Table of Contents
Title Page
Copyright
Contributors
Chapter 1: Introduction
References
Part I: Contextual Backgrounds
Chapter 1: Reproducibility, Objectivity, Invariance
1.1 Introduction
1.2 Reproducibility in the Empirical Sciences
1.3 Objectivity
1.4 Invariance and Symmetry
1.5 Summary
References
Chapter 2: Reproducibility between Production and Prognosis
2.1 Preliminary Remarks: Three Myths
2.2 How Does Reproducibility Connect with Production?
2.3 How Does Production Connect with Continuity?
2.4 How Does Continuity Connect with Scientific Rationality?
2.5 How Does Scientific Rationality Connect with Prognosis?
2.6 How Do Prediction and Prognosis Connect with Reproducibility?
2.7 Concluding Remarks
References
Chapter 3: Stability and Replication of Experimental Results: A Historical Perspective
3.1 Experiments and Their Reproduction in the Development of Science
3.2 Repetition of Experiments
3.3 The Power of Replicability
3.4 Cases of Failed Replication
3.5 Doing Science without Replication and Replicability
3.6 What Can We Learn from History?
Acknowledgments
References
Chapter 4: Reproducibility of Experiments: Experimenters' Regress, Statistical Uncertainty Principle, and the Replication Imperative
4.1 Introduction
4.2 The Experimenter's Regress
4.3 The Statistical Uncertainty Principle
4.4 The Replication Imperative
References
Part II: Statistical Issues
Chapter 5: Statistical Issues in Reproducibility
5.1 Introduction
5.2 A Random Sample
5.3 Structures of Variation
5.4 Regression Models
5.5 Model Development and Selection Bias
5.6 Big and High-Dimensional Data
5.7 Bayesian Statistics
5.8 Conclusions
Acknowledgments
References
Chapter 6: Model Selection, Data Distributions, and Reproducibility
6.1 Introduction
6.2 Bayesian Model Selection and Relation to Minimum Description Length
6.3 Extending BMS (and NML#): BMS*
6.4 Replication Variance and Reproducibility
6.5 Final Remark
References
Chapter 7: Reproducibility from the Perspective of Meta-Analysis
7.1 Introduction
7.2 Basics of Meta-Analysis
7.3 Meta-Analysis of Mind-Matter Experiments: A Case Study
7.4 Summary
References
Chapter 8: Why Are There So Many Clustering Algorithms, and How Valid Are Their Results?
8.1 Introduction
8.2 Supervised and Unsupervised Learning
8.3 Cluster Validity as Easiness in Classification
8.4 Applying Clustering-Quality Measures to Data
8.5 Other Clustering Models
8.6 Summary
References
Part III: Physical Sciences
Chapter 9: Facilitating Reproducibility in Scientific Computing: Principles and Practice
9.1 Introduction
9.2 A Culture of Reproducibility
9.3 Statistical Overfitting
9.4 Performance Reporting in High-Performance Computing
9.5 Numerical Reproducibility
9.6 High-Precision Arithmetic in Experimental Mathematics and Mathematical Physics
9.7 Reproducibility in Symbolic Computing
9.8 Why Should We Trust the Results of Computation?
9.9 Conclusions
References
Chapter 10: Methodological Issues in the Study of Complex Systems
10.1 Introduction
10.2 Definitions of Complexity
10.3 Complexity and Meaning
10.4 Beyond Stationarity and Ergodicity
10.5 Conclusions
Acknowledgments
References
Chapter 11: Rare and Extreme Events
11.1 Introduction
11.2 Statistics of Extremes
11.3 Predictions of Extreme Events
11.4 Evolving Systems Exposed to Extreme Events
11.5 Conclusions
Acknowledgments
References
Chapter 12: Science under Societal Scrutiny: Reproducibility in Climate Science
12.1 Reproducibility Challenges for Climate Science
12.2 Reproducibility in Observational Climate Science
12.3 Reproducibility in Climate Modeling
12.4 Reproducibility in Paleoclimatology
12.5 Conclusions and Recommendations
References
Part IV: Life Sciences
Chapter 13: From Mice to Men: Translation from Bench to Bedside
13.1 The Drug Development Process
13.2 Contributions of Animals to Medical Progress
13.3 Translation Challenges in Different Fields of Research
13.4 Increasing Translational Success: Summary and Conclusions
References
Chapter 14: A Continuum of Reproducible Research in Drug Development
14.1 Introduction
14.2 The Strategy of the Magic Bullet
14.3 Specialists and Generalists
14.4 From Single-Target to Multi-Target Drugs
14.5 Conclusions
References
Chapter 15: Randomness as a Building Block for Reproducibility in Local Cortical Networks
15.1 Introduction
15.2 Spike Trains and Reproducibility
15.3 Spike Trains
15.4 Neuronal Populations
15.5 Summary
References
Chapter 16: Neural Reuse and In-Principle Limitations on Reproducibility in Cognitive Neuroscience
16.1 Introduction
16.2 The Erosion of Modular Thinking
16.3 Intrinsic Limits on Reproducibility
16.4 Going Forward
References
Chapter 17: On the Difference between Persons and Things – Reproducibility in Social Contexts
17.1 The Problem of Other Minds and Its Evolutionary Dimension
17.2 Understanding the Inner Experience of Others
17.3 Identifying the Neural Mechanisms of Understanding Others
17.4 Abduction of the Functional Roles of Neural Networks
17.5 Psychopathology of the Inner Experience of Others
17.6 Conclusions
References
Part V: Social Sciences
Chapter 18: Order Effects in Sequential Judgments and Decisions
18.1 Introduction
18.2 Question Order Effects and QQ Equality
18.3 No Order Effect Model and Saturated Model
18.4 The Anchor Adjustment Model
18.5 The Repeat Choice Model
18.6 The Quantum Model
18.7 Concluding Comments
References
Chapter 19: Reproducibility in the Social Sciences
19.1 Introduction
19.2 Reproducibility as a Current Problem in the Social Sciences
19.3 “Reproductions Have No Meaningful Scientific Value”
19.4 Reaction from the Blogosphere
19.5 Conclusion
References
Chapter 20: Accurate But Not Reproducible? The Possible Worlds of Public Opinion Research
20.1 Introduction
20.2 Reproducibility: A Missing Criterion in Public Opinion Research?
20.3 Big Data versus Science: The Breakthrough of Modern Polling
20.4 The Birth of a Statistical Myth
20.5 Generating Trust 10
20.6 The Possible Worlds of Public Opinion Research
20.7 Looping Effects between Measurement and Measured
20.8 Swarms of Possible Worlds
References
Chapter 21: Depending on Numbers
21.1 Introduction
21.2 Statistical Error
21.3 Translation
21.4 Statistical and Substantive Significance
21.5 Irreproducible Numbers
21.6 Reproducing Calculations
References
Chapter 22: Science Between Trust and Control: Non-Reproducibility in Scholarly Publishing
22.1 Introduction
22.2 Reproducibility as the Touchstone for Distinguishing Science from Non-Science
22.3 Contested Claims: The Story behind STAP
22.4 The Structural Gap between the Production and Representation of Scientific Facts
22.5 The Increasing Awareness of Reproducibility Problems
22.6 The New Transparency: Bridging the Gap in Scholarly Publishing
22.7 Conclusions
References
Part VI: Wider Perspectives
Chapter 23: Repetition with a Difference: Reproducibility in Literature Studies
23.1 Introduction
23.3 Language and Difference
23.3 Mimesis, Imitatio, and Parody
23.4 Literary Translation: Domesticating versus Foreignizing
23.5 Reproducing Cultural Significance
23.6 Conclusions
Acknowledgments
References
Chapter 24: Repetition Impossible: Co-Affection by Mimesis and Self-Mimesis
24.1 Introduction
24.2 Repetition within the Philosophy of Time
24.3 Re-Presenting Forgetting
24.4 Repetition, Co-Affection and Trauma: Identity and Coping with the Past
24.5 The Dialectics of Remembering and Forgetting
24.6 Co-Affection and Memorizing Recall
24.7 Mimesis
References
Chapter 25: Relevance Criteria for Reproducibility: The Contextual Emergence of Granularity
25.1 Introduction
25.2 Contrast Classes, Coarse Grains, Partition Cells
25.3 Two Examples
25.4 Contextual Emergence
25.5 Ontological Relativity: Beyond Fundamentalism and Relativism
Acknowledgments
References
Chapter 26: The Quest for Reproducibility Viewed in the Context of Innovation Societies
26.1 Introduction
26.2 A Genealogical Sketch of “Innovation”
26.3 Reframing Scientific Ethos: Sound Science
26.4 Reproducibility and Innovation: A Regulative Dual
26.5 Making the Implicit Explicit I: Social Robustness of Science
26.6 Making the Implicit Explicit II: Responsible Research and Innovation
26.7 Mertonian Norms Challenged Anew: Institutional Reflexivity and Responsiveness
References
Index
End User License Agreement
Pages
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
Guide
Table of Contents
Introduction
Part I
Begin Reading
List of Illustrations
Chapter 5: Statistical Issues in Reproducibility
Figure 5.1 Measurements of the velocity of light by Newcomb in 1882. Three confidence intervals are shown on the right.
Figure 5.2 Histogram (left) and empirical cumulative distribution function (right) for the Newcomb data. The dashed and dotted lines represent normal distributions obtained from fits to the data without and with two outliers, respectively.
Figure 5.3 Student's measurements of the concentration of nitrogen in aspartic acid. The bars represent confidence intervals based on the -test for the first 32, 64, 96, and 128 of 131 observations.
Chapter 6: Model Selection, Data Distributions, and Reproducibility
Figure 6.1a Case A – the data distributions are the binomial distributions for the number of successes in 10 trials predicted by model instances with in the range from 0.0 to 1.0 in steps of 0.1.
Figure 6.1b Case A – the toy example imposes a geometric prior on the data distributions. Due to their one-to-one correspondence with the priors on data distributions, the model instances have the same priors.
Figure 6.2 Case B – 11 additional “partner” distributions are each a mixture of one of the binomials with a uniform distribution, and are shown to the right of the binomial partner.
Figure 6.2b Case B – prior probabilities for the 22 distributions: the binomial distribution is assumed to be three times less likely than its partner.
Figure 6.2c Case B – the prior of each model instance is just the sum of its two best matching partner distributions.
Figure 6.3 Case A – posterior probabilities for the data distributions following an outcome of three successes in 10 trials. For Case A, these are also the posterior probabilities of the corresponding model instances.
Figure 6.3 Case B – posterior probabilities for the 22 data distributions, following an outcome of three successes in 10 trials.
Figure 6.3 Case B – posterior probabilities for the model instances are the sum of the posterior probabilities for the two partner distributions.
Figure 6.4 Predicted distribution for the mean number of successes expected in ten trials, based on posterior probabilities following an outcome of three observed successes in 10 trials. Solid lines connecting dots are for Case A, and dashed lines connecting triangles are for Case B.
Chapter 7: Reproducibility from the Perspective of Meta-Analysis
Figure 7.1 Forest plot of the beta-blocker data from Carlin (1992). See text for explanation.
Figure 7.2 Forest plot of the magnesium data from Sterne and Egger (2001). See text for explanation.
Figure 7.3 Funnel plots of standard deviations of the estimates (SE) versus log odds ratio for the beta blocker (a) and the magnesium data (b) using fixed-effects estimates. See text for explanation.
Figure 7.4 Funnel plots of standard deviations of the estimates (SE) versus log odds ratio for the beta blocker (a) and the magnesium data (b) using random-effects estimates. See text for explanation.
Figure 7.5 Histogram of -scores under experimental conditions (data from Radin and Nelson (1989)). The bell curve represents the distribution expected under the null hypothesis of no effect.
Figure 7.6 Normal probability plots of the standardized residuals for the FCR, RCR1, RCR0, and RCRS model (from upper left to lower right).
Figure 7.7 (a) and (b) are histograms of 5,000 test -values simulated under the null hypotheses of no selection and no effect, respectively. (c) and (d) are profiles of weights estimated within the RCRS model, when there is selection (d), and when there is not (c). Squares: averages of estimated -profiles from 5,000 simulated data. Circles: -profiles underlying the simulations. Broken lines: 10 -profiles randomly selected from the 5,000 simulated profiles (providing an impression of the variability to be expected).
Chapter 8: Why Are There So Many Clustering Algorithms, and How Valid Are Their Results?
Figure 8.1 Percentage of incorrectly classified instances versus perturbation size for datasets of Type I (left) and of Type VI (right) for . With WEKA's simple -means there is no preferred value for (the number of clusters) as the results show no stability of the error rate with respect to perturbation size.
Figure 8.2 Analysis of one dataset of Type III on the left shows as the largest value of for which there is stability (in learned classifier accuracy) to perturbations when we consider the clustering results of WEKA's simple -means as a supervised instance. On the right we have a similar (and typical) result of our analysis, but for dataset of Type IV. Again, is the model-order identified for WEKA's simple -means matching the number of components of the mixture generating the data.
Figure 8.3 Reproducing previous studies, we determine the model order of WEKA's simple -means for the datasets of Type II (left) and Type V (right). The analysis with the CQM method shows correctly that the largest values of where there is stability in supervised learning to perturbation are and , respectively. We illustrate the plot for one randomly chosen dataset out of the 50 of each data type.
Figure 8.4 Distribution with mixture of two components with proportions as per Eq. (8.3). Left: , middle: , right: .
Figure 8.5 Plots for the analysis of the dataset Leukemia with WEKA's simple -means. The classifier is WEKA's NaiveBayes. Left: Our analysis shows the preferred values of and since the supervised learning problem shows the same error rate for until , while does not start increasing until . On the other hand, suffers from as early as . Right: Once is determined, we analyze the different clustering obtained with different seeds.
Figure 8.6 For the dataset Leukemia and the subclass ALL, obtained with an earlier clustering, we see that simple -means with preferrably (and maybe with ) yields sub-clusters.
Figure 8.7 Identification of the number of clusters (components) in the mixture learned by EM on the simple data of Estivill-Castro (2011, Figure 1). Left: Resilience to perturbation of NaiveBayes applied to a supervised-learning problem by forcing a crisp clustering of EM results selecting the most likely cluster. Middle: Resilience to perturbation of NaiveBayes applied to several supervised learning problems by choosing a class as per the distribution of EM results. Right: Resilience to perturbation of Voronoi partition classification applied to several supervised-learning problems by choosing a class as per the distribution of EM results.
Chapter 9: Facilitating Reproducibility in Scientific Computing: Principles and Practice
Figure 9.1 Performance development from 1994 to 2014 of the top 500 computers: medium curve = #1 system; lower curve = #500 system; upper curve = sum of #1 through #500. See top500.org/statistics/perfdevel/.
Figure 9.2 Minimum backtest length (in years) versus number of trials.
Figure 9.3 Final optimized strategy applied to the input dataset. Note that the Sharpe ratio is 1.32, indicating a fairly effective strategy on this dataset.
Figure 9.4 Final optimized strategy applied to the new dataset. Note that the Sharpe ratio is , indicating a completely ineffective strategy on this dataset.
Figure 9.5 Performance plot of run time as a function of number of objects: parallel system (lower curve) vs. vector system (upper curve)]. Note that for all data points from the table except for the last entry, the vector system is faster than the parallel system.
Chapter 10: Methodological Issues in the Study of Complex Systems
Figure 10.1 Three patterns used to demonstrate the notion of complexity. Typically, the pattern in the middle is intuitively judged as most complex. The left pattern is a periodic sequence of black and white pixels, whereas the pattern on the right is constructed as a random sequence of black and white pixels. (Reproduced from Grassberger (1986) with permission.)
Figure 10.2 Two classes of complexity measures: Monotonic complexity measures essentially are measures of randomness, typically based on syntactic information. Convex complexity measures vanish for complete randomness and can be related to the concept of pragmatic information.
Chapter 12: Science under Societal Scrutiny: Reproducibility in Climate Science
Figure 12.1 Monthly average maximum temperatures during August at Potsdam (gray circles and connecting lines) from 1893 to 2013 exhibiting large interannual variations due to weather phenomena and short-term climatic variability. For investigations of longer-term trends, averages over extended periods of time are more meaningful, as indicated by the 30-year climatological averages at the beginning and at the end of the time series (horizontal black lines).
Figure 12.3 Climate model intercomparison of projections for future global temperatures under different emission scenarios for the fourth (left) and fifth (right) assessment report of the Intergovernmental Panel on Climate Change. (Reproduced from Knutti and Sedláček (2013) with permission by Nature Publishing Group.)
Figure 12.2 Global surface air temperature anomalies (relative to the average over the period 1971–2000) for the four different datasets and analysis methods described in the text.
Figure 12.4 Ensemble reconstructions of Northern-hemisphere surface air temperature anomalies (with respect to the reference period 1370–1420) over the last millennium (Frank et al. 2010). The shaded regions (from light to dark gray) indicate the ranges covered by 90%, 70%, and 50% of the ensemble reconstructions. The solid black line shows a climate model simulation (Feulner 2011) which compares very well with the reconstructed temperatures.
Chapter 13: From Mice to Men: Translation from Bench to Bedside
Figure 13.1 Study design features for 232 publications for the period from 1997 to 2011. Left panel: randomization; right panel: conflict of interest statement.
Figure 13.2 Incidence of experiments with outcome “tumor progression” versus quality score for 1,538 experiments described in 232 publications.
Chapter 15: Randomness as a Building Block for Reproducibility in Local Cortical Networks
Figure 15.1 (from Carandini 2004): Response of a single cell to an identical stimulus. (a) three different trials, and (b) average over seven trials.
Figure 15.2 Different types of spike trains. The -axis shows time (arbitrary units) and each peak corresponds to a spike. (a) Poisson spiking (CV = 1), (b) regular spiking (CV = 0), and (c) random spike bursts (CV ).
Figure 15.3 Inhibition with leakage for lower voltage bounds of (solid), (dashed), or (dotted). Bold lines, left axis: CV of output spike train; thin lines, right axis: output rate for an input rate of 100Hz. On the -axis, we plot the probability that the next spike is excitatory. So corresponds to excess inhibition, while corresponds to excess excitation. The -axis is the CV of the resulting output spike train. For excess inhibition the output CV is close to 1, regardless of the value of . In contrast to the model without lower voltage bound, the rate is not fixed by insisting on a CV of 1.
Figure 15.4 (from Lengler et al. 2013): Response of a heterogeneous (dark) and a homogeneous (light) network with balanced excitation and inhibition for Poisson input of varying rates. (a) Reaction to input which is not perfectly Poisson. The plot shows that the heterogeneous network has strictly smaller CVs. Moreover, the homogeneous network shows strongly fluctuating responses to similar inputs. (b) Reaction to Poisson input that is disturbed by spontaneous bursts of input synchronization. The -axis gives the fraction of input spikes that belong to bursts. The input rate is constant. For the heterogeneous network the output rate is robust and concentrated, while the homogeneous network shows much less predictable responses. The shaded areas show the standard deviations. (c) Behavior if the output of the population is fed in as input to an additional population; populations 1, 2, and 3 are shown (from solid to dotted). For the heterogeneous network the spike trains of different neurons are decorrelated as the signal propagates, while the homogeneous network increases cross-correlations. (d) Reaction of the network to a small number of neurons ( -axis) each of which feeds one input spike into the network within a time interval of 10 ms (solid lines), 20 ms (dashed lines), or 30 ms (dotted lines). The -axis shows the time of the first spike in the population. The curves start at the input size where all 100 trials produced at least one spike. The heterogeneous network can be activated by fewer input spikes and reacts faster.
Chapter 16: Neural Reuse and In-Principle Limitations on Reproducibility in Cognitive Neuroscience
Figure 16.1 Panel A offers a spatial map of motor cortex, showing which local regions exhibited neural activity during movements of the digit, wrist, etc. Panel B summarizes the total amount of the mapped area showing neural activity during the specified movements. Note both the similarities and differences in the spatial map during the two training phases and reacquisition. The same behavior can be supported by somewhat different neural arrangements. (Reprinted from Scheiber (2001) with permission license # 3577790361692.)
Figure 16.2 Top panel shows neural activation during a theory of mind task, and bottom panel during a change detection task. The left side of each panel displays the group average activation maps, and the right side the percentage of individuals who had activation at each location in the group average map. Note that for the stricter comparisons (social vs. random, and relational vs. match, which show differences in activity in the experimental task vs. a control task, rather than vs. baseline activity, and thus represent the regions of the brain specific to the experimental task) very little of the group average activation was displayed by more than 50% of individual participants. (Reprinted from Barch et al. 2013 with permission license # 3577790735856.)
Figure 16.3 Functional fingerprints for selected regions of the brain. Clockwise from top left: left intraparietal sulcus, left auditory cortex, right auditory cortex, and right anterior cingulate. The lines represents the relative amount of the overall activity (with confidene intervals) for each region that was observed in each of 20 task domains.
Chapter 18: Order Effects in Sequential Judgments and Decisions
Figure 18.1 Test of the no order effect model. Left panel: All 72 data sets with some specifically selected to have order effects. Right panel: 66 data sets not specifically selected to have order effects. See text for discussion.
Figure 18.2 Left panel: Quantile plot for the chi-square test of the anchor-adjustment model. Values should fall on a line with unit slope and zero intercept. Right panel: Scatter plot relating the bias parameters obtained from each order. To satisfy the QQ equality, the correlation must equal +1.0.
Figure 18.3 Left panel: Quantile plot for the chi-square test of the repeat choice model. Values should fall on a line with unit slope and zero intercept. Right panel: Scatter plot relating the context effect to the difference in marginal choice probabilities. The correlation is predicted to equal +1.0.
Figure 18.4 Left panel: Quantile plot for the chi-square test of the quantum model. Values should fall on the line with unit slope and zero intercept. Right panel: Scatter plot relating the two order effects in the minor diagonal of Table 18.5c. The observed correlation of is close to the predicted slope of .
Chapter 22: Science Between Trust and Control: Non-Reproducibility in Scholarly Publishing
Figure 22.1 Types of accompanying materials for an original paper
Chapter 25: Relevance Criteria for Reproducibility: The Contextual Emergence of Granularity
Figure 25.1 Two chiral versions of the molecule thalidomide, also known under the brand name Contergan® (also Softenon® ); left: -thalidomide acts as a sedative, right: -thalidomide is acutely teratogenic and caused thousands of babies with deformed extremities in the 1950s.
List of Tables
Chapter 5: Statistical Issues in Reproducibility
Table 5.1 The probabilities of getting the same result in the replication study, for the four possible cases of correct null hypothesis and alternative hypothesis and significance of test results in the original study. * Replication of wrong results is undesirable.
Table 5.2 Levels of validation for same and different features (middle columns) in the original and follow-up study. The different terms in the left column are related to different types of validation in the right column.
Chapter 6: Model Selection, Data Distributions, and Reproducibility
Table Matrix 6.1 Prior probabilities for data distributions (left margin), distributions predicted by model instances (right margin), and model outcomes (bottom margin). A posterior version of the same matrix can be produced where the margins are conditioned on the observed data outcome . As explained in the text, virtually all the equations and predictions in this chapter can be derived and explained with reference to this matrix. Data outcomes in bolded columns are explained in Section 6.4.2.
Table 6.1 follows the form of Matrix 6.1, with probabilities corresponding to the priors and data distributions for the toy example, for Case A described in the text. Case A assumes all distributions to be binomial, based on the corresponding value of (from 0.0 to 1.0 in steps of 0.1).
Chapter 9: Facilitating Reproducibility in Scientific Computing: Principles and Practice
Table 9.1 Run times on parallel and vector systems for different problem sizes (data for Fig. 9.5).
Chapter 13: From Mice to Men: Translation from Bench to Bedside
Table 13.1 Outcome distribution for drugs in the “main focus” category and in the “comparison” category, within category “drug type.”
Chapter 18: Order Effects in Sequential Judgments and Decisions
Table 18.1a: white-black
Table 18.1b: black-white
Table 18.1c: order effects
Table 18.2 joint probabilities for the no order effect model
Table 18.3a: anchor-adjust joint probabilities for A–B order
Table 18.3b: anchor-adjust joint probabilities for B–A order
Table 18.3c: anchor-adjust predicted order effects
Table 18.4a: repeat-choice joint probabilities for A–B order
Table 18.4b: repeat-choice joint probabilities for B–A order
Table 18.4c: repeat-choice predicted order effects
Table 18.5a: quantum model joint probabilities for A–B order
Table 18.5b: quantum model joint probabilities for B–A order
Table 18.5c: quantum model predicted order effects
Reproducibility
Principles, Problems, Practices, and Prospects
Edited by
Harald Atmanspacher
Collegium Helveticum, University and ETH Zurich, Zurich, Switzerland
Sabine Maasen
Munich Center for Technology in Society, Technical University, Munich, Germany
Copyright © 2016 by John Wiley & Sons, Inc. All rights reserved
Published by John Wiley & Sons, Inc., Hoboken, New Jersey
Published simultaneously in Canada
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permissions.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.
For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com.
Library of Congress Cataloging-in-Publication Data:
Names: Atmanspacher, Harald. | Maasen, Sabine, 1960-
Title: Reproducibility : principles, problems, practices, and prospects /
edited by Harald Atmanspacher, Sabine Maasen.
Description: Hoboken, New Jersey : John Wiley & Sons, Inc., [2016] | Includes
index.
Identifiers: LCCN 2015036802 | ISBN 9781118864975 (cloth)
Subjects: LCSH: Observation (Scientific method) | Science–Methodology.
Classification: LCC Q175.32.O27 R47 2016 | DDC 001.4/2–dc23 LC record available at
http://lccn.loc.gov/2015036802
Michael Anderson
Department of Psychology
Franklin and Marshal College
Lancaster PA, USA
mljanderson@gmail.com
Harald Atmanspacher
Collegium Helveticum
University and ETH Zurich
Zurich, Switzerland
atmanspacher@collegium.ethz.ch
Sabine Baier
Collegium Helveticum
University and ETH Zurich
Zurich, Switzerland
baier@collegium.ethz.ch
David H. Bailey
Lawrence Berkeley
National Laboratory
Berkeley CA, USA
david@davidhbailey.com
Ladina Bezzola Lambert
Department of English
University of Basel
Basel, Switzerland
ladina.bezzola@unibas.ch
Jonathan Borwein
School of Mathematical
and Physical Sciences
University of Newcastle
Callaghan NSW, Australia
jon.borwein@gmail.com
Jerome Busemeyer
Department of Psychological
and Brain Sciences
Indiana University
Bloomington IN, USA
jbusemey@indiana.edu
Suyog Chandramouli
Department of Psychological
and Brain Sciences
Indiana University
Bloomington IN, USA
suchandr@indiana.edu
Harry Collins
School of Social Sciences
Cardiff University
Cardiff, UK
CollinsHM@cardiff.ac.uk
Werner Ehm
Heidelberg Institute
for Theoretical Studies
Heidelberg, Germany
wernehm@web.de
Hinderk Emrich
Psychiatric Clinic
Hannover Medical School
Hannover, Germany
emrich.hinderk@mh-hannover.de
Vladimir Estivill-Castro
Department of Information
and Communication Technologies
University Pompeu Fabra
Barcelona, Spain
vestivill@gmail.com
Georg Feulner
Earth System Analysis
Potsdam Institute for
Climate Impact Research
Potsdam, Germany
feulner@pik-potsdam.de
Gerd Folkers
Collegium Helveticum
University and ETH Zurich
Zurich, Switzerland
folkers@collegium.ethz.ch
Martina Franzen
Wissenschaftszentrum
für Sozialforschung
Reichpietschufer 50
Berlin, Germany
martina.franzen@wzb.eu
Holger Kantz
Nonlinear Dynamics
and Time Series Analysis
Max-Planck-Institute for
Physics of Compex Systems
Dresden, Germany
kantz@pks.mpg.de
Marianne Martic-Kehl
Collegium Helveticum
University and ETH Zurich
Zurich, Switzerland
martic@collegium.ethz.ch
Felix Keller
Humanities and Social Sciences
University of St. Gallen
St. Gallen, Switzerland
felix.keller@unisg.ch
Johannes Lengler
Theoretical Computer Science
ETH Zurich
Zurich, Switzerland
johannes.lengler@inf.ethz.ch
Sabine Maasen
Center for Technology in Society
Technical University
Munich, Germany
sabine.maasen@tum.de
Theodore Porter
Department of History
University of California
Los Angeles CA, USA
tporter@history.ucla.edu
Martin Reinhart
Institute for Social Sciences
Humboldt University
Berlin, Germany
martin.reinhart@hu-berlin.de
P. August Schubiger
Collegium Helveticum
University and ETH Zurich
Zurich, Switzerland
schubiger@collegium.ethz.ch
Richard Shiffrin
Department of Psychological
and Brain Sciences
Indiana University
Bloomington IN, USA
shiffrin@indiana.edu
Werner Stahel
Seminar for Statistics
ETH Zurich
Zurich, Switzerland
stahel@stat.math.ethz.ch
Angelika Steger
Theoretical Computer Science
ETH Zurich
Zurich, Switzerland
steger@inf.ethz.ch
Friedrich Steinle
Institute for Philosophy
Technical University
Berlin, Germany
friedrich.steinle@tu-berlin.de
Victoria Stodden
Graduate School of Library
and Information Sciences
University of Illinois
Urbana-Champaign IL, USA
victoria@stodden.net
Holm Tetens
Institute for Philosophy
Free University
Berlin, Germany
tetens@zedat.fu-berlin.de
Kai Vogeley
Department of Psychiatry
University Hospital
Cologne, Germany
kai.vogeley@uk-koeln.de
Zheng Wang
School of Communication
Ohio State University
Columbus OH, USA
wang.1243@osu.edu
Walther C. Zimmerli
Graduate School
Humboldt University
Berlin, Germany
walther.ch.zimmerli@hu-berlin.de