How should we adjust our beliefs in light of the testimony of those who are in a better epistemic position than ourselves, such as experts and other epistemic superiors? In this paper, I develop and defend a deference model of epistemic authority. The paper attempts to resolve the debate between the preemption view and the total evidence view of epistemic authority by taking an accuracy-first approach to the issue of how we should respond to authoritative and expert testimony. I argue (...) that when we look at the debate through the lens of accuracy, it becomes clear that matters are more complicated than either the preemption view or the total evidence view are able to account for. Consequently, a deference model, outlined within a credence-based framework, does a better job of capturing the relevant phenomena, and explaining how we should update our beliefs in response to epistemically authoritative testimony. (shrink)

In the biomedical context, policy makers face a large amount of potentially discordant evidence from different sources. This prompts the question of how this evidence should be aggregated in the interests of best-informed policy recommendations. The starting point of our discussion is Hunter and Williams’ recent work on an automated aggregation method for medical evidence. Our negative claim is that it is far from clear what the relevant criteria for evaluating an evidence aggregator of this sort are. What is the (...) appropriate balance between explicitly coded algorithms and implicit reasoning involved, for instance, in the packaging of input evidence? In short: What is the optimal degree of ‘automation’? On the positive side: We propose the ability to perform an adequate robustness analysis as the focal criterion, primarily because it directs efforts to what is most important, namely, the structure of the algorithm and the appropriate extent of automation. Moreover, where there are resource constraints on the aggregation process, one must also consider what balance between volume of evidence and accuracy in the treatment of individual evidence best facilitates inference. There is no prerogative to aggregate the total evidence available if this would in fact reduce overall accuracy. (shrink)

Many solutions of the Goodman paradox have been proposed but so far no agreement has been reached about which is the correct solution. However, I will not contribute here to the discussion with a new solution. Rather, I will argue that a solution has been in front of us for more than two hundred years because a careful reading of Hume’s account of inductive inferences shows that, contrary to Goodman’s opinion, it embodies a correct solution of the paradox. Moreover, the (...) account even includes a correct answer to Mill’s question of why in some cases a single instance is sufficient for a complete induction, since Hume gives a well-supported explanation of this reliability phenomenon. The discussion also suggests that Bayesian theory by itself cannot explain this phenomenon. Finally, we will see that Hume’s explanation of the reliability phenomenon is surprisingly similar to the explanation given lately by a number of naturalistic philosophers in their discussion of the Goodman paradox. (shrink)

Amalgamating evidence of different kinds for the same hypothesis into an overall confirmation is analogous, I argue, to amalgamating individuals’ preferences into a group preference. The latter faces well-known impossibility theorems, most famously “Arrow’s Theorem”. Once the analogy between amalgamating evidence and amalgamating preferences is tight, it is obvious that amalgamating evidence might face a theorem similar to Arrow’s. I prove that this is so, and end by discussing the plausibility of the axioms required for the theorem.

Some properties such as having a beard and being a philosopher are intuitively qualitative, while other properties such as being identical to Plato and being a student of Socrates are intuitively non-qualitative. It is often assumed that, necessarily, a property is qualitative if and only if it can be designated descriptively without the aid of directly referential devices. I argue that this linguistic thesis fails in both directions: there might be non-qualitative properties that can be designated descriptively, and there appear (...) to be qualitative properties that can only be designated directly. I conclude that while the linguistic thesis is ultimately untenable as stated, it can be plausibly recast as a thesis about our concepts rather than the properties they designate. (shrink)

The distinction between qualitative and non-qualitative properties should be familiar from discussions of the principle of the identity of indiscernibles: two otherwise exactly similar individuals, Castor and Pollux, might share all their qualitative properties yet differ with respect to their non-qualitative properties—for while Castor has the property being identical to Castor, Pollux does not. But while this distinction is familiar, there has not been much critical attention devoted to spelling out its precise nature. I argue that the class of non-qualitative (...) properties is broader than it is often taken to be. When properly construed, it will not only include properties such as being identical to Castor, which somehow make reference to particular individuals, it will also include more general properties such as identity, composition, set membership, as well as various peculiarly ontological properties. Given that some of these more general properties help to explain objective similarity, we have reason to believe that there are fundamental non-qualitative properties. (shrink)

Extant literature on Goodman’s ‘New Riddle of Induction’ deals mainly with two versions. I consider both of them, starting from the (‘epistemic’) version of Goodman’s classic of 1954. It turns out that it belongs to the realm of applications of inductive logic, and that it can be resolved by admitting only significant evidence (as I call it) for confirmations of hypotheses. Sect. 1 prepares some ground for the argument. As much of it depends on the notion of evidential significance, this (...) concept is defined and its introduction motivated. Further, I introduce and explain the distinction between support and confirmation: put in a slogan, ‘confirmation is support by significant evidence’. The second section deals with the Riddle itself. I demonstrate that, given the provisions of the first section, not ‘anything confirms anything’: significant green-evidence confirms only green-hypotheses (and no grue-hypotheses), and significant grue-evidence confirms only grue-hypotheses (and no green-hypotheses), whichever terms we use for expressing these evidences or hypotheses. The third section rounds off my treatment. First I show that Frank Jackson’s use of his counterfactual condition is unsuccessful. Further, I argue that no unwanted consequences result, if one starts from the other, ‘objective’, definition of ‘grue’, as it constitutes no more than a mere fact of logic that cannot do any harm. Finally, I present a grue-case involving both kinds of definition, where the exclusive confirmation of either the green- or the grue-hypothesis is shown. (shrink)

Amalgamating evidence from heterogeneous sources and across levels of inquiry is becoming increasingly important in many pure and applied sciences. This special issue provides a forum for researchers from diverse scientific and philosophical perspectives to discuss evidence amalgamation, its methodologies, its history, its pitfalls, and its potential. We situate the contributions therein within six themes from the broad literature on this subject: the variety-of-evidence thesis, the philosophy of meta-analysis, the role of robustness/sensitivity analysis for evidence amalgamation, its bearing on questions (...) of extrapolation and external validity of experiments, its connection with theory development, and its interface with causal inference, especially regarding causal theories of cancer. (shrink)

This paper is concerned with two concepts of qualitativeness that apply to intensional entities. I propose an account of pure qualitativeness that largely follows the traditional understanding established by Carnap, and try to shed light on its ontological presuppositions. On this account, an intensional entity is purely qualitative iff it does not ‘involve’ any particular. An alternative notion of qualitativeness—which I propose to refer to as a concept of strict qualitativeness—has recently been introduced by Chad Carmichael. However, Carmichael’s definition presupposes (...) a highly fine-grained conception of properties and relations. To eliminate this presupposition, I tentatively suggest a different definition that rests on a concept of perspicuous denotation. In the penultimate section, both concepts of qualitativeness are put to work in distinguishing between different ‘grades’ of qualitative discriminability. (shrink)

We give an account of some relationships between the principles of Constant and Atom Exchangeability and various generalizations of the Principle of Instantial Relevance within the framework of Inductive Logic. In particular we demonstrate some surprising and somewhat counterintuitive dependencies of these relationships on ostensibly unimportant parameters, such as the number of predicates in the overlying language.

The problem of collecting, analyzing and evaluating evidence on adverse drug reactions (ADRs) is an example of the more general class of epistemological problems related to scientific inference and prediction, as well as a central problem of the health-care practice. Philosophical discussions have critically analysed the methodological pitfalls and epistemological implications of evidence assessment in medicine, however they have mainly focused on evidence of treatment efficacy. Most of this work is devoted to statistical methods of causal inference with a special (...) attention to the privileged role assigned to randomized controlled trials in Evidence Based Medicine. Regardless of whether the RCT’s privilege holds for efficacy assessment, it is nevertheless important to make a distinction between causal inference of intended and unintended effects, in that the unknowns at stake are heterogonous in the two contexts. This point has been emphasized by epidemiologists in the last decade. Their main focus is methodological, and regards the fact that bias and confounding do not affect studies on intended and unintended effects in the same way. However, deeper concerns ground the intuition for such a distinction; these are related to the constraints which we impose on evidence and their epistemological justification. My thesis is that such constraints ought to be understood to be different in the case of evidence for risk vs. benefit assessment. I present the recent debate on the causal association between acetaminophen and asthma in order to illustrate the point at issue. (shrink)

Philosophical discussions have critically analysed the methodological pitfalls and epistemological implications of evidence assessment in medicine, however they have mainly focused on evidence of treatment efficacy. Most of this work is devoted to statistical methods of causal inference with a special attention to the privileged role assigned to randomized controlled trials (RCTs) in evidence based medicine. Regardless of whether the RCT’s privilege holds for efficacy assessment, it is nevertheless important to make a distinction between causal inference of intended and unintended (...) effects, in that the unknowns at stake are heterogonous in the two contexts. However, although “lower level” evidence is increasingly acknowledged to be a valid source of information contributory to assessing the risk profile of medications on theoretical or empirical grounds, current practices have difficulty in assigning a precise epistemic status to this kind of evidence because they are more or less implicitly parasitic on the (statistical) methods developed to test drug efficacy. My thesis is that (1) “lower level” evidence is justified on distinct grounds and at different conditions depending on the different epistemologies which one wishes to endorse, in that each impose different constraints on the methods we adopt to collect and evaluate evidence; (2) such constraints ought to be understood to be different in the case of evidence for risk versus benefit assessment for a series of reasons which I will illustrate on the basis of the recent debate on the causal association between acetaminophen (a.k.a. paracetamol) and asthma. (shrink)

In this paper, I consider Goodman’s new riddle of induction and how we should best respond to it. Noticing that all the emeralds so far observed are green, we infer that all emeralds are green. However, all emeralds so far observed are also grue, so we could also infer that they are grue. Only one of these inductive inferences or projections could, however, be valid. For the hypothesis that all emeralds are green predicts that the next observed emerald will be (...) green; whereas the hypothesis that they are grue predicts that it will blue. Goodman’s new riddle is the problem of saying why the inductive inference involving “green” is the valid one. Goodman’s own solution appeals to the idea of entrenchment. His idea is that “green” is a more entrenched predicate than “grue” in the sense that it has figured many more times in our past projections than has “grue”. In his view, this explains why “green” is projectible whereas “grue” isn’t. I argue that this response doesn’t go far enough and that we additionally need an explanation of why “green” is more entrenched than “grue”—that we are otherwise left with the unsatisfactory view that its superior entrenchment is a mere linguistic accident. I try to supplement Goodman’s solution with an explanation of this kind. I argue that “grue” is not entrenched be- cause past successful inductions involving “green” show that past projections that could have been made using what I call “grue-like” predicates—predicates which are like “grue” except that the times featuring in their definitions are past—would have been unsuccessful. (shrink)

Philosophical discussions on causal inference in medicine are stuck in dyadic camps, each defending one kind of evidence or method rather than another as best support for causal hypotheses. Whereas Evidence Based Medicine advocates the use of Randomised Controlled Trials and systematic reviews of RCTs as gold standard, philosophers of science emphasise the importance of mechanisms and their distinctive informational contribution to causal inference and assessment. Some have suggested the adoption of a pluralistic approach to causal inference, and an inductive (...) rather than hypothetico-deductive inferential paradigm. However, these proposals deliver no clear guidelines about how such plurality of evidence sources should jointly justify hypotheses of causal associations. We here develop such guidelines by first giving a philosophical analysis of the underpinnings of Hill’s viewpoints on causality. We then put forward an evidence-amalgamation framework adopting a Bayesian net approach to model causal inference in pharmacology for the assessment of harms. Our framework accommodates a number of intuitions already expressed in the literature concerning the EBM vs. pluralist debate on causal inference, evidence hierarchies, causal holism, relevance, and reliability. (shrink)

Charles Taber and Milton Lodge provide compelling evidence that people's minds may be closed to information that is inconsistent with their prior beliefs. This type of inconsistency has often been termed ?irrational.? However, recent research suggests that being open or closed minded is not an unchanging variable but depends on one's goals, including one's need for closure, which vary from person to person and situation to situation. In this vein, as Taber and Lodge suggest, those who have more political information (...) may, after having been open minded enough to acquire the information, become closed minded by virtue of having acquired it. At the same time, being more open minded to inaccurate information might lead one in the wrong direction; hence, open mindedness does not necessarily enhance the accuracy of one's judgments or the quality of one's decisions. We may need to rethink the concept of an enlightened, well-informed electorate, to the extent that it assumes the unqualified benefits of open mindedness. (shrink)

The extensions of Goodman’s ‘grue’ predicate and Kripke’s ‘quus’ are constructed from the extensions of more familiar terms via a reinterpretation that permutes assignments of reference. Since this manoeuvre is at the heart of Putnam’s model-theoretic and permutation arguments against metaphysical realism (‘Putnam’s Paradox’), both Goodman’s New Riddle of Induction and the paradox about meaning that Kripke attributes to Wittgenstein are instances of Putnam’s. Evidence cannot selectively confirm the green-hypothesis and disconfirm the grue-hypothesis, because the theory of which the green-hypothesis (...) is a part has an unintended model in which the grue-hypothesis is equally confirmed; and there are no meaning-facts that determine reference, because the objects referred to by the referring terms of any language or set of intentional mental states are permutable in a way that is consistent with the truth-values of all other sentences in that language or beliefs in that set. The upshot is that the three paradoxes need to be solved in a unified way. (shrink)

This paper argues for a novel account of deceitful scientific communication, as “wishful speaking”. This concept is of relevance both to philosophy of science and to discussions of the ethics of lying and misleading. Section 1 outlines a case-study of “ghost-managed” research. Section 2 introduces the concept of “wishful speaking” and shows how it relates to other forms of misleading communication. Sections 3–5 consider some complications raised by the example of pharmaceutical research; concerning the ethics of silence; how research strategies—as (...) well as the communication of results—may be misleading; and questions of multiple authorship. The conclusion suggests some more general conclusions. (shrink)

The use of the Law of Likelihood (LL) as a general tool for assessing rival hypotheses has been criticized for its ambiguous treatment of background information. The LL endorses radically different answers depending on what information is designated as background versus evidence. I argue that once one distinguishes between two questions about evidentiary support, the ambiguity vanishes. I demonstrate this resolution by applying it to a debate over the status of the ‘fine-tuning argument’.

Yehezkel presents an interesting argument to resolve Goodman’s New Riddle of Induction, and in effect to claim that a purely syntactical theory of confirmation is possible. In this paper I suggest that Yehezkel’s argument relies on two premises not proven in his paper. The first premise is that if “a New Riddle of Deduction” can be formulated then there is no New Riddle of Induction. This premise seems to be wrong as I claim in part 2 of this paper. The (...) second premise is that a New Riddle of Deduction can be formulated. Despite Yehezkel’s efforts to prove this claim, it seems to me false, or at least, unproven. I therefore conclude that there is no New Riddle of Deduction, and therefore, no actual resolution to the New Riddle of Induction. (shrink)

Subjective probability plays an increasingly important role in many fields concerned with human cognition and behavior. Yet there have been significant criticisms of the idea that probabilities could actually be represented in the mind. This paper presents and elaborates a view of subjective probability as a kind of sampling propensity associated with internally represented generative models. The resulting view answers to some of the most well known criticisms of subjective probability, and is also supported by empirical work in neuroscience and (...) behavioral psychology. The repercussions of the view for how we conceive of many ordinary instances of subjective probability, and how it relates to more traditional conceptions of subjective probability, are discussed in some detail. (shrink)

The paper shows why and how an empirical study of fast-and-frugal heuristics can provide norms of good reasoning, and thus how (and how far) rationality can be naturalized. We explain the heuristics that humans often rely on in solving problems, for example, choosing investment strategies or apartments, placing bets in sports, or making library searches. We then show that heuristics can lead to judgments that are as accurate as or even more accurate than strategies that use more information and computation, (...) including optimization methods. A standard way to defend the use of heuristics is by reference to accuracy-effort trade-offs. We take a different route, emphasizing ecological rationality (the relationship between cognitive heuristics and environment), and argue that in uncertain environments, more information and computation are not always better (the “less-can-be-more” doctrine). The resulting naturalism about rationality is thus normative because it not only describes what heuristics people use, but also in which specific environments one should rely on a heuristic in order to make better inferences. While we desist from claiming that the scope of ecological rationality is unlimited, we think it is of wide practical use. (shrink)

Although the category “race” fails as a postulated natural kind, racial, ethnic, national, linguistic, religious, and other group designations might nonetheless be considered projectible insofar as they support inductive inferences in biomedicine. This article investigates what it might mean for group concepts in population genetics and genomics to be projectible and whether the projectibility of such predicates licenses the representation of their corresponding classes as natural kinds according to currently prevailing projectibility-based accounts of natural kinds. The article draws on a (...) case study from cancer genetics, specifically, breast cancer risk in Ashkenazi Jewish women. (shrink)

It is an ongoing controversy whether natural selection is a cause of population change, or a mere statistical description of how individual births and deaths accumulate. In this paper I restate the problem in terms of the reference class problem, and propose how the structure of stable equilibrium can provide a solution in continuity with biological practice. Insofar natural selection can be understood as a tendency towards equilibrium, key statisticalist criticisms are avoided. Further, in a modification of the Newtonian-force analogy, (...) it can be suggested that a better metaphor for natural selection is that of an emergent force, similar in nature to entropic forces: with magnitude and direction, but lacking a spatiotemporal origin or point of application. (shrink)

In science and everyday life, we often infer that something is true because it would explain some set of facts better than any other hypothesis we can think of. But what if we have reason to believe that there is a better way to explain these facts that we just haven't thought of? Wouldn't that undermine our warrant for believing the best available explanation? Many philosophers have assumed that we can solve such underconsideration problems by stipulating that a hypothesis should (...) not only be 'the best' explanation available; rather, it should also be 'good enough'. Unfortunately, however, the only current suggestion for what it might mean to say that an explanation is 'good enough' is, well, not good enough. This paper aims to provide a better account of what is required for an explanatory hypothesis to be considered 'good enough'. In brief, the account holds that a `good enough' hypothesis is one that has gone through a process that I call explanatory consolidation, in which accumulating evidence and failed attempts to formulate better alternatives gradually make it more plausible that the explanation we currently have is better than any other that could be formulated. (shrink)

Historically, Nelson Goodman’s paradox involving the predicates ‘grue’ and ‘bleen’ has been taken to furnish a serious blow to Carl Hempel’s theory of confirmation in particular and to purely formal theories of confirmation in general. In this paper, I argue that Goodman’s paradox is no more serious of a threat to Hempel’s theory of confirmation than is Hempel’s own paradox of the ravens. I proceed by developing a suggestion from R. D. Rosenkrantz into an argument for the conclusion that these (...) paradoxes are, in fact, equivalent. My argument, if successful, is of both historical and philosophical interest. Goodman himself maintained that Hempel’s theory of confirmation was capable of handling the paradox of the ravens. And Hempel eventually conceded that Goodman’s paradox showed that there could be no adequate, purely syntactical theory of confirmation. The conclusion of my argument entails, by contrast, that Hempel’s theory of confirmation is incapable of handling Goodman’s paradox if and only if it is incapable of handling the paradox of the ravens. It also entails that for any adequate solution to one of these paradoxes, there is a corresponding and equally adequate solution to the other. (shrink)

It is commonly accepted that people can, and regularly do, deceive themselves. Yet closer examination reveals a set of conceptual puzzles that make self-deception difficult to explain. Applying the conditions for other-deception to self-deception generates what are known as the ‘paradoxes’ of belief and intention. Simply put, the central problem is how it is possible for me to believe one thing, and yet intentionally cause myself to simultaneously believe its contradiction. There are two general approaches taken by philosophers to account (...) for these puzzles about the self-deceptive state and the process of self-deception. ‘Partitioning’ strategies try to resolve the paradoxes by proposing that the mind is divided in some way that allows self-deception to occur. ‘Reformulation’ strategies suggest that the conditions we use to define self-deception should be modified so that the paradoxes do not arise at all. Both approaches are subject to criticism about the consequences of the strategies philosophers use, but recent cross-disciplinary analyses of self-deception may help shed light on the puzzles that underlie this phenomenon. (shrink)

Supra-Bayesianism is the Bayesian response to learning the opinions of others. Probability pooling constitutes an alternative response. One natural question is whether there are cases where probability pooling gives the supra-Bayesian result. This has been called the problem of Bayes-compatibility for pooling functions. It is known that in a common prior setting, under standard assumptions, linear pooling cannot be nontrivially Bayes-compatible. We show by contrast that geometric pooling can be nontrivially Bayes-compatible. Indeed, we show that, under certain assumptions, geometric and (...) Bayes-compatible pooling are equivalent. Granting supra-Bayesianism its usual normative status, one upshot of our study is thus that, in a certain class of epistemic contexts, geometric pooling enjoys a normative advantage over linear pooling as a social learning mechanism. We discuss the philosophical ramifications of this advantage, which we show to be robust to variations in our statement of the Bayes-compatibility problem. (shrink)

This thesis is about making decisions when we are uncertain about what will happen, how valuable it will be, and even how to make decisions. Even the most sure-footed amongst us are sometimes uncertain about all three, but surprisingly little attention has been given to the latter two. The three essays that constitute my thesis hope to do a small part in rectifying this problem. The first essay is about the value of finding out how to make decisions. Society spends (...) considerable resources funding people (like me) to research decision-making, so it is natural to wonder whether society is getting a good deal. This question is so shockingly underresearched that bedrock facts are readily discoverable, such as when this kind of information is valuable. My second essay concerns whether we can compare value when we are uncertain about value. Many people are in fact uncertain about value, and how we deal with this uncertainty hinges on these comparisons. I argue that value comparisons are only sometimes possible; I call this weak comparability. This essay is largely a synthesis of the literature, but I also present an argument which begins with a peculiar view of the self: it is as if each of us is a crowd of different people separated by time (but connected by continuity of experience). I’m not the first to endorse this peculiar view of the self, but I am the first to show how it supports the benign view that value is sometimes comparable. We may be uncertain of any decision rules, even those that would tell us how to act when we face uncertainty in decision rules. We may be uncertain of how to decide how to decide how to... And so on. If so, we might have to accept infinitely many decision rules just to make any mundane decision, such as whether to pick up a five-cent piece from the gutter. My third essay addresses this problem of regress. I think all of our decisions are forced: we must decide now or continue to deliberate. Surprisingly, this allows us to avoid the original problem. I call this solution “when forced, do your best”. (shrink)

This paper surveys and critically assesses existing theories of evidence with respect to four desiderata. A good theory of evidence should be both a theory of evidential support (i.e., be informative about what kinds of facts speak in favour of a hypothesis), and of warrant (i.e., be informative about how strongly a given set of facts speaks in favour of the hypothesis), it should apply to the non-ideal cases in which scientists typically find themselves, and it should be ‘descriptively adequate’, (...) i.e., able to adequately represent typical episodes of evidentiary reasoning. The theories surveyed here—Bayesianism, hypotheticodeductivism,satisfaction theories, error statistics as well as Achinstein’s and Cartwright’s theories—are all found wanting in important respects. I finally argue that a deficiency all these theories have in common is a neglect or underplaying of the epistemic context in which the episode of evidentiary reasoning takes place.Este artículo describe y valora críticamente diversas teorías de la evidencia en relación a cuatro desiderata. Una buena teoría de la evidencia debería ser tanto una teoría sobre el apoyo evidencial [evidential support] (ser informativa sobre qué tipos de hechos hablan a favor de la hipótesis) como sobre la justificación [warrant]; debería aplicarse en las situaciones no ideales en las que normalmente se encuentran los científicos; y debería ser ‘descriptivamente adecuada’, esto es, capaz de representar correctamente episodios típicos de razonamiento evidencial. Las teorías aquí revisadas—bayesianismo, hipotético-deductivismo, teorías de la satisfacibilidad, la estadística del error, así como las propuestas de Achinstein y Cartwright—se consideran deficientes en aspectos básicos. Argumentaré que un defecto común en todas ellas es que olvidan, o minusvaloran, el contexto epistémico en el que el episodio de razonamiento evidencial tiene lugar. (shrink)

The thesis consists of an introduction and five papers. The introduction gives a brief historical survey of empirical investigations into the effectiveness of medicinal interventions, as well as surveys of the concept of evidence and of the history and philosophy of experiments. The main ideas of the EBM movement are also presented. Paper I: Concerns have been raised that clinical trials do not offer reliable evidence for some types of treatment, in particular for highly individualised treatments, for example traditional homeopathy. (...) With respect to individualised treatments, it is argued that such concerns are unfounded. There are two minimal conditions related to the nature of the treatments that must be fulfilled for evaluability in a clinical trial, namely the proper distinction of treatment groups and the elimination of confounding variables or variations. These conditions do not preclude the testing of individualised medicine. Paper II: Traditionally, mechanistic reasoning has been assigned a negligible role in the EBM literature. When discussed, mechanistic reasoning has almost exclusively been positive – both in an epistemic sense of claiming that there is a mechanistic chain and in a health-related sense of there being claimed benefits for the patient. Negative mechanistic reasoning has been neglected. I distinguish three main types of negative mechanistic reasoning and subsume them under a new definition. One of the three distinguished types, which is negative only in the health-related sense, has a corresponding positive counterpart, whereas the other two, which are epistemically negative, do not have such counterparts, at least not that are particularly interesting as evidence. Accounting for negative mechanistic reasoning in EBM is therefore partly different from accounting for positive mechanistic reasoning. Paper III: Evidence hierarchies are lists of investigative strategies ordered with regard to the claimed strength of evidence. They have been used for a couple of decades within EBM, particularly for the assessment of evidence for treatment recommendations, but they remain controversial. An under-investigated question is what the order in the hierarchy means. Four interpretations of the order are distinguished and discussed. The two most credible ones are, in rough terms, “typically stronger” and “ideally stronger”. The GRADE framework seems to be based on the “typically stronger” reading. Even if the interpretation of an evidence hierarchy were established, hierarchies appear to be rather unhelpful for the task of evidence aggregation. However, specifying the intended order relation may help sort out disagreements. Paper IV: There are three main arguments for randomisation that connect inseparably to theoretical concepts: Randomisation is useful for performing null hypothesis testing. Randomisation is needed for plausible causal inferences from treatment to effect. Randomisation is acceptable and computationally convenient in a Bayesian setting. A critical scrutiny of these arguments shows that is acceptable in the context of clinical trials. As for, it is argued that randomisation only provides weak reasons for drawing causal inferences in the context of real clinical trials. Argument is weak because it is controversial among Bayesians, and because formally Bayesian analyses of trial results are rarely asked for. Paper V: Practical arguments for randomisation are arguments with no necessary connections to theoretical frameworks like null hypothesis testing or causal inferences. Four common practical arguments in the context of clinical trials are distinguished and assessed: Randomisation contributes to allocation concealment. Randomisation contributes to the baseline balance of treatment groups. Randomisation decreases self-selection bias. Randomisation removes allocation bias. Argument is rejected. Arguments and are approved. Argument is rejected if it is formulated so as to be independent from and, but it is true that randomisation contributes to balance through the mechanisms mentioned in and. It is judged that may be the strongest single argument. (shrink)

Policy-makers face an uncertain world. One way of getting a handle on decision-making in such an environment is to rely on evidence. Despite the recent increase in post-fact figures in politics, evidence-based policymaking takes centre stage in policy-setting institutions. Often, however, policy-makers face large volumes of evidence from different sources. Robustness analysis can, prima facie, handle this evidential diversity. Roughly, a hypothesis is supported by robust evidence if the different evidential sources are in agreement. In this thesis, I strengthen the (...) case for the use of robustness analysis in evidence-based policymaking by answering open research questions about this inference technique. First, I argue that existing taxonomies miss a fruitful category of robustness reasoning, that is predictive stability. Second, I claim that derivational robustness analysis – the investigation of whether the results of different models are in agreement – can yield interesting insights even if not the entire relevant model space is covered by available models or if the model results are only partially in agreement. Third, I claim that expert knowledge is necessary to address questions that arise when one applies measurement robustness analysis – the investigation into whether multiple means of measurement yield the same result. Finally, I argue that, in situations where evidence from different measurements is not in agreement, it can be advisable to no longer take all of the evidence into account. This can be done in a rationally defensible way by choosing the most adequate theory or model underlying parts of the evidence set. I discuss examples from climate, medical, and economic policy-making to establish my claims. (shrink)