Severity as a basic concept in philosophy of statistics

Severity as a basic concept in
philosophy of statistics
Deborah G Mayo
Dept of Philosophy, Virginia Tech
October 9, 2024: Neyman Seminar
Dept of Statistics, UC Berkeley 1

I want to begin by thanking the
statistics department!
For elaborate organizational arrangements
Ryan Giordano, Amanda Coston,
Xueyin (Snow) Zhang
Panel discussants:
Ben Recht
Philip Stark
Bin Yu
Xueyin (Snow) Zhang 2

In a conversation with Sir David Cox
COX: Deborah, in some fields foundations do not
seem very important, but we both think foundations of
statistical inference are important; why do you think?
MAYO: …in statistics …we invariably cross into
philosophical questions about empirical knowledge
and inductive inference.
Sir David Cox: 15 July 1924 - 18 January 2022
(“A Statistical Scientist Meets a Philosopher of Science” 2011)
3

4
At one level, statisticians and
philosophers of science ask
similar questions:
• What should be observed and what may
justifiably be inferred from data?
• How well do data confirm or test a model?

Two-way street
• Statistics is a kind of “applied philosophy of
science” (Oscar Kempthorne 1976).
“Inductive Behavior as a Basic Concept in
Philosophy of Science” (Neyman 1957)
5

6
Statistics → Philosophy
Statistical accounts are used in philosophy
of science to:
1) Model Scientific Inference
2) Solve (or reconstruct) Philosophical
Problems about inference and evidence
(e.g., problem of induction)

7
Problem of Induction
• Failure to justify (enumerative) induction led to
building logics of confirmation C(H,e), like
deductive logic: Carnap
• Evidence e is given, calculating C(H,e) is formal.
• Scientists could come to the inductive logician for
rational degree of confirmation
They did not succeed, and the program is
challenged in the 80s (Popper, Lakatos, Kuhn)

8
Popperians
Evidence e is theory-laden, not given:
Collected with difficulty, must start with a problem
Conjecture and Refutation: EI is a myth
Corroboration, not Confirmation: Those that
survive stringent or severe tests

9
Popper never adequately cashed out the
idea—tried adding requirements like novel
predictive success to C(h, e)
(his anti-inductivism and unfamiliarity with
statistical methods)

Neyman and Pearson (N-P): Inductive
performance
“We may look at the purpose of tests from another
view-point...We may search for rules to govern our
behavior…in following which we insure that in the long
run of experience, we shall not be too often wrong”
(Neyman and Pearson 1933, 141-2)
• Probability is assigned, not to hypotheses, but to
methods 10

Series of models
It is a fundamental contribution of modern mathematical
statistics to have recognized the explicit need of a
model in analyzing the significance of experimental
data. (Patrick Suppes 1969, 33)
11

I suspected that understanding how these
statistical methods worked would offer up
solutions to the vexing problem of how we
learn about the world in the face of error.
(Mayo 1996, Error and the Growth of
Experimental Knowledge)
12

13
Philosophy of Science → Statistics
• An obstacle to appealing to a methodology is
if its own foundations are problematic
• A central job for philosophers of science:
illuminate conceptual and logical problems of
sciences

14
• Especially when widely used methods are
said to be causing a crisis (and should be
“abandoned” or “retired”)
• A separate challenge concerns the “new
(data science) paradigm” about the nature
and goals of statistics

Central Issue: Role of probability:
performance or probabilism?
(Frequentist vs. Bayesian)
• End of foundations? (we do what works)
• Long-standing battles simmer below the surface
in today’s “statistical (replication) crisis in
science”
• What’s behind it?
15

I set sail with a minimal principle of
evidence
• We don’t have evidence for a claim C if little if
anything has been done that would have found C
flawed, even if it is
16

17
Severity Requirement
• We have evidence for a claim C only to the
extent C has been subjected to and passes a
test that would probably have found C flawed,
just if it is.
• This probability is the stringency or severity
with which it has passed the test.

18
Error Statistics
This underwrites the aim of statistical significance
tests:
• To bound the probabilities of erroneous
interpretations of data: error probabilities
A small part of a general methodology which I call
error statistics
(statistical tests, confidence intervals, resampling,
randomization)

• (Formal) error statistics is a small part of severe
testing
• The severity concept is sufficiently general to
apply to any methods, including just using data to
solve an error-prone problem (formal or informal)
• But let’s go back to statistical significance tests
19

20
Paradox of replication
• Statistical significance tests are often seen as the
culprit of the replication crisis
• It’s too easy to get small P-values—critics say
• Replication crisis: It’s too hard to get small P-
values when others try to replicate with stricter
controls

21
Fisher & Neyman and Pearson: it’s
easy to lie with biasing selection
effects
• R.A. Fisher: it’s easy to lie with statistics by
selective reporting, (“political principle that anything
can be proved by statistics” (1955, 75))
• Cherry-picking, significance seeking, multiple
testing, post-data subgroups, trying and trying
again—may practically guarantee an impressive-
looking effect, even if it’s unwarranted

Simple statistical significance tests
(Fisher)
“to test the conformity of the particular data under
analysis with H0 in some respect:
…we find a function T = t(y) of the data, the test
statistic, such that
• the larger the value of T the more inconsistent
are the data with H0;
p = Pr(T ≥ t0bs; H0)”
(Mayo and Cox 2006, 81) 22

23
Testing reasoning
• Small P-values indicate* some underlying
discrepancy from H0 because very probably (1- P)
you would have seen a less impressive
difference were H0 true.
• This still isn’t evidence of a genuine statistical
effect H1 yet alone a scientific claim C
*(until an audit is conducted testing assumptions, and
checking biasing selection effects I use “indicate”)

24
Neyman and Pearson tests (1928, 1933)
put Fisherian tests on firmer ground:
• Introduces alternative hypotheses that exhaust the
parameter space: H0, H1
• Trade-off between Type I errors and Type II errors
• Restricts the inference to the statistical alternative
within a model

25
E. Pearson calls in Neyman “an Anglo-
Polish Collaboration: 1926-1934”!
“I was torn between conflicting emotions: a. finding it
difficult to understand R.A.F., b. hating [Fisher] for his
attacks on my paternal ‘god,’ c. realizing that in some
things at least he was right” (Constance Reid 1982,
1998).

26
Fisher-Neyman split: 1935
• Initially, it was Fisher, Neyman and Pearson vs
the “old guard” at University College
• The success of N-P optimal error control led to a
new paradigm in statistics, overshadows Fisher
• “being in the same building at University College
London brought them too close to one another”!
(Cox 2006, 195)

27
Main issue: Fisher’s fiducial fallacy
In short: Fisher claimed the fiducial level measures
both error control and post-data probability on
statistical hypotheses without prior probabilities—
in special cases
Fiducial frequency distribution (to distinguish it
from inverse [Bayesian] inference)
Neyman’s confidence intervals capture what he
assumed Fisher means

28
“[S]o many people assumed for so long that the
[fiducial] argument was correct. They lacked the
daring to question it.” (Good 1971, p.138).
• Neyman did, resulting in the break-up

Construed as a deep philosophical
difference: p’s vs ’s (Fisher vs N-P)
Fisher claimed that N-P turned “my” tests into
acceptance-sampling tools that
“confuse the process appropriate for drawing
correct conclusions, with those aimed rather at,
let us say, speeding production, or saving
money.” (Fisher 1955)
Also, “no scientific worker has a fixed level of
significance” used habitually (Fisher 1956)
29

30
Lehmann (1993):“The Fisher, Neyman-
Pearson Theories of Testing
Hypotheses: One Theory or Two?
“Unlike Fisher, Neyman and Pearson ...did not
recommend a standard [significance] level but
suggested that ‘how the balance [between the two
kinds of error] should be struck must be left to the
investigator.’” (1244)…
“Fisher, although originally recommending the use of
such levels, later strongly attacked any standard
choice”. (1248)
Earlier phrases are changed

31
Lehmann’s view might surprise: “you
can have your cake and eat it too”
“Should [the report consist] merely of a statement of
significance or nonsignificance at a given level, or
should a p value be reported? ...One should routinely
report the p value and, where desired, combine this
with a statement on significance at any stated level.”
(1247)
“it gives an idea of how strongly the data contradict the
hypothesis” (Lehmann and Romano 2005, 63-4)

32
p-value as post data error
probability
“pobs [the observed p-value] is the probability that
we would mistakenly declare there to be
evidence against H0, were we to regard the data
under analysis as just decisive against H0.” (Cox
and Hinkley 1974, p. 66)
error probabilities are counterfactual

33
evidence (p’s) vs error (’s)?
Neymanian Lehmann is viewing the p-value as evidence
Fisherian Cox, as, performance (or calibration)
Critics say this is to confuse evidence (p’s) and error
(’s)
Severity says it is not.

34
…
Neyman’s applied work is more
evidential: decision or conclusion
“A study of any serious substantive problem involves
a sequence of incidents at which one is forced to
pause and consider what to do next. In an effort to
reduce the frequency of misdirected activities one
uses statistical tests” (Neyman 1976).
Since he regards the only logic to be deductive, he
uses “inductive behavior”.
The “act” can be to declare evidence against H0

Severity gives an evidential twist
• What is the epistemic value of good
performance relevant for inference in the
case at hand?
• What bothers you with selective reporting,
cherry picking, stopping when the data look
good, P-hacking
• Not just long-run quality control
35

36
We cannot say the test has done its job in the
case at hand in avoiding sources of
misinterpreting data
The successes are due to the biasing selection
effects, not C’s truth (e.g., marksmanship)

A claim C is not warranted _______
• Probabilism: unless C is true or probable (gets
a probability boost, made comparatively firmer)
• Performance: unless it stems from a method
that would rarely err
• Probativism (severe testing) unless C
passes a test highly capable of uncovering
mistakes
37

The view of evidence underlying
‘probabilism’ is based on the
Likelihood Principle (LP)
All the evidence is contained in the ratio of
likelihoods:
Pr(x0;H0)/Pr(x0;H1)
x supports H0 less well than H1 if H0 is less likely
than H1 in this technical sense
Ian Hacking (1965) “Law of Likelihood” 38

Comparative logic of support
• “there always is such a rival hypothesis viz.,
that things just had to turn out the way they
actually did” (Barnard 1972, 129)
• Pr(H0 is less well supported than H1;H0) is
high for some H1 or other
(comes from the sampling distribution of T)
39

Hacking: “There is no such thing
as a logic of statistical inference”
“I now believe that Neyman, Peirce, and
Braithwaite were on the right lines to follow in
the analysis of inductive arguments” (1980,
141)
40

All error probabilities
violate the LP:
“Sampling distributions, significance levels,
power, all depend on something more [than
the likelihood function]–something that is
irrelevant in Bayesian inference–namely the
sample space” (Lindley 1971, 436)
41

42
Data Dredging and multiplicity
Replication researchers (re)discovered that data-
dependent hypotheses, multiplicity, optional
stopping are a major source of spurious
significance levels.
“Repeated testing complicates the interpretation of
significance levels. If enough comparisons are
made, random error almost guarantees that some
will yield ‘significant’ findings, even when there is no
real effect . . .” (Kaye and Freedman 2011, 127)

Optional Stopping
43
• “if an experimenter uses this [optional stopping]
procedure, then with probability 1 he will
eventually reject any sharp null hypothesis,
even though it be true”. (Edwards, Lindman, and
Savage 1963, 239)
• “the import of the...data actually observed will be
exactly the same as it would had you planned to
take n observations.” (ibid)

44
A question
If a method is insensitive to error probabilities, does
it escape inferential consequences of gambits that
inflate error rates?
It depends on a critical understanding of “escape
inferential consequences”

45
It would not escape consequences for a severe
tester:
• What alters error probabilities alters the
method’s error probing capability
• Alters the severity of what’s inferred

46
• Bayesians and frequentists use sequential
trials—difference is whether/how to adjust
• Post-data selective inference is a major
research area of its own
• With Big Data, new methods to deal with
multiplicity are needed, and Juliet Shaffer
(1995), a faculty member here, is among the
first

Probabilists may block intuitively
unwarranted inferences
(without error probabilities)
• Give high prior probability to “no effect” (spike
prior)
47

48
Can work in some cases, but :
• The believability of data-dredged
hypotheses is what makes them so seductive
• Additional source of flexibility, can also be
data-dependent
“Why should one's knowledge, or ignorance,
of a quantity depend on the experiment being
used to determine it” (Lindley 1972)

49
Data-dredging need not be
pejorative
• It’s a biasing selection effect only when it
injures severity
• It can even increase severity.
• There’s a difference between ruling out chance
variability and explaining a known effect (e.g.,
DNA testing)

50
Neyman initially sought a
Bayesian account
• rare availability of frequentist priors led him
(and Pearson) to develop tools whose validity
did not depend on them
• opposes rules for the ideal rational mind
(fiducial or Bayesian)--dogmatic, not testable
[Aside: the end of Lehmann’s paper describes
where Lehmann thinks Fisher and Neyman do
disagree]

51
Turning to more recent
“statistics wars”
Statistics has kept its reputation as being a
subject of philosophical debate marked by
unusual heights of passion and controversy

American Statistical Association
(ASA):
2015: Concerned about the “statistical crisis of
replication,” it assembles ~2 dozen researchers
Phillip Stark was one
I was a “philosophical observer”
52

ASA: 2016 Statement on P-values
and Statistical significance 1
• Warns: p-values aren’t effect-size measures or
posteriors
• Data dredging “renders the reported p-values
essentially uninterpretable”
1 Wasserstein & Lazar 2016
53

ASA Executive Director’s Editorial
(2019) Abandon ‘statistical
significance’: No threshold
• “The 2016 Statement “stopped just short of recommending that
declarations of ‘statistical significance’ be abandoned”
• “We take that step here ...Whether a p-value passes any
arbitrary threshold should not be considered at all when
deciding which results to present or highlight”
• Not ASA policy
54

• Many think removing P-value thresholds,
researchers lose an incentive to data dredge
and multiple test
• It is also much harder to hold data-dredgers
accountable
• No thresholds, no tests, no falsification
55

2019 The ASA (President’s) Task
Force on Statistical Significance and
Replicability (2019-2021) :
“The use of P-values and significance testing,
properly applied and interpreted,
• are important tools that should not be
abandoned.
• Much of the controversy surrounding statistical
significance can be dispelled through a better
appreciation of uncertainty, variability, multiplicity,
and replicability” (Benjamini et al. 2021)
56

57
The ASA President’s (Karen Kafadar) Task Force:
Linda Young, National Agric Stats, U of Florida (Co-Chair)
Xuming He, University of Michigan (Co-Chair)
Yoav Benjamini, Tel Aviv University
Dick De Veaux, Williams College (ASA Vice President)
Bradley Efron, Stanford University
Scott Evans, George Washington U (ASA Pubs Rep)
Mark Glickman, Harvard University (ASA Section Rep)
Barry Graubard, National Cancer Institute
Xiao-Li Meng, Harvard University
Vijay Nair, Wells Fargo and University of Michigan
Nancy Reid, University of Toronto
Stephen Stigler, The University of Chicago
Stephen Vardeman, Iowa State University
Chris Wikle, University of Missouri

58
Phil Stat underlying criticisms
There are plenty of misinterpretations, but there is
also a deep difference in philosophies of
evidence:
P-values
• exaggerate evidence
• are not evidence,
• must be misinterpreted to provide evidence

59
P-values
• exaggerate evidence, (if equated with a
posterior probability of H0)
• are not evidence (if the evidence requires the
LP)
• must be misinterpreted to provide evidence
(if evidence must be probabilist)

60
• Highly probable differs from being highly
well-probed (in the error statistical sense)
• The goals are sufficiently different that we
may accommodate both
• Trying to reconcile creates confusion—
members of rival school talk past each
other

61
Severity Reformulates tests
in terms of discrepancies (effect sizes) that are
and are not severely-tested
SEV(Test T, data x, claim C)
• In a nutshell: one tests several discrepancies
from a test hypothesis and infers those well or
poorly warranted
Mayo 1991-; Mayo and Spanos (2006, 2011); Mayo
and Cox (2006); Mayo and Hand (2022)

63
Severity requires auditing
Check for
1. biasing selection effects,
2. violations of the statistical model assumptions
• generally rely on simple significance tests
(e.g., IID)
• “diagnostic checks and tests of fit which, I
will argue, require frequentist theory
significance tests for their formal justification”
(Box 1983, p. 57)

64
Auditing
3. unwarranted substantive interpretations
The most serious problem, I concur with Philip
Stark (2022) is “testing statistical hypotheses
that have no connection to the scientific
hypotheses”

65
Falsifying inquiries
We should stringently test (and perhaps falsify)
some of the canonical types of inquiry
An inquiry is falsified by showing its inability to
severely probe the question of interest.
e.g., cleanliness and morality: Does
unscrambling soap words make you less
judgmental on moral dilemmas?

66
Falsifying inquiries
We should stringently test (and perhaps falsify)
some of the canonical types of inquiry
An inquiry is falsified by showing its inability to
severely probe the question of interest.
e.g., cleanliness and morality: Does
unscrambling soap words make you less
judgmental on moral dilemmas?

Concluding remark
• I begin with a simple tool: the minimal
requirement for evidence
• We have evidence for C only to the extent C
has been subjected to and passes a test it
probably would have failed if false
• In formal tests, (relevant) error probabilities
can be used to assess this capacity
67

68
In the severe testing context,
statistical methods should be:
• directly altered by biasing selection effects
• able to falsify claims statistically,
• able to test statistical model assumptions.
• able to block inferences that violate minimal
severity

69
• The goals of the probabilist and the
probativist differ
• There are also differences in the goals of
formal epistemology and philosophy of
statistics
(I’m keen to learn what Snow Zhang thinks)

Statistical inference is
broadened
• Statistical inference: using data to solve
problems of error prone inquiry, qualified
with an assessment of the capability of the
method to have avoided erroneous
interpretations of data
• (Usually) using data to learn about the data
generating mechanism qualified...
70

“inductive learning [in] the frequentist philosophy is
that inductive inference requires building up incisive
arguments and inferences by putting together
several different piece-meal results; ...Although the
complexity ...makes it more difficult to set out
neatly, ...the payoff is an account that approaches
the kind of arguments that scientists build up in
order to obtain reliable knowledge and
understanding of a field” (Mayo and Cox 2006).
“Frequentist Statistics as a Theory of Inductive Inference” in the
proceedings of the second Lehmann Symposium
71

72
I see connections with all the
panelists:
• With Snow Zhang, the importance of clarifying
concepts of knowledge philosophically
• I find the severity requirement in sync with a number
of papers by Philip Stark

73
Bin Yu’s veridical data science, with its goal
ensuring that every step of the data-to-inference
pipeline is robust and trustworthy
Ben Recht’s emphasis on probing algorithms for
potential weaknesses
How far can data science, AI/ML, explainable AI and
black box modeling have error control, at least
qualitatively?

Thank you! I’ll be glad to have
questions and learn from the panelists!
74

References
Barnard, G. (1972). The logic of statistical inference (Review of “The Logic of Statistical Inference” by Ian
Hacking). British Journal for the Philosophy of Science 23(2), 123–32.
Benjamin, D., Berger, J., Johannesson, M., et al. (2018). Redefine statistical significance. Nature Human
Behaviour, 2, 6–10. https://doi.org/10.1038/s41562-017-0189-z
Benjamini, Y., De Veaux, R., Efron, B., et al. (2021). The ASA President’s task force statement on statistical
significance and replicability. The Annals of Applied Statistics. https://doi.org/10.1080/09332480.2021.2003631.
Berger, J. O. and Wolpert, R. (1988). The Likelihood Principle, 2nd ed. Vol. 6 Lecture Notes-Monograph Series.
Hayward, CA: Institute of Mathematical Statistics.
Birnbaum, A. (1969). ‘Concepts of Statistical Evidence’, in Morgenbesser, S., Suppes, P., and White, M. (eds.),
Philosophy, Science, and Method: Essays in Honor of Ernest Nagel, New York: St. Martin’s Press, pp. 112–43.
Box, G. (1983). ‘An Apology for Ecumenism in Statistics’, in Box, G., Leonard, T., and Wu, D. (eds.), Scientific
Inference, Data Analysis, and Robustness, New York: Academic Press, 51–84.
Cox, D. R. (2006). Principles of statistical inference. Cambridge University Press.
https://doi.org/10.1017/CBO9780511813559
Cox, D. and Hinkley, D. (1974). Theoretical Statistics. London: Chapman and Hall.
Cox, D. R., and Mayo, D. G. (2010). Objectivity and conditionality in frequentist inference. In D. Mayo & A.
Spanos (Eds.), Error and Inference pp. 276–304. Cambridge: CUP.
Cox, D. and Mayo, D. (2011). ‘A Statistical Scientist Meets a Philosopher of Science: A Conversation between
Sir David Cox and Deborah Mayo’, in Rationality, Markets and Morals (RMM) 2, 103–14.
Edwards, W., Lindman, H., and Savage, L. (1963). Bayesian statistical inference for psychological research.
Psychological Review, 70(3), 193-242.
75

Fisher, R. A., (1955), Statistical Methods and Scientific Induction, J R Stat Soc (B) 17: 69-78.
Fisher, R. A. (1958). Scientific Methods for Research Workers 13th ed., New York: Hafner.
Good, I. J. (1971). ‘The Probabilistic Explication of Information, Evidence, Surprise, Causality, Explanation, and
Utility’ and ‘Reply’, in Godambe, V. and Sprott, D. (eds.), pp. 108–22, 131–41. Foundations of Statistical Inference.
Toronto: Holt, Rinehart and Winston of Canada
Hacking, I. (1965). Logic of Statistical Inference. Cambridge: Cambridge University Press.
Hacking, I. (1972). ‘Review: Likelihood’, British Journal for the Philosophy of Science 23(2), 132–7.
Hacking, I. (1980). The theory of probable inference: Neyman, Peirce and Braithwaite. In D. Mellor (Ed.), Science,
Belief and Behavior: Essays in Honour of R. B. Braithwaite, Cambridge: Cambridge University Press, pp. 141–60.
Kafadar, K. (2019). The year in review…And more to come. President’s corner. Amstatnews, 510, 3–4.
Kaye, D. and Freedman, D. (2011). ‘Reference Guide on Statistics’, in Reference Manual on Scientific Evidence,
3rd edn. pp. 83–178.
Kempthorne, O. (1976). ‘Statistics and the Philosophers’, in Harper, W. and Hooker, C. (eds.), Foundations of
Probability Theory, Statistical Inference and Statistical Theories of Science, Volume II, pp. 273–314. Boston, MA:
D. Reidel.
Kuhn, Thomas S. (1970). The Structure of Scientific Revolutions. Enlarged (2nd ed.). University of Chicago Press.
Lakatos, I. (1978). The methodology of scientific research programmes. Cambridge University Press.
Lehmann, E. (1993). ‘The Fisher, Neyman-Pearson Theories of Testing Hypotheses: One Theory or Two?’, Journal
of the American Statistical Association 88 (424), 1242–9.
Lehmann, E. and Romano, J. (2005). Testing Statistical Hypotheses, 3rd edn. New York: Springer.
Lindley, D. V. (1971). The estimation of many parameters. In V. Godambe & D. Sprott, (Eds.),Foundations of
Statistical Inference, pp. 435–455. Toronto: Holt, Rinehart and Winston.
Lindley, D. V. (1972). Bayesian statistics, a review. Society for Industrial and Applied Mathematics.
76

77
Mayo, D. G. (1996). Error and the Growth of Experimental Knowledge. Science and Its Conceptual Foundation.
Chicago: University of Chicago Press.
Mayo, D. (2014). On the Birnbaum Argument for the Strong Likelihood Principle (with discussion). Statistical Science
29(2), 227–39; 261–6.
Mayo, D. G. (2018). Statistical Inference as Severe Testing: How to Get Beyond the Statistics Wars, Cambridge:
Cambridge University Press.
Mayo, D. G. (2019). P-value Thresholds: Forfeit at Your Peril. European Journal of Clinical Investigation 49(10): e13170.
(https://doi.org/10.1111/eci.13170
Mayo, D. G. (2022). The statistics wars and intellectual conflicts of interest. Conservation Biology : The Journal of the
Society for Conservation Biology, 36(1), 13861. https://doi.org/10.1111/cobi.13861.
Mayo, D.G. (2023). Sir David Cox’s Statistical Philosophy and its Relevance to Today’s Statistical Controversies. JSM
2023 Proceedings, DOI: https://zenodo.org/records/10028243.
Mayo, D. G. and Cox, D. R. (2006). Frequentist statistics as a theory of inductive inference. In J. Rojo, (Ed.) The Second
Erich L. Lehmann Symposium: Optimality, 2006, pp. 247-275. Lecture Notes-Monograph Series, Volume 49, Institute of
Mathematical Statistics.
Mayo, D.G., Hand, D. (2022). Statistical significance and its critics: practicing damaging science, or damaging scientific
practice?. Synthese 200, 220.
Mayo, D. G., and A. Spanos. (2006). Severe testing as a basic concept in a Neyman–Pearson philosophy of induction.”
British Journal for the Philosophy of Science 57(2) (June 1), 323–357.
Mayo, D. G., and A. Spanos (2011). Error statistics. In P. Bandyopadhyay and M. Forster (Eds.), Philosophy of
Statistics, 7, pp. 152–198. Handbook of the Philosophy of Science. The Netherlands: Elsevier.
Neyman, J. (1934). ‘On the Two Different Aspects of the Representative Method: The Method of Stratified Sampling and
the Method of Purposive Selection’, The Journal of the Royal Statistical Society 97(4), 558–625. Reprinted 1967 Early
Statistical Papers of J. Neyman, 98–141.
Neyman, J. (1937). ‘Outline of a Theory of Statistical Estimation Based on the Classical Theory of Probability’,
Philosophical Transactions of the Royal Society of London Series A 236(767), 333–80. Reprinted 1967 in Early
Statistical Papers of J. Neyman, 250–90.

78
Neyman, J. (1941). ‘Fiducial Argument and the Theory of Confidence Intervals’, Biometrika 32(2), 128–50.
Reprinted 1967 in Early Statistical Papers of J. Neyman: 375–94.
Neyman , J. (1956). Note on an Article by Sir Ronald Fisher, J R Stat Soc (B) 18: 288-294.
Neyman, J. (1957). ‘“Inductive Behavior” as a Basic Concept of Philosophy of Science’, Revue de l‘Institut
International de Statistique/Review of the International Statistical Institute 25(1/3), 7–22.
Neyman, J. (1976). ‘Tests of Statistical Hypotheses and Their Use in Studies of Natural Phenomena’,
Communications in Statistics: Theory and Methods 5(8), 737–51.
Neyman, J. and Pearson, E. (1928). ‘On the Use and Interpretation of Certain Test Criteria for Purposes of
Statistical Inference: Part I’, Biometrika 20A(1/2), 175–240. Reprinted in Joint Statistical Papers, 1–66.
Neyman, J. and Pearson, E. (1933). On the problem of the most efficient tests of statistical hypotheses.
Philosophical Transactions of the Royal Society of London Series A 231, 289–337. Reprinted in Joint Statistical
Papers, 140–85.
Neyman, J. & Pearson, E. (1967). Joint statistical papers of J. Neyman and E. S. Pearson. University of California
Press.
Pearson, E., (1955). Statistical Concepts in Their Relation to Reality, J R Stat Soc (B) 17: 204-207.
Pearson, E., (1970). The Neyman Pearson Story: 1926–34. In Pearson, E. and Kendall, M. (eds.), Studies in
History of Statistics and Probability, I. London: Charles Griffin & Co., 455–77.
Popper, K. (1962). Conjectures and Refutations: The Growth of Scientific Knowledge. New York: Basic Books.
Recht, B. (current). Blog: Arg min (https://www.argmin.net/).
Reid, C. (1998). Neyman. New York: Springer Science & Business Media.
Savage, L. J. (1962). The Foundations of Statistical Inference: A Discussion. London: Methuen.
Shaffer, J. (1995). Multiple hypothesis testing. Ann. Rev. Psychol. 46: 561-84.

79
Simmons, J. Nelson, L. and Simonsohn, U. (2018). False-positive citations. Perspectives on Psychological
Science 13(2) 255–259.
Stark, P. B. (2022). Reproducibility, p‐values, and type III errors: response to Mayo (2022). Conservation
Biology, 36(5), e13986-n/a. https://doi.org/10.1111/cobi.13986
Stark, P.B. (2022). Pay No Attention to the Model Behind the Curtain. Pure Appl. Geophys. 179, 4121–4145.
https://doi.org/10.1007/s00024-022-03137-2
Suppes, P. (1969). Models of Data, in Studies in the Methodology and Foundations of Science, Dordrecht, The
Netherlands: D. Reidel, pp. 24–35.
Wasserstein, R., & Lazar, N. (2016). The ASA’s statement on p-values: Context, process and purpose (and
supplemental materials). The American Statistician, 70(2), 129–133.
https://doi.org/10.1080/00031305.2016.1154108
Wasserstein, R., Schirm, A., & Lazar, N. (2019). Moving to a world beyond “p < 0.05” (Editorial). The American
Statistician 73(S1), 1–19. https://doi.org/10.1080/00031305.2019.1583913
Yu, B. and Barter, R. (2024).Veridical Data Science: The Practice of Responsible Data Analysis and Decision
Making Adaptive Computation and Machine Learning series, The MIT Press.(https://vdsbook.com/)

Severity as a basic concept in philosophy of statistics

More Related Content

Similar to Severity as a basic concept in philosophy of statistics (20)

More from jemille6 (20)

Recently uploaded (20)

Severity as a basic concept in philosophy of statistics