIS TUTORIAL ARE TO PROVIDE AN OVERVIEW OF PATH analysis, step-by- One ANONYMOUS re- VIEWER STEP INSTRUCTIONS FOR CONDUCTING PATH ANALYSIS IN Mplus, AND GUIDELINES AND TOOLS FOR RESEARCHERS TO CONSTRUCT AND EVALUATE PATH MODELS IN THEIR RESPECTIVE fiELDS OF STUDY. These OBJECTIVES WILL BE MET BY USING DATA TO GENERATE A CONCEPTUAL MODEL OF THE psychological, cognitive, AND BEHAVIORAL ANTECEDENTS OF BULIMIC SYMPTOMS IN A SAMPLE OF YOUNG adults. KEYWORDS TOOLS Path analysis; Tutorial; Eating Disorders. Mplus. B [email protected] KB: 0000-0001-5596-6214; KB: 0000-0002-1132-5789; FS:NA; KS:NA 10.20982/tqmp.15.1.p038 Introduction QUESTIONS AND IS COMMONLY EMPLOYED FOR GENERATING AND EVALUATING CAUSAL models. This TUTORIAL AIMS TO PROVIDE ba- Path ANALYSIS (also KNOWN AS CAUSAL modeling) IS A multi- SIC KNOWLEDGE IN EMPLOYING AND INTERPRETING PATH models, VARIATE STATISTICAL TECHNIQUE THAT IS OFTEN USED TO DETERMINE GUIDELINES FOR CREATING PATH models, AND UTILIZING Mplus TO IF A PARTICULAR A PRIORI CAUSAL MODEL fiTS A RESEARCHER’S DATA CONDUCT PATH analysis.
    [Show full text]
  • STATS 361: Causal Inference
    STATS 361: Causal Inference Stefan Wager Stanford University Spring 2020 Contents 1 Randomized Controlled Trials 2 2 Unconfoundedness and the Propensity Score 9 3 Efficient Treatment Effect Estimation via Augmented IPW 18 4 Estimating Treatment Heterogeneity 27 5 Regression Discontinuity Designs 35 6 Finite Sample Inference in RDDs 43 7 Balancing Estimators 52 8 Methods for Panel Data 61 9 Instrumental Variables Regression 68 10 Local Average Treatment Effects 74 11 Policy Learning 83 12 Evaluating Dynamic Policies 91 13 Structural Equation Modeling 99 14 Adaptive Experiments 107 1 Lecture 1 Randomized Controlled Trials Randomized controlled trials (RCTs) form the foundation of statistical causal inference. When available, evidence drawn from RCTs is often considered gold statistical evidence; and even when RCTs cannot be run for ethical or practical reasons, the quality of observational studies is often assessed in terms of how well the observational study approximates an RCT. Today's lecture is about estimation of average treatment effects in RCTs in terms of the potential outcomes model, and discusses the role of regression adjustments for causal effect estimation. The average treatment effect is iden- tified entirely via randomization (or, by design of the experiment). Regression adjustments may be used to decrease variance, but regression modeling plays no role in defining the average treatment effect. The average treatment effect We define the causal effect of a treatment via potential outcomes. For a binary treatment w 2 f0; 1g, we define potential outcomes Yi(1) and Yi(0) corresponding to the outcome the i-th subject would have experienced had they respectively received the treatment or not.
    [Show full text]
  • Basic Concepts of Statistical Inference for Causal Effects in Experiments and Observational Studies
    Basic Concepts of Statistical Inference for Causal Effects in Experiments and Observational Studies Donald B. Rubin Department of Statistics Harvard University The following material is a summary of the course materials used in Quantitative Reasoning (QR) 33, taught by Donald B. Rubin at Harvard University. Prepared with assistance from Samantha Cook, Elizabeth Stuart, and Jim Greiner. c 2004, Donald B. Rubin Last update: 22 August 2005 1 0-1.2 The perspective on causal inference taken in this course is often referred to as the “Rubin Causal Model” (e.g., Holland, 1986) to distinguish it from other commonly used perspectives such as those based on regression or relative risk models. Three primary features distinguish the Rubin Causal Model: 1. Potential outcomes define causal effects in all cases: randomized experiments and observational studies Break from the tradition before the 1970’s • Key assumptions, such as stability (SUTVA) can be stated formally • 2. Model for the assignment mechanism must be explicated for all forms of inference Assignment mechanism process for creating missing data in the potential outcomes • Allows possible dependence of process on potential outcomes, i.e., confounded designs • Randomized experiments a special case whose benefits for causal inference can be formally stated • 3. The framework allows specification of a joint distribution of the potential outcomes Framework can thus accommodate both assignment-mechanism-based (randomization-based or design- • based) methods and predictive (model-based or Bayesian) methods of causal inference One unified perspective for distinct methods of causal inference instead of two separate perspectives, • one traditionally used for randomized experiment, the other traditionally used for observational studies Creates a firm foundation for methods for dealing with complications such as noncompliance and dropout, • which are especially flexible from a Bayesian perspective 2 0-1.1 Basic Concepts of Statistical Inference for Causal Effects in Experiments and Observational Studies I.
    [Show full text]
  • Causal and Path Model Analysis of the Organizational Characteristics of Civil Defense System Simon Wen-Lon Tai Iowa State University
    Iowa State University Capstones, Theses and Retrospective Theses and Dissertations Dissertations 1972 Causal and path model analysis of the organizational characteristics of civil defense system Simon Wen-Lon Tai Iowa State University Follow this and additional works at: https://lib.dr.iastate.edu/rtd Part of the Sociology Commons Recommended Citation Tai, Simon Wen-Lon, "Causal and path model analysis of the organizational characteristics of civil defense system " (1972). Retrospective Theses and Dissertations. 5280. https://lib.dr.iastate.edu/rtd/5280 This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. INFORMATION TO USERS This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted. The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image.
    [Show full text]
  • Generalized Correlations and Kernel Causality Using R Package Generalcorr
    Generalized Correlations and Kernel Causality Using R Package generalCorr Hrishikesh D. Vinod* October 3, 2017 Abstract Karl Pearson developed the correlation coefficient r(X,Y) in 1890's. Vinod (2014) develops new generalized correlation coefficients so that when r∗(Y X) > r∗(X Y) then X is the \kernel cause" of Y. Vinod (2015a) arguesj that kernelj causality amounts to model selection be- tween two kernel regressions, E(Y X) = g1(X) and E(X Y) = g2(Y) and reports simulations favoring kernelj causality. An Rj software pack- age called `generalCorr' (at www.r-project.org) computes general- ized correlations, partial correlations and plausible causal paths. This paper describes various R functions in the package, using examples to describe them. We are proposing an alternative quantification to extensive causality apparatus of Pearl (2010) and additive-noise type methods in Mooij et al. (2014), who seem to offer no R implementa- tions. My methods applied to certain public benchmark data report a 70-75% success rate. We also describe how to use the package to assess endogeneity of regressors. Keywords: generalized measure of correlation, non-parametric regres- sion, partial correlation, observational data, endogeneity. *Vinod: Professor of Economics, Fordham University, Bronx, New York, USA 104 58. E-mail: [email protected]. A version of this paper was presented at American Statistical Association's Conference on Statistical Practice (CSP) on Feb. 19, 2016, in San Diego, California, and also at the 11-th Greater New York Metro Area Econometrics Colloquium, Johns Hopkins University, Baltimore, Maryland, on March 5, 2016. I thank Dr Pierre Marie Windal of Paris for pointing out an error in partial correlation coefficients.
    [Show full text]
  • Elements of Causal Inference
    Elements of Causal Inference Foundations and Learning Algorithms Adaptive Computation and Machine Learning Francis Bach, Editor Christopher Bishop, David Heckerman, Michael Jordan, and Michael Kearns, As- sociate Editors A complete list of books published in The Adaptive Computation and Machine Learning series appears at the back of this book. Elements of Causal Inference Foundations and Learning Algorithms Jonas Peters, Dominik Janzing, and Bernhard Scholkopf¨ The MIT Press Cambridge, Massachusetts London, England c 2017 Massachusetts Institute of Technology This work is licensed to the public under a Creative Commons Attribution- Non- Commercial-NoDerivatives 4.0 license (international): http://creativecommons.org/licenses/by-nc-nd/4.0/ All rights reserved except as licensed pursuant to the Creative Commons license identified above. Any reproduction or other use not licensed as above, by any electronic or mechanical means (including but not limited to photocopying, public distribution, online display, and digital information storage and retrieval) requires permission in writing from the publisher. This book was set in LaTeX by the authors. Printed and bound in the United States of America. Library of Congress Cataloging-in-Publication Data Names: Peters, Jonas. j Janzing, Dominik. j Scholkopf,¨ Bernhard. Title: Elements of causal inference : foundations and learning algorithms / Jonas Peters, Dominik Janzing, and Bernhard Scholkopf.¨ Description: Cambridge, MA : MIT Press, 2017. j Series: Adaptive computation and machine learning series
    [Show full text]
  • The Book of Why Presentation Transcription
    “The Book of Why” For Epidemiologists Prof. George Davey Smith and Prof. Judea Pearl UCLA May 29, 2019 Start: 3:20 Finish: 46:07 Thanks again for the invite to UCLA. I would first like to ask, and you’ve got to not be influenced by the fact that Judea is here... I want to see who have got ‘The Book of Why’? So that is the majority of people. Keep those hands up, people are holding them up .. they want to get it signed. Who has read any of it? Who has read it all the way through? Okay so, some hands have gone down. Okay, so I have structured this around talking about ’The Book of Why’ from the perspective of an empirical researcher. Also I would imagine some of you who haven’t read ‘The Book of Why’, students might also know the primer that Judea and colleagues wrote as well. So if you have read either of them then this is fine. (Slide 2) I’m going to outline as I see it the value of DAG theory to epidemiology. I’m also going to say a bit about the reality of use of DAGs in epidemiology, some of which has come from my experience of having to read a very large number of papers as the editor of the International Journal of Epidemiology. We were getting 1000 papers a year by the time I was worn out, well Shah Ebrahim and I (there were two of us co-editing it, with a team including Karen Michaels as associate editors).
    [Show full text]
  • Tutorial: Causal Model Search
    Tutorial: Causal Model Search Richard Scheines Carnegie Mellon University Peter Spirtes, Clark Glymour, Joe Ramsey, others 1 Goals 1) Convey rudiments of graphical causal models 2) Basic working knowledge of Tetrad IV 2 Tetrad: Complete Causal Modeling Tool 3 Tetrad 1) Main website: http://www.phil.cmu.edu/projects/tetrad/ 2) Download: http://www.phil.cmu.edu/projects/tetrad/current.html 3) Data files: workshop.new.files.zip in www.phil.cmu.edu/projects/tetrad_download/download/workshop/Data/ 4) Download from Data directory: • tw.txt • Charity.txt • Optional: • estimation1.tet, estimation2.tet • search1.tet, search2.tet, search3.tet 4 Outline 1) Motivation 2) Representing/Modeling Causal Systems 3) Estimation and Model fit 4) Causal Model Search 5 Statistical Causal Models: Goals 1) Policy, Law, and Science: How can we use data to answer a) subjunctive questions (effects of future policy interventions), or b) counterfactual questions (what would have happened had things been done differently (law)? c) scientific questions (what mechanisms run the world) 2) Rumsfeld Problem: Do we know what we do and don’t know: Can we tell when there is or is not enough information in the data to answer causal questions? 6 Causal Inference Requires More than Probability Prediction from Observation ≠ Prediction from Intervention P(Lung Cancer 1960 = y | Tar-stained fingers 1950 = no) ≠ P(Lung Cancer 1960 = y | Tar-stained fingers 1950set = no) In general: P(Y=y | X=x, Z=z) ≠ P(Y=y | Xset=x, Z=z) Causal Prediction vs. Statistical Prediction: Non-experimental data P(Y,X,Z) P(Y=y | X=x, Z=z) (observational study) P(Y=y | X =x, Z=z) Background Knowledge Causal Structure set 7 Causal Search Causal Search: 1.
    [Show full text]
  • “Structural Vs. Reduced Form” Language
    \Structural vs. Reduced Form" Language, Confusion, and Models in Empirical Economics Phil Haile November 2020 1 Types of Empirical Economics? 1. Measurement I how much do prices change with the number of competitors? I how has productivity (e.g., TFP) in the U.S. auto industry changed over the last 30 years? I what is the effect of college attendance on expected wage? I what is the elasticity of aggreg demand for health insurance? 2. Model Testing I Is there moral hazard in auto insurance? I Does BNE do well in predicting bidding at oil auctions? I Do actual contracts resemble optimal contracts? 3. Model Estimation for Counterfactuals I How much would prices rise two firms merged? if the sales tax increased by 1%? I How would student outcomes differ under a different school choice mechanism? 2 Measurement vs. \Model Estimation for Counterfactuals" This distinction is usually false: Most \measurement" questions in economics concern a counterfactual of the form \how would the world have been different if XXX were changed, all else fixed?” • effect of entry on prices: we can't just compare markets with differing numbers of competitors • productivity: we can't just look at inputs and outputs to learn the production function (and thus TFP), since unobservables affect both the input levels chosen by firms and their realized output • returns to college: we can't just compare wages for those who attended and those who did not • elasticity of demand: we can't just compare quantities purchased for goods with high prices vs. those with low prices. 3 Causality and Counterfactuals Much of empirical economics is about \causal effects” (what we once just called “effects”) Causality is a concept defined by a counterfactual question of the form \what would happen if certain things were changed while others were held fixed?" This is true even for an RCT.
    [Show full text]