Muutke küpsiste eelistusi

Data Analytics for the Social Sciences: Applications in R [Kõva köide]

(North Carolina State University, Raleigh, USA)
  • Formaat: Hardback, 686 pages, kõrgus x laius: 280x210 mm, kaal: 2100 g, 163 Halftones, color; 163 Illustrations, color
  • Ilmumisaeg: 30-Nov-2021
  • Kirjastus: Routledge
  • ISBN-10: 036762429X
  • ISBN-13: 9780367624293
Teised raamatud teemal:
Data Analytics for the Social Sciences: Applications in RSuurem pilt
  • Formaat: Hardback, 686 pages, kõrgus x laius: 280x210 mm, kaal: 2100 g, 163 Halftones, color; 163 Illustrations, color
  • Ilmumisaeg: 30-Nov-2021
  • Kirjastus: Routledge
  • ISBN-10: 036762429X
  • ISBN-13: 9780367624293
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9780367624293.html
Märksõnad:
This book presents a complete exploration of statistical data analysis in R for a wide variety of social science disciplines and quantitative methods courses.

Data Analytics for the Social Sciences is an introductory, graduate-level treatment of data analytics for social science. It features applications in the R language, arguably the fastest growing and leading statistical tool for researchers.

The book starts with an ethics chapter on the uses and potential abuses of data analytics. Chapters 2 and 3 show how to implement a broad range of statistical procedures in R. Chapters 4 and 5 deal with regression and classification trees and with random forests. Chapter 6 deals with machine learning models and the "caret" package, which makes available to the researcher hundreds of models. Chapter 7 deals with neural network analysis and Chapter 8 with network analysis and visualization of network data. A final chapter treats text analysis, including web scraping, comparative word frequency tables, word clouds, word maps, sentiment analysis, topic analysis, and more. All empirical chapters have two "Quick Start" exercises designed to allow quick immersion in chapter topics, followed by "In Depth" coverage. Data are available for all examples and runnable R code is provided in a "Command Summary". An appendix provides an extended tutorial on R and RStudio. Over 30 online supplements for each chapter provide "books within the book" on a variety of topics, such as agent-based modelling.

Rather than focusing on equations, derivations and proofs, this book emphasises hands-on obtaining of output for various social science models and on how to interpret the output. It is suitable for all advanced level undergraduate and postgraduate students learning statistical data analysis.

Acknowledgments xvi
Preface xvii
1 Using and abusing data analytics in social science 1(21)
1.1 Introduction
1(2)
1.2 The promise of data analytics for social science
3(1)
1.2.1 Data analytics in public affairs and public policy
3(1)
1.2.2 Data analytics in the social sciences
3(1)
1.2.3 Data analytics in the humanities
4(1)
1.3 Research design issues in data analytics
4(6)
1.3.1 Beware the true believer
4(1)
1.3.2 Pseudo-objectivity in data analytics
4(1)
1.3.3 The bias of scholarship based on algorithms using big data
5(3)
1.3.4 The subjectivity of algorithms
8(1)
1.3.5 Big data and big noise
9(1)
1.3.6 Limitations of the leading data science dissemination models
9(1)
1.4 Social and ethical issues in data analytics
10(9)
1.4.1 Types of ethical issues in data analytics
10(1)
1.4.2 Bias toward the privileged
11(1)
1.4.3 Discrimination
12(1)
1.4.4 Diversity and data analytics
13(1)
1.4.5 Distortion of democratic processes
14(1)
1.4.6 Undermining of professional ethics
14(1)
1.4.7 Privacy, profiling, and surveillance issues
15(3)
1.4.8 The transparency issue
18(1)
1.5 Summary: Technology and power
19(2)
Endnotes
21(1)
2 Statistical analytics with R, Part 1 22(69)
Part I: Overview Of Statistical Analysis With R
22(2)
2.1 Introduction
22(1)
2.2 Data and packages used in this chapter
22(2)
2.2.1 Example data
22(1)
2.2.2 R packages used
23(1)
Part II: Quick Start On Statistical Analysis With R
24(9)
2.3 Descriptive statistics
24(2)
2.4 Linear multiple regression
26(7)
Part III: Statistical Analysis With R In Detail
33(58)
2.5 Hypothesis testing
33(3)
2.5.1 One-sample test of means
34(1)
2.5.2 Means test for two independent samples
35(1)
2.5.3 Means test for two dependent samples
35(1)
2.6 Crosstabulation, significance, and association
36(2)
2.7 Loglinear analysis for categorical variables
38(1)
2.8 Correlation, correlograms, and scatterplots
38(5)
2.9 Factor analysis (exploratory)
43(1)
2.10 Multidimensional scaling
44(1)
2.11 Reliability analysis
44(5)
2.11.1 Cronbach's alpha and Guttman's lower bounds
46(1)
2.11.2 Guttman's lower bounds and Cronbach's alpha
46(2)
2.11.3 Krippendorff's alpha and Cohen's kappa
48(1)
2.12 Cluster analysis
49(11)
2.12.1 Hierarchical cluster analysis
50(1)
2.12.2 K-means clustering
50(9)
2.12.3 Nearest neighbor analysis
59(1)
2.13 Analysis of variance
60(13)
2.13.1 Data and packages used
60(1)
2.13.2 GLM univariate: ANOVA
61(5)
2.13.3 GLM univariate: ANCOVA
66(1)
2.13.4 GLM multivariate: MANOVA
67(3)
2.13.5 GLM multivariate: MANCOVA
70(3)
2.14 Logistic regression
73(6)
2.14.1 ROC and AUC analysis
77(1)
2.14.2 Confusion table and accuracy
77(2)
2.15 Mediation and moderation
79(10)
2.16
Chapter 2 command summary
89(1)
Endnotes
89(2)
3 Statistical analytics with R, Part 2 91(45)
Part I: Overview Of Statistical Analytics With R
91(1)
3.1 Introduction
91(1)
3.2 Data and packages used in this chapter
91(1)
3.2.1 Example data
91(1)
3.2.2 R Packages used
92(1)
Part II: Quick Start On Statistical Analysis Part 2
92(9)
3.3 Quick start: Linear regression as a generalized linear modeling (GZLM)
92(7)
3.3.1 Background to GZLM
92(1)
3.3.2 The linear model in glm()
92(1)
3.3.3 GZLM output
93(1)
3.3.4 Fitted value, residuals, and plots
94(3)
3.3.5 Noncanonical custom links
97(1)
3.3.6 Multiple comparison tests
98(1)
3.3.7 Estimated marginal means (EMM)
98(1)
3.4 Quick start: Testing if multilevel modeling is needed
99(2)
Part III: Statistical Analysis, Part 2, In Detail
101(35)
3.5 Generalized linear models (GZLM)
101(14)
3.5.1 Introduction
101(2)
3.5.2 Setup for GZLM models in R
103(1)
3.5.3 Binary logistic regression example
104(1)
3.5.4 Gamma regression model
105(3)
3.5.5 Poisson regression model
108(5)
3.5.6 Negative binomial regression
113(2)
3.6 Multilevel modeling (MLM)
115(4)
3.6.1 Introduction
115(1)
3.6.2 Setup and data
115(1)
3.6.3 The random coefficients model
116(3)
3.6.4 Likelihood ratio test
119(1)
3.7 Panel data regression (PDR)
119(15)
3.7.1 Introduction
119(1)
3.7.2 Types of PDR model
120(2)
3.7.3 The Hausman test
122(1)
3.7.4 Setup and data
123(1)
3.7.5 PDR with the plm package
124(9)
3.7.6 PDR with the panelr package
133(1)
3.8 Structural equation modeling (SEM)
134(1)
3.9 Missing data analysis and data imputation
134(1)
3.10
Chapter 3 command summary
134(1)
Endnotes
134(2)
4 Classification and regression trees in R 136(79)
Part I: Overview Of Classification And Regression Trees With R
136(9)
4.1 Introduction
137(1)
4.2 Advantages of decision tree analysis
137(1)
4.3 Limitations of decision tree analysis
138(1)
4.4 Decision tree terminology
139(1)
4.5 Steps in decision tree analysis
140(1)
4.6 Decision tree algorithms
140(2)
4.7 Random forests and ensemble methods
142(1)
4.8 Software
143(1)
4.8.1 R language
143(1)
4.8.2 Stata
144(1)
4.8.3 SAS
144(1)
4.8.4 SPSS
144(1)
4.8.5 Python language
144(1)
4.9 Data and packages used in this chapter
144(1)
4.9.1 Example data
144(1)
4.9.2 R packages used
145(1)
Part II: Quick Start - Classification And Regression Trees
145(7)
4.10 Classification tree example: Survival on the Titanic
145(4)
4.11 Regression tree example: Correlates of murder
149(3)
Part III: Classification And Regression Trees, In Detail
152(63)
4.12 Overview
152(1)
4.13 The rpart() program
153(5)
4.13.1 Introduction
153(2)
4.13.2 Training and validation datasets
155(1)
4.13.3 Setup for rpart() trees
156(2)
4.14 Classification trees with the rpart package
158(31)
4.14.1 The basic rpart classification tree
158(2)
4.14.2 Printing tree rules
160(1)
4.14.3 Visualization with prp() and draw.tree()
161(2)
4.14.4 Visualization with fancyRpartPlot()
163(1)
4.14.5 Interpreting tree summaries
164(5)
4.14.6 Listing nodes by country and countries by node
169(1)
4.14.7 Node distribution plots
170(1)
4.14.8 Saving predictions and residuals
171(2)
4.14.9 Cross-validation and pruning
173(3)
4.14.10 The confusion matrix and model performance metrics
176(6)
4.14.11 The ROC curve and AUC
182(2)
4.14.12 Lift plots
184(2)
4.14.13 Gains plots
186(1)
4.14.14 Precision vs. recall plot
186(3)
4.15 Regression trees with the rpart package
189(23)
4.15.1 Setup
189(1)
4.15.2 Creating an rpart regression tree
189(3)
4.15.3 Printing tree rules
192(1)
4.15.4 Visualization with prp() and fancyRpartPlot()
192(2)
4.15.5 Interpreting tree summaries
194(3)
4.15.6 The CP table
197(1)
4.15.7 Listing nodes by country and countries by node
198(1)
4.15.8 Saving predictions and residuals
199(1)
4.15.9 Plotting residuals
200(1)
4.15.10 Cross-validation and pruning
201(1)
4.15.11 R-squared for regression trees
202(3)
4.15.12 MSE for regression trees
205(1)
4.15.13 The confusion matrix
206(1)
4.15.14 The ROC curve and AUC
206(1)
4.15.15 Gains plots
206(3)
4.15.16 Gains plot with OLS comparison
209(3)
4.16 The tree package
212(1)
4.17 The ctree() program for conditional decision trees
212(1)
4.18 More decision trees programs for R
212(1)
4.19
Chapter 4 command summary
213(1)
Endnotes
213(2)
5 Random forests 215(76)
Part I: Overview Of Random Forests In R
215(3)
5.1 Introduction
215(3)
5.1.1 Social science examples of random forest models
215(1)
5.1.2 Advantages of random forests
216(1)
5.1.3 Limitations of random forests
217(1)
5.1.4 Data and packages
217(1)
Part II: Quick Start - Random Forests
218(8)
5.2 Classification forest example: Searching for the causes of happiness
218(3)
5.3 Regression forest example: Why so much crime in my town?
221(5)
Part III: Random Forests, In Detail
226(65)
5.4 Classification forests with randomForest()
226(27)
5.4.1 Setup
226(1)
5.4.2 A basic classification model
227(3)
5.4.3 Output components of randomForest() objects for classification models
230(8)
5.4.4 Graphing a randomForest tree?
238(1)
5.4.5 Comparing randomForest() and rpart() performance
239(2)
5.4.6 Tuning the random forest model
241(9)
5.4.7 MDS cluster analysis of the RF classification model
250(3)
5.5 Regression forests with randomForest()
253(19)
5.5.1 Introduction
253(1)
5.5.2 Setup
254(1)
5.5.3 A basic regression model
254(2)
5.5.4 Output components for regression forest models
256(4)
5.5.5 Graphing a randomForest tree?
260(1)
5.5.6 MDS plots
260(1)
5.5.7 Quartile plots
261(1)
5.5.8 Comparing randomForest() and rpart() regression models
262(1)
5.5.9 Tuning the randomForest() regression model
263(5)
5.5.10 Outliers: Identifying and removing
268(4)
5.6 The randomForestExplainer package
272(14)
5.6.1 Setup for the randomForestExplainer package
272(1)
5.6.2 Minimal depth plots
273(1)
5.6.3 Multiway variable importance plots
274(3)
5.6.4 Multiway ranking of variable importance
277(1)
5.6.5 Comparing randomForest and OLS rankings of predictors
278(2)
5.6.6 Which importance criteria?
280(1)
5.6.7 Interaction analysis
281(5)
5.6.8 The explain_forest() function
286(1)
5.7 Summary
286(1)
5.8 Conditional inference forests
287(1)
5.9 MDS plots for random forests
287(1)
5.10 More random forest programs for R
287(2)
5.11 Command summary
289(1)
Endnotes
289(2)
6 Modeling and machine learning 291(64)
Part I: Overview Of Modeling And Machine Learning
291(6)
6.1 Introduction
291(6)
6.1.1 Social science examples of modeling and machine learning in R
292(2)
6.1.2 Advantages of modeling and machine learning in R
294(1)
6.1.3 Limitations of modeling and machine learning in R
294(1)
6.1.4 Data, packages, and default directory
295(2)
Part II: Quick Start - Modeling And Machine Learning
297(19)
6.2 Example 1: Bayesian modeling of county-level poverty
297(10)
6.2.1 Introduction
297(1)
6.2.2 Setup
297(1)
6.2.3 Correlation plot
298(2)
6.2.4 The Bayes generalized linear model
300(7)
6.3 Example 2: Predicting diabetes among Pima Indians with mlr3
307(9)
6.3.1 Introduction
307(1)
6.3.2 Setup
307(1)
6.3.3 How mlr3 works
307(2)
6.3.4 The Pima Indian data
309(7)
Part III: Modeling And Machine Learning In Detail
316(39)
6.4 Illustrating modeling and machine learning with SVM in caret
316(4)
6.4.1 How SVM works
317(1)
6.4.2 SVM algorithms compared to logistic and OLS regression
317(1)
6.4.3 SVM kernels, types, and parameters
318(1)
6.4.4 Tuning SVM models
319(1)
6.4.5 SVM and longitudinal data
319(1)
6.5 SVM versus OLS regression
320(1)
6.6 SVM with the caret package: Predicting world literacy rates
320(6)
6.6.1 Setup
321(1)
6.6.2 Constructing the SVM regression model with caret
322(1)
6.6.3 Obtaining predicted values and residuals
323(1)
6.6.4 Model performance metrics
323(1)
6.6.5 Variable importance
324(1)
6.6.6 Other output elements
324(1)
6.6.7 SVM plots
325(1)
6.7 Tuning SVM models
326(7)
6.7.1 Tuning for the train() command from the caret package
327(1)
6.7.2 Tuning for the svm() command from the e1071 package
328(2)
6.7.3 Cross-validating SVM models
330(1)
6.7.4 Using e1071 in caret rather than the default kern package
331(2)
6.8 SVM classification models: Classifying U.S. Senators
333(8)
6.8.1 The "senate" example and setup
333(1)
6.8.2 SVM classification with alternative kernels: Senate example
333(5)
6.8.3 Tuning the SVM binary classification model
338(3)
6.9 Gradient boosting machines (GBM)
341(4)
6.9.1 Introduction
341(1)
6.9.2 Setup and example data
342(1)
6.9.3 Metrics for comparing models
343(1)
6.9.4 The caret control object
343(1)
6.9.5 Training the GBM model under caret
344(1)
6.10 Learning vector quantization (LVQ)
345(2)
6.10.1 Introduction
345(1)
6.10.2 Setup and example data
346(1)
6.10.3 Metrics for comparing models
346(1)
6.10.4 The caret control object
346(1)
6.10.5 Training the LVQ model under caret
346(1)
6.11 Comparing models
347(2)
6.12 Variable importance
349(3)
6.12.1 Leave-one-out modeling
349(1)
6.12.2 Recursive feature elimination (RFE) with caret
350(2)
6.12.3 Other approaches to variable importance
352(1)
6.13 SVM classification for a multinomial outcome
352(1)
6.14 Command summary
352(1)
Endnotes
352(3)
7 Neural network models and deep learning 355(46)
Part I: Overview Of Neural Network Models And Deep Learning
355(9)
7.1 Overview
355(1)
7.2 Data and packages
356(1)
7.3 Social science examples
357(1)
7.4 Pros and cons of neural networks
358(1)
7.5 Artificial neural network (ANN) concepts
359(5)
7.5.1 ANN terms
359(3)
7.5.2 R software programs for ANN
362(1)
7.5.3 Training methods for ANN
363(1)
7.5.4 Algorithms in neuralnet
363(1)
7.5.5 Algorithms in nnet
363(1)
7.5.6 Tuning ANN models
364(1)
Part II: Quick Start - Modeling And Machine Learning
364(11)
7.6 Example 1: Analyzing NYC airline delays
364(6)
7.6.1 Introduction
364(1)
7.6.2 General setup
364(1)
7.6.3 Data preparation
364(1)
7.6.4 Modeling NYC airline delays
365(5)
7.7 Example 2: The classic iris classification example
370(5)
7.7.1 Setup
370(1)
7.7.2 Exploring separation with a violin plot
371(1)
7.7.3 Normalizing the data
371(1)
7.7.4 Training the model with nnet in caret
372(2)
7.7.5 Obtain model predictions
374(1)
7.7.6 Display the neural model
375(1)
Part III: Neural Network Models In Detail
375(26)
7.8 Analyzing Boston crime via the neuralnet package
375(11)
7.8.1 Setup
376(1)
7.8.2 The linear regression model for unscaled data
377(2)
7.8.3 The neuralnet model for unscaled data
379(1)
7.8.4 Scaling the data
379(1)
7.8.5 The linear regression model for scaled data
379(1)
7.8.6 The neuralnet model for scaled data
380(1)
7.8.7 Neuralnet results for the training data
381(1)
7.8.8 Model performance plots
382(1)
7.8.9 Visualizing the neuralnet model
383(1)
7.8.10 Variable importance for the neuralnet model
384(2)
7.9 Analyzing Boston crime via neuralnet under the caret package
386(1)
7.10 Analyzing Boston crime via nnet in caret
386(9)
7.10.1 Setup
387(1)
7.10.2 The nnet/caret model of Boston crime
388(4)
7.10.3 Variable importance for the nnet/caret model
392(1)
7.10.4 Further tuning the nnet model outside caret
393(2)
7.11 A classification model of marital status using nnet
395(5)
7.11.1 Setup
395(2)
7.11.2 The nnet classification model of marital status
397(3)
7.12 Neural network analysis using "mlr3keras"
400(1)
7.13 Command summary
400(1)
Endnotes
400(1)
8 Network analysis 401(102)
Part I: Overview Of Network Analysis With R
401(4)
8.1 Introduction
401(1)
8.2 Data and packages used in this chapter
401(2)
8.3 Concepts in network analysis
403(1)
8.4 Getting data into network format
404(1)
Part II: Quick Start On Network Analysis With R
405(11)
8.5 Quick start exercise 1: The Medici family network
405(4)
8.6 Quick start exercise 2: Marvel hero network communities
409(7)
Part III: Network Analysis With R In Detail
416(87)
8.7 Interactive network analysis with visNetwork
416(13)
8.7.1 Undirected networks: Research team management
417(4)
8.7.2 Clustering by group: Research team grouped by gender
421(1)
8.7.3 A larger network with navigation and circle layout
422(3)
8.7.4 Visualizing classification and regression trees: National literacy
425(1)
8.7.5 A directed network (asymmetrical relationships in a research team)
426(3)
8.8 Network analysis with igraph
429(24)
8.8.1 Term adjacency networks: Gubernatorial websites and the covid pandemic
429(7)
8.8.2 Similarity/distance networks with igraph: Senate interest group ratings
436(4)
8.8.3 Communities, modularity, and centrality
440(7)
8.8.4 Similarity network analysis: All senators
447(6)
8.9 Using intergraph for network conversions
453(4)
8.10 Network-on-a-map with the diagram and maps packages
457(5)
8.11 Network analysis with the statnet and network packages
462(11)
8.11.1 Introduction
462(5)
8.11.2 Visualization
467(3)
8.11.3 Neighborhoods
470(2)
8.11.4 Cluster analysis
472(1)
8.12 Clique analysis with sna
473(8)
8.12.1 A simplified clique analysis
473(2)
8.12.2 A clique analysis of the DHHS formal network
475(6)
8.12.3 K-core analysis of the DHHS formal network
481(1)
8.13 Mapping international trade flow with statnet and Intergraph
481(1)
8.14 Correlation networks with corrr
481(3)
8.15 Network analysis with tidygraph
484(10)
8.15.1 Introduction
484(1)
8.15.2 A simple tidygraph example
484(6)
8.15.3 Network conversions with tidygraph
490(1)
8.15.4 Finding community clusters with tidygraph
491(3)
8.16 Simulating networks
494(6)
8.16.1 Agent-based network modeling with SchellingR
494(5)
8.16.2 Agent-based network modeling with RSiena
499(1)
8.16.3 Agent-based network modeling with NetLogoR
499(1)
8.17 Summary
500(1)
8.18 Command summary
501(1)
Endnotes
501(2)
9 Text analytics 503(110)
Part I: Overview Of Text Analytics With R
503(13)
9.1 Overview
503(1)
9.2 Data used in this chapter
503(1)
9.3 Packages used in this chapter
504(1)
9.4 What is a corpus?
505(1)
9.5 Text files
505(11)
9.5.1 Overview
505(1)
9.5.2 Archived texts
505(1)
9.5.3 Project Gutenberg archive
506(3)
9.5.4 Comma-separated values (.csv) files
509(1)
9.5.5 Text from Word .docx files with the textreadr package
509(3)
9.5.6 Text from other formats with the readtext package
512(2)
9.5.7 Text from raw text files
514(2)
Part II: Quick Start On Text Analytics With R
516(7)
9.6 Quick start exercise 1: Key word in context (kwic) indexing
516(2)
9.7 Quick start exercise 2: Word frequencies and histograms
518(5)
Part III: Network Analysis With R In Detail
523(90)
9.8 Web scraping
523(8)
9.8.1 Overview
523(1)
9.8.2 Web scraping: The "htm2txt" package
524(3)
9.8.3 Web scraping: The "rvest" package
527(4)
9.9 Social media scraping
531(8)
9.9.1 Analysis of Twitter data: Trump and the New York Times
532(4)
9.9.2 Social media scraping with twitter
536(3)
9.10 Leading text formats in R
539(15)
9.10.1 Overview
539(1)
9.10.2 Formats related to the "tidytext" package
540(3)
9.10.3 Formats related to the "tm" package
543(4)
9.10.4 Formats related to the "quanteda" package
547(5)
9.10.5 Common text file conversions
552(2)
9.11 Tokenization
554(3)
9.11.1 Overview
554(1)
9.11.2 Word tokenization
554(3)
9.12 Character encoding
557(2)
9.13 Text cleaning and preparation
559(1)
9.14 Analysis: Multigroup word frequency comparisons
559(8)
9.14.1 Multigroup analysis in tidytext
559(4)
9.14.2 Multigroup analysis with quanteda's textstat_keyness() command
563(3)
9.14.3 Multigroup analysis with textstat frequency() in quanteda and ggplot2
566(1)
9.15 Analysis: Word clouds
567(5)
9.16 Analysis: Comparison clouds
572(2)
9.17 Analysis: Word maps and word correlations
574(13)
9.17.1 Working with the tdm format
574(1)
9.17.2 Working with the dtm format
575(1)
9.17.3 Word frequencies and word correlations
576(1)
9.17.4 Correlation plots of word and document associations
577(4)
9.17.5 Plotting word stem correlations for word pairs
581(3)
9.17.6 Word correlation maps
584(3)
9.18 Analysis: Sentiment analysis
587(9)
9.18.1 Overview
587(1)
9.18.2 Example: Sentiment analysis of news articles
587(9)
9.19 Analysis: Topic modeling
596(14)
9.19.1 Overview
596(1)
9.19.2 Topic analysis example 1: Modeling topic frequency over time
597(6)
9.19.3 Topic analysis example 2: LDA analysis
603(7)
9.20 Analysis: Lexical dispersion plots
610(1)
9.21 Analysis: Bigrams and ngrams
611(1)
9.22 Command summary
612(1)
Endnotes
612(1)
Appendix 1: Introduction to R and RStudio 613(45)
Appendix 2: Data used in this book 658(10)
References 668(10)
Index 678
G. David Garson teaches advanced research methodology in the School of Public and International Affairs, North Carolina State University, USA. Founder and longtime editor emeritus of the Social Science Computer Review, he is president of Statistical Associates Publishing, which provides free digital texts worldwide. His degrees are from Princeton University (BA, 1965) and Harvard University (PhD, 1969).