book

Behavioral Data Analysis with R and Python

by Florent Buisson

June 2021

Beginner to intermediate

358 pages

10h

English

O'Reilly Media, Inc.

Read now

Unlock full access

Who This Book Is ForWho This Book Is Not ForR and Python CodeCode EnvironmentsCode ConventionsFunctional-Style Programming 101Using Code ExamplesNavigating This BookConventions Used in This BookO’Reilly Online LearningHow to Contact UsAcknowledgments
Why We Need Causal Analytics to Explain Human BehaviorThe Different Types of AnalyticsHuman Beings Are ComplicatedConfound It! The Hidden Dangers of Letting Regression Sort It OutDataWhy Correlation Is Not Causation: A Confounder in ActionToo Many Variables Can Spoil the BrothConclusion
A Basic Model of Human BehaviorPersonal CharacteristicsCognition and EmotionsIntentionsActionsBusiness BehaviorsHow to Connect Behaviors and DataDevelop a Behavioral Integrity MindsetDistrust and VerifyIdentify the CategoryRefine Behavioral VariablesUnderstand the ContextConclusion
Causal Diagrams and the Causal-Behavioral FrameworkCausal Diagrams Represent BehaviorsCausal Diagrams Represent DataFundamental Structures of Causal DiagramsChainsForksCollidersCommon Transformations of Causal DiagramsSlicing/Disaggregating VariablesAggregating VariablesWhat About Cycles?PathsConclusion
Business Problem and Data SetupData and PackagesUnderstanding the Relationship of InterestIdentify Candidate Variables to IncludeActionsIntentionsCognition and EmotionsPersonal CharacteristicsBusiness BehaviorsTime TrendsValidate Observable Variables to Include Based on DataRelationships Between Numeric VariablesRelationships Between Categorical VariablesRelationships Between Numeric and Categorical VariablesExpand Causal Diagram IterativelyIdentify Proxies for Unobserved VariablesIdentify Further CausesIterateSimplify Causal DiagramConclusion
Business Problem: Ice Cream and Bottled Water SalesThe Disjunctive Cause CriterionDefinitionFirst BlockSecond BlockThe Backdoor CriterionDefinitionsFirst BlockSecond BlockConclusion
Data and PackagesVisualizing Missing DataAmount of Missing DataCorrelation of MissingnessDiagnosing Missing DataCauses of Missingness: Rubin’s ClassificationDiagnosing MCAR VariablesDiagnosing MAR VariablesDiagnosing MNAR VariablesMissingness as a SpectrumHandling Missing DataIntroduction to Multiple Imputation (MI)Default Imputation Method: Predictive Mean MatchingFrom PMM to Normal Imputation (R Only)Adding Auxiliary VariablesScaling Up the Number of Imputed Data SetsConclusion

Intro to the Bootstrap: “Polling” Oneself UpPackagesThe Business Problem: Small Data with an OutlierBootstrap Confidence Interval for the Sample MeanBootstrap Confidence Intervals for Ad Hoc StatisticsThe Bootstrap for Regression AnalysisWhen to Use the BootstrapConditions for the Traditional Central Estimate to Be SufficientConditions for the Traditional CI to Be SufficientDetermining the Number of Bootstrap SamplesOptimizing the Bootstrap in R and PythonR: The BehavioralDataAnalysis PackagePython OptimizationConclusion
Planning the Experiment: Theory of ChangeBusiness Goal and Target MetricInterventionBehavioral LogicData and PackagesDetermining Random Assignment and Sample Size/PowerRandom AssignmentSample Size and Power AnalysisAnalyzing and Interpreting Experimental ResultsConclusion
Planning the ExperimentBusiness Goal and Target MetricDefinition of the InterventionBehavioral LogicData and PackagesDetermining Random Assignment and Sample Size/PowerRandom AssignmentPower Analysis with Bootstrap SimulationsAnalyzing and Interpreting Experimental ResultsIntention-to-Treat Estimate for Encouragement InterventionComplier Average Causal Estimate for Mandatory InterventionConclusion
Planning the ExperimentBusiness Goal and Target MetricDefinition of the InterventionBehavioral LogicData and PackagesIntroduction to Hierarchical ModelingR CodePython CodeDetermining Random Assignment and Sample Size/PowerRandom AssignmentPower AnalysisAnalyzing the ExperimentConclusion
Data and PackagesBehavioral Varieties of ModerationSegmentationInteractionsNonlinearitiesHow to Apply ModerationWhen to Look for Moderation?Multiple ModeratorsValidating Moderation with BootstrapInterpreting Individual CoefficientsConclusion
MediationUnderstanding Causal MechanismsCausal BiasesIdentifying MediationMeasuring MediationInstrumental VariablesDataPackagesUnderstanding and Applying IVsMeasurementApplying IVs: Frequently Asked QuestionsConclusion

Content preview from Behavioral Data Analysis with R and Python

Bibliography

Aberson, Christopher L. Applied Power Analysis for the Behavioral Sciences. Abingdon, UK: Routledge, 2019.
Angrist, Joshua D., and Jörn-Steffen Pischke. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton, NJ: Princeton University Press, 2009. A go-to reference for applied econometrics, with graduate-level math and statistics.
Angrist, Joshua D., and Jörn-Steffen Pischke. Mastering “Metrics”: The Path from Cause to Effect. Princeton, NJ: Princeton University Press, 2014. A more accessible version of their 2009 classic, with added Kung Fu Panda references!
Antonio, Nuno, Ana de Almeida, and Luis Nunes. “Hotel booking demand datasets.” Data in Brief 22 (Feb. 2019): 41-49. https://doi.org/10.1016/j.dib.2018.11.126.
Bertrand, Marianne, and Sendhil Mullainathan. “Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination.” American Economic Review 94, no. 4 (2004): 991-1013.
Cohen, Jacob. Statistical Power Analysis for the Behavioral Sciences, 2nd Edition. Abingdon, UK: Routledge, 2013. A classic on power analysis. Like most classics, it is somewhat dated on computer implementation, but still a great resource to think more deeply about what statistical power is.
Cunningham, Scott. Causal Inference: The Mixtape. New Haven, CT: Yale University Press, 2021. A very accessible variation on the Angrist/Pischke books, still targeted primarily to academics.
Davison, A. C., and D. V. Hinkley. Bootstrap ...