My first article to be published in a statistics journal appeared in June 1990. I recently happened to look at the table of contents of that issue. It’s a fascinating time capsule. Maybe not so fascinating to you, but compelling to me, in that I recognize so many of the names, also the topics induce some nostalgia:
Applications and Case Studies
Estimating the Electoral Consequences of Legislative Redistricting (pp. 274-282)
Andrew Gelman and Gary KingEstimating Fecundability From Data on Waiting Times to First Conception (pp. 283-294)
James J. Heckman and James R. WalkerThe Deterrent Effect of Capital Punishment: An Analysis of Daily Homicide Counts (pp. 295-303)
Jeffrey GroggerInference from Coarse Data Via Multiple Imputation with Application to Age Heaping (pp. 304-314)
Daniel F. Heitjan and Donald B. RubinMultinomial Runs Tests to Detect Clustering in Constrained Free Recall (pp. 315-320)
Gail Rubin, Charles E. McCulloch and Michael A. ShapiroEvaluating Screening for the Early Detection and Treatment of Cancer Without Using a Randomized Control Group (pp. 321-327)
Stuart G. Baker and Kenneth C. ChuModeling and Monitoring Biomedical Times Series (pp. 328-337)
K. Gordon and A. F. M. SmithAn Application of the Seasonal Fractionally Differenced Model to the Monetary Aggregates (pp. 338-344)
Susan Porter-HudakSliding-Spans Diagnostics for Seasonal and Related Adjustments (pp. 345-355)
David F. Findley, Brian C. Monsell, Holly B. Shulman and Marian G. PughHomophily and Social Distance in the Choice of Multiple Friends: An Analysis Based on Conditionally Symmetric Log-Bilinear Association Model (pp. 356-366)
Kazuo YamaguchiThe Relationship Between the Length of the Base Period and Population Forecast Errors (pp. 367-375)
Stanley K. Smith and Terry SincichTheory and Methods
A Multivariate Generalization of Quantile-Quantile Plots (pp. 376-386)
George S. Easton and Robert E. McCullochTesting the Goodness of Fit of a Linear Model Via Nonparametric Regression Techniques (pp. 387-392)
R. L. Eubank and C. H. SpiegelmanInfluence on Confidence Regions for Regression Coefficients in Generalized Linear Models (pp. 393-397)
William ThomasSampling-Based Approaches to Calculating Marginal Densities (pp. 398-409)
Alan E. Gelfand and Adrian F. M. SmithKernel Quantile Estimators (pp. 410-416)
Simon J. Sheather and J. S. MarronRefining Bootstrap Simultaneous Confidence Sets (pp. 417-426)
Rudolf BeranSmall-Sample Confidence Intervals (pp. 427-434)
Maureen Tingley and Christopher FieldExtension of the Stein Estimating Procedure Through the Use of Estimating Functions (pp. 435-440)
Kung-Yee Lian and Myron A. WaclawiwComponents of Pearson’s Phi-Squared Distance Measure for the k-Sample Problem (pp. 441-445)
R. L. Eubank and V. N. LaRicciaBreakdown Robustness of Tests (pp. 446-452)
Xuming He, Douglas G. Simpson and Stephen L. PortnoyExact Inference for Contingency Tables with Ordered Categories (pp. 453-458)
Alan Agresti, Cyrus R. Mehta and Nitin R. PatelThe Relative Efficiency of Goodness-of-Fit Statistics in the Simple and Composite Hypothesis-Testing Problem (pp. 459-463)
Martin T. WellsThe Evaluation of Integrals of the Form ∫+∞-∞ f(t)exp (- t2) dt: Application to Logistic-Normal Models (pp. 464-469)
Edmund A. C. Crouch and Donna SpiegelmanThe Maximal Smoothing Principle in Density Estimation (pp. 470-477)
George R. TerrellSigned-Rank Tests for Censored Matched Pairs (pp. 478-485)
Dorota M. DabrowskaBootstrap Prediction Intervals for Autoregression (pp. 486-492)
Lori A. Thombs and William R. SchucanySensitivity of Two-Sample Permutation Inferences in Observational Studies (pp. 493-498)
Paul R. Rosenbaum and Abba M. KriegerModeling Time-Varying Dynamical Systems (pp. 499-507)
Carlo GrillenzoniBounded-Influence Rank Regression: A One-Step Estimator Based on Wilcoxon Scores (pp. 508-513)
Mara TablemanThe Accuracy of Approximate Intervals for a Binomial Parameter (pp. 514-518)
Hanfeng ChenEfficiencies of Interblock Rank Statistics for Repeated Measures Designs (pp. 519-528)
G. L. Thompson and L. P. AmmannTwo-Sample Inference for Median Survival Times Based on One-Sample Procedures for Censored Survival Data (pp. 529-536)
Jane-Ling Wang and Thomas P. HettmanspergerA Simulation Study of the Analysis of Sets of 2 × 2 Contingency Tables Under Cluster Sampling: Estimation of a Common Odds Ratio (pp. 537-543)
Alan Donald and Allan DonnerConfidence Curves in Nonlinear Regression (pp. 544-551)
R. Dennis Cook and Sanford WeisbergA Multivariate Signed-Rank Test for the One-Sample Location Problem (pp. 552-557)
Dawn Peters and Ronald H. RandlesModels for Distributions on Permutations (pp. 558-564)
Hal SternTests of Hypotheses in Overdispersed Poisson Regression and Other Quasi-Likelihood Models (pp. 565-571)
Norman BreslowStatistical Densities, Cumulatives, Quantiles, and Power Obtained by S-System Differential Equations (pp. 572-578)
Philip F. Rust and Eberhard O. Voitomparison of Linear Estimators Using Pitman’s Measure of Closeness (pp. 579-581)
Robert L. Mason, Jerome P. Keating, Pranab K. Sen and Neil W. Blaylock, Jr.Testing the Mixture of Exponentials Hypothesis and Estimating the Mixing Distribution by the Methods of Moments (pp. 582-589)
James J. Heckman, Richard Robb and James R. WalkerBook Reviews
The Evolving Role of Statistical Assessments as Evidence in Courts. by Stephen E. Fienberg (p. 591)
Review by: Denis J. HauptlyNews & Numbers: A Guide to Reporting Statistical Claims and Controversies in Health and Related Fields. by Victor Cohn (pp. 591-592)
Review by: Judith M. TanurThe Empire of Chance: How Probability Changed Science and Everyday Life. by Gerd Gigerenzer, Zeno Swijtink, Theodore Porter, Lorraine Daston, John Beatty, Lorenz Kruger (p. 592)
Review by: Peter GuttorpCounterexamples in Probability. by Jordan M. Stoyanov (p. 592)
Review by: R. W. HammingStatistical Analysis and Mathematical Modelling of Aids. by J. C. Jager, E. J. Ruitenberg (p. 593)
Review by: Jeremy M. G. TaylorTransformation and Weighting in Regression. by Raymond J. Carroll, David Ruppert (pp. 593-594)
Review by: Peter A. LachenbruchNonlinear Regression Analysis and Its Applications. by Douglas M. Bates, Donald G. Watts;Nonlinear Regression. by G. A. F. Seber, C. J. Wild (pp. 594-595)
Review by: Robert E. KassApplied Regression Analysis in Econometrics. by Howard E. Doran (pp. 595-596)
Review by: Farid KianifardTheoretical and Computational Aspects of Simulated Annealing. by P.J.M. van Laarhoven (p. 596)
Review by: J. Michael SteeleOrder Restricted Statistical Inference. by Tim Robertson, F. T. Wright, R. L. Dykstra (pp. 596-597)
Review by: Thomas SantnerThe Design of Experiments: Statistical Principles for Practical Applications. by R. Mead (p. 597)
Review by: Larry WassermanThe Analysis of Categorical Data Using GLIM. by James K. Lindsey (pp. 597-598)
Review by: Peter A. LachenbruchNonparametric Estimation of Probability Densities and Regression Curves. by E. A. Nadaraya, Samuel Kotz (p. 598)
Review by: David W. ScottRobustness of Statistical Tests. by Takeaki Kariya, Bimal K. Sinha (pp. 598-599)
Review by: John I. MardenDynamic Graphics for Statistics. by William S. Cleveland, Marylyn E. McGill (p. 599)
Review by: William F. EddyComputer Intensive Methods for Testing Hypotheses: An Introduction. by Eric W. Noreen (pp. 599-600)
Review by: Virginia F. FlackMultivariate Statistics: A Practical Approach. by Bernhard Flury, Hans Riedwyl (p. 600)
Review by: Christine WaternauxData Analysis: A Model-Comparison Approach. by Charles M. Judd, Gary H. McClelland (pp. 600-601)
Review by: Virginia ClarkQuality Control, Robust Design, and the Taguchi Method. by Khosrow Dehnad (p. 601)
Review by: Thomas C. HsiangPanel Surveys. by Daniel Kasprzyk, Greg J. Duncan, Graham Kalton, M. P. Singh (pp. 601-602)
Review by: Seymour SudmanThe Politics of Numbers. by William Alonso, Paul Starr;The American Census: A Social History. by Marjo J. Anderson (pp. 602-603)
Review by: Daniel MelnickAmerican Women in Poverty. by Paul E. Zopf, Jr. (pp. 603-604)
Review by: A. Dianne SchmidelyAmerican Neighborhoods and Residential Differentiation. by Michael J. White (pp. 604-605)
Review by: Anne B. ShlayWorkshop on Statistical Uses of Microcomputers in Federal Agencies by Subcommittee on Statistical Uses of Microcomputers in Federal Agencies (pp. 605-606)
Review by: Robert F. TeitelIntroduction to Probability Models (4th ed.). by Sheldon M. Ross (p. 606)
Review by: Roger CarlsonEncyclopedia of Statistical Sciences (supp.). by Samuel Kotz, Norman L. Johnson, Campbell B. Read (p. 606)
Review by: Donald Guthrie
Lots of social science and lots of hypothesis testing.
As a comparison, here’s the most recent issue currently available on Jstor, from December 2018. This time, there’s much more of a focus on high-dimensional data (notably in biology), statistical computing, and Bayesian modeling. No book reviews.
It is already 24 hours since this was first posted, yet no one has as yet chimed in to offer a comment regarding how the focus of attention has shifted since June, 1990. I suggest that Andrew offer an opinion as to why the absence and the reticence. Perhaps it would help if a prize were awarded for the most coherent, cogent explanation.
I didn’t comment because I don’t feel I have the expertise to intelligently say anything about the changes within the statistics profession. But I do have a reaction to comparing the titles then and now. I found the 1990 titles more direct and applied and focused on interesting questions. The more recent titles seemed more focused on technique and less directly tied to application (although it was clear that there were practical applications, the titles seemed to me to push the application to be secondary to the methods). I’ll be interested to see what those within the statistics community say.
Paul, Dale:
I dunno. Just in general these historical studies get less attention than I’d expect. Aki and I put a lot of effort into writing our article, What are the most important statistical ideas of the past 50 years?. We had the idea of publishing it along with discussions from different people offering their perspectives, but the journal wasn’t interested in doing discussions, and that paper received less attention than we’d expected.
Part of it might just be an age thing: most practitioners are too young to remember what was going on 30 or 40 or 50 years ago, so these retrospective comparisons are just not of interest to them.
I think another thing is there’s been a profound change in the power of computing since 1990. In 1990 a Mac SE with a motorola 68000 processor running at 7.8 MHz having 1MB of RAM with two 1.44MB floppy drives cost $2900 (todays dollars around $8000).
Today, a Raspberry Pi 5 with 8000 MB of RAM running 4 cores at 2400 MHz and a micro SD card holding 128000 MB of storage cost around $100
In 1990 buying a compiler for say Pascal or C or Fortran cost hundreds of dollars, probably equivalent to maybe $1000 today. Today you can download Julia for free and get full compilation to hardware instructions, or download Python or R for free and get interpreted language with compiled primitives.
In 1990 you could run an appletalk cable between two Macs and get 14400 bits per second of network file sharing. Today every computer has 1000000000 bits per second network ports as a matter of course including the ~$100 raspberry Pi 5.
So the focus has shifted in statistics away from some kinds of questions and towards others usually related to utilizing incredible amounts of computing power compared to what you could muster in 1990. In 1990 a Raspberry Pi level of performance would have been accessible maybe to the NSA or something.
Daniel:
Good point. Also there’s the dead hand of tradition. Even back in 1990 I wasn’t the only person to think that hypothesis testing was a bad foundation for statistics.
Interesting that you bring up computation — one of the papers in the 1990 journal is “Sampling-Based Approaches to Calculating Marginal Densities”. Computation in 1990 was just at the point where MCMC was usable. How best to use the massive increase in computation power is still, at least to me, an open question.
Robin, yes MCMC type stuff could just begin its development. Radford Neal’s famous article on HMC was 1996, where computing was already ~8-10x faster and better than 1990.
Robin:
Also, my paper—the very first article in that issue—used Gibbs sampling to fit a hierarchical Bayesian mixture model.
> In 1990 buying a compiler for say Pascal or C or Fortran cost hundreds of dollars, probably equivalent to maybe $1000 today.
Turbo C and Turbo Pascal costed less than $100 – not hundreds of dollars. It’s true that the cheapest FORTRAN was probably Microsoft’s and it was over $200 (still below $1000 today).
>Today you can download Julia for free and get full compilation to hardware instructions, or download Python or R for free and get interpreted language with compiled primitives.
Back then you could download XLispStat for free, or download Perl for free.
Daniel:
Back in 1990 I programmed in Fortran on my workstation and I’m pretty sure the Fortran was free.
Andrew, yeah all you had to do was buy a UNIX workstation, which cost maybe $10k at the time (about $25k today).
Carlos, I was a Mac guy until the mid 1990’s (used BeOS for a couple years and then started using Debian in probably 1995-1997?). In general Mac stuff was a bit more expensive than MS Dos stuff for x86. I vaguely remember Lightspeed Pascal being about $250? So that’d be $600 today. But also you had to buy any upgrades, so it was a little more like a subscription unless you were willing to constantly use older stuff. So if you bought a compiler in 1990 by 1995 you’d probably spent multiple hundreds of nominal mid 1990’s dollars for upgrades etc. Eventually Metroworks had a really affordable C/C++ compiler for the Mac (and for BeOS) that was like $99 in 1995 ish. That’s still $240 today which is a lot more than what you’d pay to get Julia, R, Python, gcc, g++, clang, gnu fortran, gprolog, VSCode, and git on your desktop today (all of that put together would cost nothing)
XLispStat seemed cool but I never actually used it. Common Lisp compilers like Apple’s Macintosh Common Lisp were pretty expensive at the time, around $500 which would today be $1200
Anyway, computing in 1990’s was a lot more expensive I think we all agree on that.