This blog post series explores methodological questions and challenges in the use of historical research by Sentience Institute, the effective altruism community, and the farmed animal movement. It is broader in scope, more theoretical, and more meta than most of Sentience Institute’s research, but I have tried to regularly bring the discussion back to concrete examples (e.g. WWI and the French Revolution) with concrete implications for advocates. The other posts are on “How tractable is changing the course of history?” and “How is SI research different from existing social movement literature and relevant historical works?”
Edited by Jacy Reese Anthis. Many thanks to Kelly Anthis, J. Mohorčich, Samara Mendez, Steven Rouk, Kelly McNamara, and Carl Shulman for reviewing and providing feedback.
The effective altruism community has looked to historical evidence from previous social movements to inform their tactics and intervention prioritization. Groups considering historical evidence include Open Philanthropy Project, the Centre for Effective Altruism, the Future of Humanity Institute, Animal Charity Evaluators, Sentience Institute, and individuals. Animal advocates more widely have often done the same, such as Direct Action Everywhere, which has many blog posts citing social movements as evidence, Steven Wise and the Nonhuman Rights Project, who have looked at antislavery measures for specific legal precedents, and individual advocates. Advocates frequently cite historical issues and movements to draw support for their cause, such as with the frequent quoting of respected figures and leaders from past social movements, though this is usually not done in reference to the effectiveness of particular strategies. There is ample room in this space for more extensive research that more rigorously analyzes causation and strategic implications.
Sentience Institute’s 2017 survey of effective animal advocacy (EAA) researchers found moderately weaker to moderately greater valuation of social movement evidence than other types of evidence for answering foundational questions in EAA, but individual answers varied widely. This suggests that a deeper look at the questions, “To what extent can we actually glean significant strategic knowledge from history?” and, “How much weight should the EAA community place on insights from history?” are important for EAA intervention prioritization and strategic decision-making. Hereafter, terms such as “strategic knowledge” and “insights” from history will refer primarily to information from history that can be used to answer foundational questions in EAA and to understand how to maximize the impact of interventions used in the farmed animal movement, although the considerations in this post are relevant other applications of knowledge from history.
Philosophers have argued that conclusively proving historical causation is impossible, but most historians have continued to posit causal relationships. This post reflects on four historical case studies which are analyzed in greater depth in a previous post on How tractable is changing the course of history? In each of the case studies considered, historians have been unable to reach consensus on the presence or absence of certain factors, and struggled even more to agree about the relative importance of these factors in causing particular outcomes. These disagreements, as well as the inherent difficulty in forming accurate judgements about causation suggested by hindsight bias and chaos theory, suggest that we should not place too much weight on hypothesized historical cause and effect relationships in general. We can of course be more or less confident in hypotheses about causation in specific case studies and rigorous use of counterfactual analyses may lead to firmer conclusions. Given that the conclusions we form from history about causal relationships will always be quite uncertain, this should reduce the weight that we place on the strategic knowledge gained from any individual historical case study. Historical evidence from individual case studies might still be useful in providing inspiration for potential advocacy actions and building intuitions, however.
Correlations that occur more than once in a similar form across historical contexts and movements—such as when certain tactics seem successful across multiple movements—provide much stronger evidence in support of hypothesized cause and effect relationships. The larger the dataset, the more confidence we should place in significant findings. Studies such as William Gamson’s Strategy of Social Protest can show us which internal characteristics of social movement organizations (SMOs) seem to increase the likelihood of successful outcomes, even if causation cannot be proven. Quantitative and comparative studies of the importance of factors external to SMOs can also increase our confidence in conclusions about the importance of broad factors in causing social change, at least for specific historical movements. They can also increase our confidence in conclusions about the causes of social change across contexts (see “External constraints on the impact of social movements” in “How tractable is changing the course of history?”), although empirical evidence from sociologists suggesting that such generalized theories of causation do not necessarily hold consistently should reduce the weight that we place on such studies.
Given the varying intuitions and existing views of people on the topic of historical tractability, we can’t present conclusions from this work that would apply to all readers, so this section instead specifically details how Jamie Harris (author) and Jacy Reese (editor) changed their views during the course of this project.
Sentience Institute staff had slightly different intuitions about the questions, “To what extent can we actually glean significant strategic knowledge from history?” and, “How much weight should the EAA community place on insights from history?” My views about these key questions did not change significantly over the course of my research into them, although my views about smaller issues within these key questions did.
I will now place significantly less weight on knowledge from most individual, historical case studies after becoming aware of theoretical arguments and psychological evidence that I had been unaware of. My previous intuition (that there was usually sufficient evidence contained within historians' narratives to mostly justify their seemingly confident views about causal relationships in particular historical case studies) was challenged by reanalyzing the narratives of historians in light of the theoretical arguments and psychological evidence. In contrast, I will now place more weight on knowledge from comparative analyses across countries, states, or movements. This is because after reflecting on the strengths and weaknesses of the different types of evidence available, I became more confident that when correlations between movement tactics and outcomes are replicated across multiple social movements they will replicate to the context of the farmed animal movement. I was also impressed by some of the quantitative and comparative methods used by sociologists that I was previously unaware of, as well as the potential for counterfactual analysis in history, to provide stronger evidence on historical causation.
This suggests that where trade-offs arise in our use of time, SI and other organizations using historical evidence should prioritize analyzing where correlations occur across multiple movements over trying to reach firmer conclusions about causation in individual movements.
Jacy Reese, our Research Director who edited this article, did not discernibly change his views on the two key questions we discussed beforehand: how easily we can glean strategic knowledge from history research and how much weight historical knowledge should be given compared to other research results (e.g. RCTs). However, Jacy did change his views on historical tractability based on the first of these posts, “How tractable is it for thoughtful actors to change the course of history?” He says this was because the first post had more empirical research results and more discussion of scholarly opinion where Jacy was uncertain about the extent of agreement.
See this spreadsheet for quantitative versions of these views, before and after this research project.
One factor that should change the weight that we place on strategic knowledge from history is the amount of confidence that we have in our hypotheses about causal relationships in any individual case study. For example, looking at the American civil rights movement, we might hypothesize that the March on Washington in August 1963 was crucial in driving through the passage of the Civil Rights Act of 1964. If this hypothesis is correct, it suggests that peaceful mass demonstrations are an effective way to encourage beneficial legislation. But if we are unsure whether the march itself, other contextual factors, other peaceful demonstrations, or more violent resistance were more important in causing the passage of the Act, then this is a less robust strategic implication.
E. J. Tapp argued that causation “is the great central pillar, not only of historical thought, but of all thinking… Without it there could be no history, in fact no means of knowing anything, still less a past.” Tapp argued for the consideration of “necessary and sufficient conditions” to explain historical causation. Previous philosophers like Hume had argued that proving causation was impossible, yet subsequent thinkers have continued to consider causation in history or to propose epistemic methods to achieve consensus on historical disagreements, including over causation. While causation cannot be proven in history, correlations offer some evidence of causation, patterns of correlations increase the weight of this evidence, and our speculation about counterfactual timelines and possible chains of causation (e.g. if a legislator votes for an important bill and states that it was because of a certain book they read) can increase the weight of our evidence further. So we can still arrive at weak views of causation from such evidence and speculation but need to be careful to not overstate our confidence that causation exists in particular relationships.
Some historians see the contributions of philosophers to historical discussion as so theoretical to be meaningless for the study of history. Others, reflecting philosophers’ doubts about the tractability of establishing causation, have avoided in-depth analyses of causation. Historian Richard J. Evans analyzes some such writings and notes that some consideration of causation is inevitable in history, even if historians do not tend to take this so far as to focus on the role of “the accidental.” Many historians accept that they cannot prove causation, but can use evidence to reach relatively high confidence in their conclusions. Others have implied that drawing clear conclusions from the past is too difficult for learning from the past to be achievable, such as A. J. P. Taylor’s remark that “the only safe explanation in history is that things happen because they happen.” Note that this discussion almost always lacks quantification, so it’s difficult to know whether or to what extent historians really disagree on the topic.
A number of sociologists have written about the difficulties in assessing the impact of social movements, reflecting the issues faced by historians and noted by philosophers in distinguishing the extent to which particular factors were responsible for particular changes. Even after rigorous data collection and detailed analysis, it may be difficult to be certain of causation in correlations between certain factors and outcomes.
“Hindsight bias” is “the tendency for people with outcome knowledge to believe falsely that they would have predicted the reported outcome of an event.” The validity of this effect was confirmed in a variety of experiments, including those trying to avoid the restrictions of laboratory settings. However, some research has found that hindsight bias exerts little impact on our views: Jay Christensen-Szalanski and Cynthia Willham’s meta-analysis showed that “the average weighted effect size of all 122 studies was r = .17.” The effect is perhaps smaller for experts, too, though not necessarily when assessing real historical outcomes. Unfortunately, knowledge of hindsight bias does not necessarily help us to avoid it, so the prevalence of this bias should slightly reduce our confidence in our ability to accurately assess historical causation.
We should be wary of the causal relationships implied by the coherent narratives that historians tend to create despite sometimes lacking convincing supporting evidence (see “Consideration of causation” in “How is SI research different from existing social movement literature and relevant historical works?”). Summarizing the implications of several psychological experiments and statistical analyses, Daniel Kahneman notes that “The confidence that individuals have in their beliefs depends mostly on the quality of the story they can tell about what they see, even if they see little.”
Social change is too complex for us to expect to account for all the potential causes, so in a sample of successful social movements, several might be successful only due to undiscoverable factors comparable to a butterfly flapping its wings. In these cases, there will essentially be no meaningful “causes” we can point to for their success, and failing to account for these unknown factors would be misleading.
To put it another way, if a dataset has a small number of data points, this is relatively weak evidence of trends and causal relationships; when trying to infer causation from a single historical case study, we are usually relying on a single data point in the face of many potential confounding variables.
The difficulty of being able to confidently establish and predict causation is attested by chaos theory, which suggests that in certain complex systems “there are limits to knowledge and prediction because some futures may be unknowable with any precision.” Human interaction across varying contexts and time periods is always complex, and in such complex systems, a small change in initial conditions can spark a chain of events which leaves the final outcomes almost unpredictable. The implication of this for history is that patterns of causation may exist, but with so many variables and interactions between these variables, clearly assessing the importance of each of these on outcomes is nearly impossible, as is making confident predictions based on them. Thus, chaos theory implies that we should slightly reduce our confidence in judgements about causation, although it is important to remember that this mathematical theory does not necessarily translate to historical situations.
In spite of these theoretical difficulties, historians have continued to analyze the causes of events and place them in hierarchies of importance. Looking to specific historical debates provides opportunity to assess the extent to which historians have been able to reach consensus about the causes of change in well-studied contexts. While “consensus in a uniquely heterogeneous, large, and uncoerced group of historians is a likely indicator of knowledge,” an inability for historians to reach consensus would reduce the confidence that we should place in historical evidence. However, one’s view on how much it should reduce confidence depends on how much one expects historians to reliably develop consensus when there exists compelling evidence. For example, several of us at Sentience Institute share the intuition that academic historians are strongly incentivized to contest rather than confirm prevailing interpretations; the first is more likely to garner publications and tenure. This means, somewhat paradoxically, that areas of study with scant total literature on them are more likely to have consensus, while areas with abundant coverage are more likely to have abundant competing interpretations. This suggests that an inability for historians to reach consensus should not reduce the confidence that we place in historical evidence by much—it should reduce the confidence that we place in assessments by historians.
Nick Beckstead questioned historians about the extent to which the industrial revolution and the “Great Divergence" in development between Europe and China were inevitable. Beckstead found “no consensus about which of these factors played an essential role, or even that all potentially essential factors have been identified," and quite dramatic disagreement between expert historians about the extent to which general aspects of the industrial revolution were inevitable.
Looking at the historical debate on the causes of World War I (see the brief case study in our blog post on "How tractable is changing the course of history?”) leads to the potentially unsettling conclusion that even after a century of focused debate, comprising the combined career focus of many different academics from many different countries, historians have been unable to reach widespread, stable agreement about the relative importance of the different factors that led to the war. Debates over this issue have been quite volatile, and historians have still not reached full consensus, even if a slight trend towards contingency can be discerned. Of course, different historical topics have different amounts of disagreement or consensus on different issues, but the debate around the causes of World War I may be quite representative of the difficulties of considering causation in any complex, multicausal situation.
Similarly, there is no consensus on the most important causes of the French Revolution (see the brief case study in our blog post on "How tractable is changing the course of history?”), though explanations have variously lost or gained academic support. Here, the issue is more about tying together the various factors into a coherent explanatory framework (and hierarchy of significance) than about total rejection of any of the broad factors identified. The debate appears less cyclical than the debate over the causes of World War I. While no universally accepted hierarchy of causes has been established, there appears to be greater agreement that a multiplicity of causes were important in bringing about revolution than among the historians of the causes of World War I. Within this, there has been a move away from class conflict-focused aspects of Marxist explanations of the causes of the French Revolution, towards a more balanced approach which emphasizes political, institutional and cultural factors, without being dogmatic about their primacy, but some dissenting views persist. The historiography of the causes of the French Revolution evidences the difficulty of reaching firm conclusions or consensus about the relative importance of causal factors.
The complications of establishing causation in these historical debates also apply to the study of modern social movements. Relative to older movements, more information regarding modern social movements is likely to become available as time goes on and historians’ views are more likely to change. My impression from a limited reading of the literature on the rise of American environmentalism in the 1960s (see the brief case study in our blog post on "How tractable is changing the course of history?”) is that there are too few scholars who have explicitly and rigorously evaluated the comparative importance of its causes for there to be either strong divides or clear consensus within academia.
Comparing the extent of correlation between certain variables with successful outcomes for a particular movement across different states or countries seems likely to be one of the more rigorous methods for more confidently understanding the causes of social change in that movement. The stronger the correlation between the variable and successful outcomes, the more likely it is that this represents a causal relationship. Some scholars have evaluated the extent of such correlations; for example, Sarah A. Soule and Susan Olzak tested the importance of a variety of variables on the ratification of the Equal Rights Amendment in different U.S. states, 1972-82. They concluded that “elite allies may have not been all that important to this particular policy outcome,” except when it combined with the presence of campaigning organizations. Such conclusions seem useful for building an understanding of the causes of legislative success or failure in the women’s rights movement in America, even if we are unsure how useful such studies are for understanding the causes of success for social movements more generally (see the section on “How likely is it that correlations will replicate across movements” below).
Similar comparisons can be made across movements, within a particular country, to help understand the causes of social change in that particular country. This can be done either by examining in-depth case studies or through quantitative methods.
Of course, there are methodological pitfalls that might reduce the value of such comparative analyses. For example, Felix Kolb notes that “some quantitative studies have conceptualized and measured the political outcomes of a specific social movement so narrowly and one-dimensionally that they are likely to have missed important aspects.” We might be wary of such limitations when using comparative analyses to understand causation in particular social movements.
Analysis of counterfactuals has been done implicitly, but rarely explicitly and rigorously in historical and social scientific efforts to assess causation. An explicit counterfactual analysis would involve analyzing how outcomes would have been different if there was a slight variation in causes and conditions. Some counterfactual analyses have been done poorly, so we should be skeptical of their conclusions. However, some scholars, such as Richard Ned Lebow and Philip E. Tetlock and Aaron Belkin, have proposed more rigorous methods of counterfactual analysis. Counterfactual analyses may have more historical evidence available for understanding the considerations of historical actors than one might assume, such as discussion government officials had about multiple strategies before deciding on one. Better use of counterfactual analysis methods might create opportunities to develop conclusions about causation that we can place greater confidence in, although there are a variety of limitations to these methods too.
Despite the difficulties of reaching firm conclusions about causation, historians often argue or imply that by studying history, we can improve our intuition and general understanding through an indirect widening of experience. It would be possible to cite many examples where historical actors have looked to the past and successfully imitated political tools or avoided dangers. As one example, many leaders have self-consciously adopted unifying devices, rhetoric, and propaganda tools from historical precedent; Napoleon and Hitler both developed powerful cults of personality and even explicitly copied Roman symbols like the imperial eagle.
The implications of history can be multifaceted and contradictory, however, even in situations where we are relatively confident in our judgements about causation. Historian John Tosh notes, for example, that looking at the histories of World War I suggests that escalation of tension is a poor way to deal with an aggressive power, while studying World War II might suggest that appeasement was just as bad. These contradictions have led to individuals taking unhelpful lessons from the past; Richard J. Evans argues that attempts to learn from appeasement led to poor foreign policy decisions by post-war British politicians, such as the attack on Egypt in 1956. Evans argues that “history never repeats itself” and that “while many people, especially politicians, try to learn lessons from history, history itself shows that in retrospect, very few of these lessons have been the right ones.”
Individual historical cases can therefore provide inspiration for potential tactics and perhaps build our intuition, but we should not place much weight on strategic knowledge gained from a single case, because causal relationships may not replicate in different contexts and may seem to work in contradictory ways. Note, however, that weak evidence can still be useful and should not be disregarded as it is often all we have available.
Even if we are not very confident about individual hypothesized causal relationships, we may be able to place significant weight on the strategic knowledge gleaned from history if we see that certain correlations reliably replicate across different movements and across different contexts. For example, in the British Antislavery Movement, it seems that the passage of legislation that might have significantly curtailed the institution of slavery (the 1807 Abolition of the Transatlantic Slave Trade) increased the tractability of the full abolition of slavery. As a single example, this provides relatively weak evidence that a similar correlation between significant reform and full abolition would be replicated in the farmed animal movement. Similar correlations occurring across multiple successful movements and absent in movements that failed to achieve their goals would constitute much stronger evidence.
Observing similar correlations across multiple contexts and hypothesizing that such correlations will reoccur in other contexts where they have not been observed is variously referred to as establishing patterns, generalizations, or laws in history. If strong correlations exist between certain factors and successful outcomes of social movements or particular social movement organizations (SMOs), then this would be useful to know; even if we aren’t certain that there is a causal relationship between success and any of the independent variables identified, attempting to mimic the characteristics and engender the conditions that are most strongly correlated with success could increase the likelihood of successful outcomes. Historians and sociologists vary in their views about to what extent such patterns can be established, however.
Some historians are optimistic about the predictiveness of historical generalizations, though some historians are more critical and many are imprecise in their discussions of such topics:
While historians have moved to integrate the linguistic analysis and methodological tools that postmodernists advocate, many have rejected the more fundamental critiques posed by postmodernists, such as that summarized by Alun Munslow that “historians cannot represent the past as it truly was," and that their conclusions will “reflect their ontology as much as the data.” Notwithstanding the fundamental criticisms offered by postmodernism, there seems substantial overlap among many of the most widely cited historians writing on the general nature of the discipline, which I would summarize as: History cannot predict the future with certainty, but it can still provide valuable insights into processes, the present, and perhaps the future. Of course, this is only a very broad, agreeable claim that doesn’t get us very far in, for example, knowing how to weigh strategic evidence from history against other evidence like randomized controlled trials (RCTs), a topic discussed in our Effective Animal Advocacy Researcher Survey and Foundational Questions Summaries.
Sociologists have advanced wider theories about the causes of social change and the factors which lead to social movements being successful. Such an approach implies that correlations do replicate across movements. Studies like that by Sarah A. Soule and Susan Olzak (discussed above) test wider theories about the external factors that influence the success or failure of social movements, i.e. about those factors that the movement has not exerted much influence on themselves, such as the structure of government in the countries that the movement operates in. Such studies provide insights for the particular movement studied, but they are usually framed in terms of their contribution towards wider sociological understanding of the causes of change across social movements.
Sociologists have so far mostly only found conclusions that hold for the particular contexts they examine, although this is partially a result of the methodologies used (see the section on “External constraints on the impact of social movements” in “How tractable is it for thoughtful actors to change the course of history?”).
Social movement scholars have also studied the importance, in aggregate, of internal characteristics of SMOs to their success. William Gamson’s Strategy of Social Protest, for example, studied a “representative collection of  American voluntary groups that, between 1800 and 1945, have challenged some aspect of the status quo.” One of Gamson’s many conclusions was that groups “that aim to displace one or more antagonists” are much less successful in both winning acceptance for themselves and in gaining new advantages for their intended beneficiaries than those that do not. Gamson here defines “antagonists” as “the target of influence” of the challenging group. Only 12% of the “displacing groups” studied gained acceptance and 6% gained new advantages. In contrast, 62% of non-displacing groups gained acceptance and 68% gained new advantages. Such strong correlations seem likely to replicate across further untested movements and SMOs; even if there is high variation, Gamson’s data suggest that SMOs would be better off not aiming to displace their antagonists if they want to be successful. Most of the correlations in Gamson’s data were not this strong, however, and subsequent scholars have disagreed about the effects of some movement characteristics on the likelihood of success. Note that Gamson’s study was conducted using American data; to increase our confidence that the correlations identified hold across countries, as well across movements and SMOs, we would need additional data and analyses.
Of course the efforts to create general theory or “invariant models” about social movements or their impacts have been criticized on both theoretical and empirical grounds. Marco Giugni analyzed the importance of variables including public opinion and political alliances on the success of three different social movements in three countries; he notes that his findings were not consistent across movements or across countries. However, it is worth noting that Giugni’s model compares broad contexts rather than specific interventions such as protests or leafleting. Additionally, comparative analyses of social movements are vulnerable to bias in the process of selecting case studies, and case studies that at the outset appear to have the clearest implications for the questions of interest for researchers—which are therefore more likely to be selected for analysis—may be unrepresentative. Nevertheless, if certain correlations consistently occur across both different social movements and different contexts, then it seems likely that these correlations would be replicated in other movements and contexts too. Failing that, they may still help us to understand the causes of social change in specific movements or contexts, as noted above.
Some correlations between movement tactics and successes might be more replicable to the farmed animal movement than others because of relevant contextual differences. For example, contextual differences in the openness of a political system to democracy and pluralism should reduce the weight we place on a case study’s strategic implications for the importance of winning support from key politicians for causing legislative change. However, those same contextual differences might not have as much bearing on our confidence in the case study’s strategic implications for the extent to which gradual reform will encourage either complacency or momentum for more radical change, given that this question seems less related to democracy and pluralism. Similarly, whether a social movement was driven primarily by the intended recipients of new advantages, as with American civil rights from the 1950s onwards, or primarily by allies, as with the British antislavery movement, probably affects the weight that we should place on strategic knowledge from those case studies more than a variable like the average age of advocates.
 Of course, each of these three questions is large in scope, and one could devote entire books — or even careers — to answering the first two. This post is intended to provide useful insights from a variety of disciplines and examples. In some instances (most notably the historical case studies), I have therefore sometimes relied on summary works, literature reviews and meta-analyses, only occasionally consulting the original individual articles and books. In other cases, where the work is only partially relevant (such as the philosophical contributions to discussion on causation), my coverage will only be based on a few examples. To avoid overcomplicating the main text, I have put most of the detail from these works in the footnotes.
 At Effective Altruism Global, London, 10/28/18, Rose Hadshar, a Project Manager at the Future of Humanity Institute, gave a talk entitled: "From the neolithic revolution to the far future: how to do EA history."
 See for example:
 A talk I attended by Stephen Wise, on “Animal Rights and Legal Protections: A Perspective from North America,” 20th July 2018, at the Temple Chambers in London, suggested that Wise spent a significant amount of time conducting this research.
This article notes that “Wise’s approach tries to provide an answer to a paradox: How can a legal thing sue to challenge its thinghood? Wise eventually found inspiration by looking back at the history of slavery, specifically the famous Somerset v. Stewart case. James Somerset, a black slave purchased in Virginia, accompanied his owner, Charles Stewart, on a journey to England, where slavery was less entrenched than in America. Somerset tried to escape but was captured and returned to his owner; as property, he could not sue for his release. In 1772 the English abolitionist Granville Sharp, serving as a legal proxy, filed a writ of habeas corpus in his stead, which the justice, Lord Mansfield, upheld against the commercial interests of slaveholders. A precedent was set, as a man who was formerly property became a free person.”
 See for example:
See here for a summary of the usefulness of different types of effective animal advocacy evidence.
 Relevant correlations could also between success or failure and the presence of more indirect factors that the movement has not exerted much influence on, such as the structure of government in the countries that the movement operates in.
 William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975) studied 53 different American “challenging groups”
 E. J. Tapp, “Some Aspects of Causation in History” The Journal of Philosophy 49, no. 3 (January 1952), 67-8
 Ibid, 67
 James Brien “The Role of Causation in History," History in the Making 2 (2013), 75, summarizes Hume’s views: “During the eighteenth century philosopher David Hume questioned the prevailing views of causal relationships, which would have a significant impact on the philosophy of causation in history. He argued that a cause and effect relationship could not be proven; merely the relationship between two objects or events observed. Even if event A was always followed by event B, it could only be said that it is likely to occur. Hume believed that human mind formed the causal link between the two events. Hume was sceptical of the ability to define a law which identified a causal link between two events, but even Hume was reluctant to fully reject the notion of causation.”
 E. J. Tapp, “Some Aspects of Causation in History” The Journal of Philosophy 49, no. 3 (January 1952), 68, notes that “so deeply and firmly is the notion of causation embedded in our thinking that in spite of Hume's destructive analysis it has remained, and is likely to do so, as an essential assumption, basic to the study of the past."
James Brien “The Role of Causation in History," History in the Making 2 (2013), 75-6, summarises some of the subsequent explorations.
 Aviezer Tucker, Our Knowledge of the Past: A Philosophy of Historiography (Cambridge: Cambridge University Press, 2004), 92-140, outlines “theories and methods” to “achieve the status of probable knowledge." On page 141 Tucker argues that “the best explanation of dissent on historiographic issues within the uncoerced heterogeneous group of historians who work within the Rankean paradigm is the absence of knowledge of history and the alternative dominance of historiographic opinion. Historiographic opinions are historiographic hypotheses that are not better confirmed than inconsistent competing hypotheses.”
 Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997), 10, notes that “very few historians in practice have possessed the necessary expertise to discuss the theory of history at a level that a trained philosopher would consider acceptable, On the other hand, the level of abstraction at which most studies of historical epistemology operate is so theoretical, so far removed from actual problems experienced by working historians, that the subject in general is of little practical relevance to what historians actually do," and suggests reading J. H. Hexter, Doing History (London, 1971), 67-71, for further discussion of this.
 Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997), 135-8. For example, Evans summarises the work of Theodore Zeldin, which is “remarkable for its refusal to engage in large-scale theories or explanations” but still “delivers numerous quite cogent and persuasive explanations of French political history." Evans also disagrees with Geoffrey Roberts, who included an emphasis on “the role of accident, miscalculation and unintended consequences in shaping historical outcomes” by arguing that “historians by and large do not concentrate on the accidental, do not assume that individuals have unfettered freedom of choice, and do not confine themselves to reconstructing the past actor’s point of view.”
 John Lewis Gaddis, The Landscape of History: How Historians Map the Past (Oxford and New York: Oxford University Press, 2002), 102-3, for example, writes: “this still leaves the question, though, of how historians know when they’ve established, once and for all, the causes of any past event. The answer is, of course, that they don’t. Because not all sources survive, because not everything gets recorded in the sources in the first place, because the memories of participants can be unreliable, and because even if they were reliable no participant would have witnessed all of an event from all possible angles, we can never expect to get the full story of what actually happened… None of this means, though, that we lack a basis for determining causes in history: it only means that our basis is a provisional one.”
Ludmilla Jordanova, History in Practice (London: Arnold, 2000), 94, writing more generally on historical knowledge, echoes Gaddis’ sentiment: “It is healthy to acknowledge the provisional nature of our knowledge. However, it does not then follow that the quality of historical knowledge is unimportant….my stress on reliable rather than on objective, truthful knowledge is thus intended to be realistic and honest. Historians should not promise what they cannot deliver.” On pages 108-9, Jordanova emphasises the complexity of historical causation: “We no longer imagine that causes were either political or economic or social, but rather see all of these at work, and find them all dependent upon, or at the very least bound up with, cultural shifts, with changes in ways of understanding and feeling about the world. In other words, conventional hierarchies of explanations no longer seem as plausible as they once did. Nor do we like the idea of invoking single causes or one type of cause, preferring to chart a wide range of factors. Hence questions like ‘was the French Revolution caused more by economic than by political factors?’ have come to seem rather artificial.”
 A.J.P. Taylor, War by Time-Table: How the First World War Began (London and New York: Macdonald & Co. Ltd, 1969), 45. Taylor was, however, probably being deliberately contrarian here: Taylor still offers explanations of what led to World War I breaking out, which are considered in more depth in the post on “How fragile is history?”
 Donatella della Porta and Mario Diani, Social Movements: An Introduction (Oxford and Malden, MA: Blackwell Publishers, 1999), 230-3, notes that “the simultaneous presence of a large number of objectives also makes it difficult to estimate what has been achieved on various fronts…. The attribution of credit for obtaining substantive successes also faces a series of obstacles (Tarrow 1994; Rucht 1992; Giugni 1997; Diani 1997). The principal problem is one well known to social scientists: the existence of such close relationships between a set of variables that it becomes impossible to identify cause and effect… Second, the presence of a plurality of actors makes it more difficult to attribute success or failure to one particular strategy (Diani 1997)... Thirdly, the difficulties created by a plurality of actors are added to by the difficulty of reconstructing the causal dynamics underlying particular public decisions… Particularly where comparison is being made between different movements or countries, the problems outlined above hinder an evaluation of the relative effectiveness of particular movement strategies.”
Felix Kolb, Protest and Opportunities: The Political Outcomes of Social Movements (Frankfurt: Campus Verlag, 2007), 1-2, asks “Why is it so difficult to explain the political impact of social movements? I am convinced that causal complexity is an important part of the answer (cf. Earl 2000; Giugni 1999). In the context of social movement outcomes, causal complexity refers to the fact that not only social movements but also interest groups, public opinion, mass media, political parties and many other factors are likely to exert causal influence on the process of social and political change. In order to sort out the various influences, we first need clear theories of the causal processes by which social movements produce their effects (Tilly, 1999). It is a key premise of this book that such causal theories are still underdeveloped in the social movement literature - despite the existence of much empirical work on the outcomes of social movements.”
 Marco Giugni, Social Protest and Policy Change: Ecology, Antinuclear, and Peace Movements in Comparative Perspective, (Lanham, MD: Rowan and Littlefield, 2004), 113-6, in conclusion to part I of the book, shows the difficult to attributing causation in the outcomes which may, or may not, have stemmed from the actions of the ecology and antinuclear movements in the USA, Switzerland and Italy. On page 113, on the ecology movement in the USA, for example, Giugni notes that “Spending by the U.S. government for environmental protection seems to have followed a development opposed to that of protest, in particular with a strong decline during the early 1980s. This might lead us to conclude that the ecology movement was instrumental in bringing about changes in environmental policy, at least as far as spending is concerned; yet, the reverse causal path is just as likely. Changes in environmental spending might have provoked a decline of the mobilization. Furthermore, both developments over time largely reflect shifts in the presidency. Protest went up when a Republican president was in power and down with a Democrat in the White House. Similarly, spending increased under Democratic presidencies and diminished under Republican ones.”
 S. A. Hawkins and R. Hastie, "Hindsight: Biased Judgments of Past Events after the Outcomes Are Known," Psychological Bulletin 107, no. 3 (1990), 311
 See S. A. Hawkins and R. Hastie, "Hindsight: Biased Judgments of Past Events after the Outcomes Are Known," Psychological Bulletin 107, no. 3 (1990), 311-27, for a review of the studies, especially 317-9 For example, 313 notes that “Pennington doubted that the typical sizable laboratory hindsight effect would be obtained when judgments were made of important current news events. In one set of judgments, similar to the standard Fischhoff paradigm, the experimenter provided the subjects with descriptions of possible outcomes (defined by settlement terms and duration of the strike) and elicited probability estimates both during the strike (foresight) and at two points in time after it was over (hindsight). These estimates revealed the typical hindsight pattern of greater probabilities associated with the actual outcome in hindsight as compared with foresight.”
 Jay J. J. Christensen-Szalanski and Cynthia Fobian Willham, “The hindsight bias: A meta-analysis," Organizational Behavior and Human Decision Processes, 48, no. 1 (1991), 153-4, “with a 95% confidence interval = .14-.20 (correction for unreliability yields r = .25; 95% confidence interval = .21-.29).” On 150, they note that in this study “r is the degree to which the presence of hindsight information is related to a change in a person’s probability assessments.”
Jacob Cohen, Statistical Power Analysis for the Behavioral Sciences (New Jersey: Lawrence Erlbaum Associates, 1988; first published 1969), 79-81, defined r = 0.1 as a “small effect size,” d = 0.3 as a “medium effect size,” and d = 0.5 as a “large effect size”
 Jay J. J. Christensen-Szalanski and Cynthia Fobian Willham, “The hindsight bias: A meta-analysis," Organizational Behavior and Human Decision Processes, 48, no. 1 (1991), 150, notes that “in the present study, d corresponds to the standardized difference between the hindsight (experimental) and foresight (control) probability assessments, while r is the degree to which the presence of hindsight information is related to a change in a person’s probability assessments.” On 154, they report that when corrected for unreliability, the overall results were r = 0.25, d = 0.52. When given “familiar tasks," the effect of hindsight bias on participants’ predictions that an event would not occur (if the outcome was indeed that it did not occur) was much smaller, at r = .09 and d = .18. The effect of hindsight bias on predictions for events that did occur, for “familiar tasks," was however, closer to the overall effect, at r = .24 and d = .48.
 S. A. Hawkins and R. Hastie, "Hindsight: Biased Judgments of Past Events after the Outcomes Are Known," Psychological Bulletin 107, no. 3 (1990), 314, summarise their literature review by saying that “our initial review of the major laboratory studies of hindsight phenomena shows that there is a substantial, reliable effect of outcome information on subsequent judgments, even when the subject is given unambiguous instructions to ignore the outcome information.”
Rüdiger F. Pohl, Wolfgang Hell, "No reduction in Hindsight Bias after Complete Information and repeated Testing," in Organizational Behaviour and Human Decision Processes 67, No. 1 (July 1996), 49: “In an attempt to reduce the typically found bias, the participants of one group in Experiment 1 were informed in advance about the experimental design and the bias phenomenon, while the participants of another group were not. In Experiment 2 all participants received at the end individual feedback about their recall performance and were then subjected to the same experimental procedure again, twice in succession. The amount of hindsight bias, however, remained unaffected by either of these two manipulations.”
 Daniel Kahneman, Thinking, Fast and Slow (London; Penguin Books, 2012; first published 2011), 87. Studies and analyses of relevance to this conclusion are summarized at various points throughout the book. For example, on 86-7, Kahneman summarizes a study by Amos Tversky and others, where participants were exposed to legal scenarios and “participants who saw one-sided evidence were more confident of their judgements than those who saw both sides.”
 Geoff Boeing, “Visual Analysis of Nonlinear Dynamical Systems: Chaos, Fractals, Self-Similarity and the Limits of Prediction,” Systems 4, no. 4 (2016), 1, “Chaos theory is a branch of mathematics that deals with nonlinear dynamical systems. A system is simply a set of interacting components that form a larger whole. Nonlinear means that due to feedback or multiplicative effects between the components, the whole becomes something greater than the mere sum of its individual parts. Lastly, dynamical means the system changes over time based on its current state. Nearly every nontrivial real-world system is a nonlinear dynamical system. Chaotic systems are a type of nonlinear dynamical system that may contain very few interacting parts and may follow simple rules, but all have a very sensitive dependence on their initial conditions.”
 Geoff Boeing, “Visual Analysis of Nonlinear Dynamical Systems: Chaos, Fractals, Self-Similarity and the Limits of Prediction,” Systems 4, no. 4 (2016), 1. This is because “despite their deterministic simplicity, over time, these systems can produce wildly unpredictable, divergent and fractal (i.e., infinitely detailed and self-similar without ever actually repeating) behavior due to that sensitivity. Forecasting such systems’ futures thus requires an impossible precision of measurement and computation.”
 Geoff Boeing, “Visual Analysis of Nonlinear Dynamical Systems: Chaos, Fractals, Self-Similarity and the Limits of Prediction,” Systems 4, no. 4 (2016), 1, “Nearly all nontrivial real-world systems are nonlinear dynamical systems.”
Note, however, that some elements are more predictable than others. John Lewis Gaddis, The Landscape of History: How Historians Map the Past (Oxford and New York: Oxford University Press, 2002), 76, reflects on chaos theory to argue that “The micro-level phenomena within our system are, for the most part, linear in character, in that there’s a predictable relationship between input and output… But the macro-level behaviour of our system as a whole… is non-linear… Poincaré’s great insight was to show that linear and non-linear relationships could coexist: that the same system can be simple and complex at the same time.”
 Aviezer Tucker, Our Knowledge of the Past: A Philosophy of Historiography (Cambridge: Cambridge University Press, 2004), 222, notes that “If “‘Chaos’... is essentially a mathematical concept, and ‘chaos theory’ is a mathematical theory” (Smith, 1998, p. vii), then it makes little sense to discuss “chaos,” “criticality,” or “linearity” in history.” 223-5 then summarises the views of historians who have variously suggested that chaos theory does or doesn’t translate well to history. Tucker notes that those who have argued that it does have not demonstrated their case convincingly: “Reisch [1991, p. 9] provided no empirical evidence for which historical processes are obviously chaotic. McCloskey (1991) acknowledged as well that it is impossible to know whether many systems are linear or not. Still, McCloskey opined that history may well be chaotic and suggested a few anecdotical qualitative examples from the historiography of the Civil War in the United States. But though it seems likely that the exact course of some of the events of the U.S. Civil War may have been sensitive to initial conditions, McCloskey did not prove that somethings similar would not have been brought about by different initial conditions and certainly neither McCloskey nor Reisch introduced the kind of mathematical modeling that scientists associate with chaos.”
Aviezer Tucker, “Historiographical Counterfactuals and Historical Contingency," History and Theory 32, no. 2 (May 1999), 269, cites Yemima Ben-Menahem, “Historical Contingency,” Ratio 10 (1997), 102, as not committing herself to the application of chaos theory to history: “the sensitivity of historical events to initial conditions may or may not be as radical as that of chaotic phenomena, but it can certainly be very significant.”
 Aviezer Tucker, Our Knowledge of the Past: A Philosophy of Historiography (Cambridge: Cambridge University Press, 2004), 39
 Annika Mombauer, “The Fischer Controversy 50 years on,” Journal of Contemporary History 48, issue 2 (April, 2013), 239, wrote that the controversy around the causes of World War I “is a perfect case study for the possibilities and limits of empirical historical analysis” and “is an excellent example of how historical research is always inevitably tied to the contemporary political context within which it is conducted."
 Gail Bossenga, “Origins of the French Revolution," History Compass 5, no. 4 (July 2007), 1296, notes that “rather than one overarching interpretation of the origins of the Revolution, there has tended to be multiple perspectives, which are not necessarily incompatible with one another but tend to remain partial: cultural origins, religious origins, financial origins, political origins, geopolitical origins, and even a renewed Marxist plea for classic ‘social’ origins. Historians also face the question of how to reconcile long-term origins, which tend to be structural in nature, with short-term origins, which emphasize the contingent nature of specific events and decisions by individuals.”
Peter R. Campbell, “Redefining the French Revolution. New directions, 1989–2009," H-France Salon 1, Issue 1, No. 2 (2009), 7, agrees that “the bicentennial vision of the Revolution has foundered, to be replaced by complexity but not coherence.” On 14, Campbell adds that “once upon a time, the origins were thought to characterize the whole Revolution, whether as class struggle or as leading to the moment at which the ideas of the Enlightenment were to put into practice. Today, the classic interpretation has been undermined, there is ambivalence about the Enlightenment, and no new paradigm for explaining revolutionary change has emerged.”
 Gwynne Lewis, The French Revolution: Rethinking the Debate (London: Routledge, 2003; first published 1999),106-113, argues for “a postrevisionist consensus," where “marxisant historians were moving to accommodate the valid criticisms of revisionists concerned about too determinist, too structuralist an approach to the study of history." On page 107, Lewis summarizes one example: “it is somewhat ironic that Albert Soboul, a constant target of revisionist historians, condemned out of hand as a ‘marxist-leninist’ ideologue, should have painted a far less rosy picture of industrial growth in France before the Revolution than did [the revisionist] Chaussinand-Nogaret, repeatedly reminding us that ‘Capitalism [before 1789] was still essentially commercial.’” 109-10 notes support for the importance of “an intra-class conflict over basic political relations," rather than an emphasis on class conflict as the main driver of change.
Peter R. Campbell, “Redefining the French Revolution. New directions, 1989–2009," H-France Salon 1, Issue 1, No. 2 (2009), 9, on the other hand, notes that “Furet’s exclusively political-philosophical interpretation [has been undermined] with an effective, if not quite orchestrated, counter-attack. Whether we can yet talk of the “return of the social” in a new guise is difficult to say.” On page 21, Campbell adds that “the political-philosophical approach has been undermined over the last two decades by the breadth of approaches spurred on by the bicentenary and by a transformation in the very definition of History as a discipline. Several of our essayists appear to be every bit as skeptical of a totalizing explanatory notion of political discourse as they once were of class conflict.”
 Henry Heller, The Bourgeois Revolution in France, 1789-1815 (New York and Oxford: Berghahn Books, 2006), viii-ix, agrees that the “revisionist school” has “dominated” the historiography since the 1960s, but notes that “against the dominant revisionist conception of the French Revolution, the following work reasserts the view that the Revolution - the capital event of the modern age - was a capitalist and bourgeois revolution.”
 Soule, Sarah A., and Susan Olzak. "When do movements matter? The politics of contingency and the equal rights amendment," American Sociological Review 69, issue 4 (August, 2004), 489.
 Soule, Sarah A., and Susan Olzak. "When do movements matter? The politics of contingency and the equal rights amendment," American Sociological Review 69, issue 4 (August, 2004), 490: when incorporating the presence of AAUW chapters, the percent of legislature that was democratic and an interaction coefficient as separate predictors (model 4), “taking all three relevant coefficients for the main and interaction terms into account at different levels of each of the two indicators implies that the rate increases over fourfold when both the number of AAUW chapters and the percentage of Democrats in the state legislature are high, compared to when they are both at average levels in a year. Clearly, allies matter.” In the opposite direction, the interaction of Republican Party strength and anti-ERA organizational strength was also important: “when these two forces are one standard deviation higher than the observed average, the chances of ratification are diminished by 88 percent (the multiplier on the rate decreases from 1.0 to .12)”
 See, for example, Frances Fox Piven and Richard A. Cloward, Poor People’s Movements: Why They Succeed, How They Fail (New York: Pantheon Books, 1977), which considers 1) The Unemployed Workers’ Movement (41-95), 2) The Industrial Workers’ Movement (96-180), 3) The Civil Rights Movement (181-263), and 4) the Welfare Rights Movement (264-362)
 See, for example, William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975)
 Felix Kolb, Protest and Opportunities: The Political Outcomes of Social Movements (Frankfurt: Campus Verlag, 2007), 9. As a specific example, Kolb notes that “Meyer and Minkoff (2004) exclusively defined the policy outcomes of the civil rights movement in terms of annual outlays for the Commission on Civil Rights between 1955 and 1985. They found only two of their twelve indicators of political opportunities to be significantly correlated with the outcomes of the civil rights movement.” Kolb adds further examples: Costain 1994; Giugni 2004b; Landman 2000; Soule et al. 1999.
 Richard Ned Lebow, “What's so different about a counterfactual?," World Politics 52, issue 4 (July 2000), 564, notes that “counterfactuals are a key component of evaluation. Was the development of nuclear weapons a blessing or a curse for humankind? What about affirmative action, free trade, or the growing economic and political integration of Europe? Serious and thoughtful scholars can be found on all sides of these controversies. Their arguments share one thing in common: the use of counterfactual benchmarks—most often, implicitly—to assess the merits of real-world policies, outcomes, or trends.”
 See the section on “Consideration of causation” in “How is SI research different from existing social movement literature and relevant historical works?” for detail on this.
 Niall Ferguson (ed.), Virtual History: Alternatives and Counterfactuals (New York: Basic Books, 1999), 9-17, summarises some previous efforts at counterfactual history, describing them as “whimsical," “trivial," “reductive," “supposed to be funny” (as opposed to “plausible”), “facetious," or having “only the sketchiest of supporting evidence.”
Although I haven’t read all the works that Ferguson cites, Ferguson’s judgement seems plausible, since my impression of other historical counterfactual analyses is that they do not take themselves seriously as contributions to academic understanding of the topics that they consider. Robert Cowley (ed.) What If? The World’s Foremost Military Historians Imagine What Might Have Been (London: Macmillan, 2000; first published 1999), for example, uses literary titles for its contributions, such as “Unlikely Victory” and “If Only It Had Not been Such a Wet Summer.” Interspersed among the main chapters, are some smaller one or two page “what ifs,” which are too short for anyone to consider them to be methodologically rigorous, and give the impression of being included as if to add some drama, such as Williamson Murray, “What a Taxi Driver Wrought,” 306-7.
Ferguson’s own work, has, in turn, been seriously criticised. Richard Ned Lebow, “What's so different about a counterfactual?," World Politics 52, issue 4 (July 2000), 570, summarises the work of Niall Ferguson, The Pity of War: Explaining World War I, (New York: Basic Books, 1999), by arguing that “Ferguson relies heavily on counterfactual arguments on the grounds that they help recapture the uncertainty of the actors, to whom the future was merely a set of possibilities, and that they allow us to assess the extent to which they made the right choices. Ferguson's questions are good ones; his answers are not. He appears to be following the A. J. P. Taylor approach to the writing of history: make outrageous arguments that stand history on its head, infuriate the critics, gain publicity, and with any luck, sell a lot of books.”
Aviezer Tucker, “Historiographical Counterfactuals and Historical Contingency," History and Theory 32, no. 2 (May 1999) adds that “the eight historical counterfactual studies that compose Ferguson’s anthology have varying levels of significance. The most interesting studies demonstrate where, when and how history was necessary or contingent. Other studies are no better or worse than conventional historiography, while still others are irrelevant for the kind of problems in which historians and philosophers of history are interested.”
 Richard Ned Lebow, “What's so different about a counterfactual?," World Politics 52, issue 4 (July 2000), 581-4, proposed “eight criteria for plausible-world counterfactuals”: “1. Clarity… 2. Legal consistency or cotenability… 3. Enabling counterfactuals should not undercut the antecedent… 4. Historical consistency… 5. Theoretical consistency… 6. Avoid the conjunction fallacy… 7. Recognize the interconnectedness of causes and outcomes… 8. Consider second-order counterfactuals.” Note that several of these were drawn from Tetlock and Belkin (see footnote 47).
 Philip E. Tetlock and Aaron Belkin, “Counterfactual Thought Experiments in World Politics: Logical, Methodological, and Psychological Perspectives," in Philip E. Tetlock and Aaron Belkin, (eds.), Counterfactual Thought Experiments in World Politics: Logical, Methodological, and Psychological Perspectives (Princeton: Princeton University Press, 1996), 18, lay out “Six Criteria for Judging Counterfactual Arguments”: ““1. Clarity... 2. Logistical consistency or cotenability… 3. Historical consistency (minimal-rewrite rule)... 4. Theoretical consistency… 5. Statistical consistency… 6. Projectability.”
 Giovanni Capoccia and R. Daniel Kelemen, “The Study of Critical Junctures: Theory, Narrative, and Counterfactuals in Historical Institutionalism," World Politics 59, No. 3 (April 2007), 356, notes that “the difference between a factual and a counterfactual argument as regards the availability of historical sources should not be exaggerated: there may be as much historical evidence about decisions that were taken as about those that were considered, discussed, and ultimately discarded.”
Richard Ned Lebow, “What's so different about a counterfactual?," World Politics 52, issue 4 (July 2000), 553, notes that “even when evidence is meager or absent, the difference between counterfactual and "factual" history may still be marginal. Documents are rarely smoking guns that allow researchers to establish motives or causes beyond a reasonable doubt.”
 As general explanations of the merits of counterfactual analysis for understanding causation in particular case studies, Giovanni Capoccia and R. Daniel Kelemen, “The Study of Critical Junctures: Theory, Narrative, and Counterfactuals in Historical Institutionalism," World Politics 59, No. 3 (April 2007), 368, notes that “We identify contingency as the key element of critical junctures. During critical junctures change is substantially less constrained than it is during the phases of path dependence that precede and follow them. In critical junctures contingency is enhanced, as the structural constraints imposed on actors during the path-dependent phase are substantially relaxed. Only by taking counterfactual analysis seriously can contingency be studied.”
Richard Ned Lebow, “What's so different about a counterfactual?," World Politics 52, issue 4 (July 2000), 561, notes that “even when such evidence is available, it may still not be possible to determine the relative weight of the several hypothesized causes and which of them, if any, might have produced the outcome in the absence of others or in combination with other causes not at work in the case. To sustain causal inference it is generally necessary to engage in comparative analysis. Within the single-case format comparative analysis can take two forms: intracase comparison and counterfactual analysis.”
 See for example: Gary King and Langche Zeng, “When Can History Be Our Guide? The Pitfalls of Counterfactual Inference," in International Studies Quarterly 51, no. 1 (March 2007), 183-210
See also Philip E. Tetlock and Aaron Belkin, “Counterfactual Thought Experiments in World Politics: Logical, Methodological, and Psychological Perspectives," in Philip E. Tetlock and Aaron Belkin, (eds.), Counterfactual Thought Experiments in World Politics: Logical, Methodological, and Psychological Perspectives (Princeton: Princeton University Press, 1996), 32-7, lay out psychological biases which pose challenges for counterfactual analysis. As an example, 35-6 lists 4 “motivational biases”: “(1) Needs for predictability and controllability… (2) Needs to avoid blame and to claim credit… (3) Needs for consolation and inspiration… (4) Need for cognitive consistency.”
 Jorma Kalela, Making History: The Historian and Uses of the Past (Basingstoke and New York: Palgrave Macmillan, 2012), ix, argues, for example, that “Why history?’... People need knowledge of the past, and everybody uses this cognition in his or her own way.”
John Lewis Gaddis, The Landscape of History: How Historians Map the Past (Oxford and New York: Oxford University Press, 2002), 10-11, argues that “There is no “correct” interpretation of the past, but that the act of interpreting is itself a vicarious enlargement of experience from which you can benefit. It would ill serve any prince to be told that the past offers simple lessons… Studying the past is no sure guide to predicting the future. What it does do, though, is to prepare you for the future by expanding experience, so that you can increase your skills, your stamina - and, if all goes well, your wisdom.”
Margaret Macmillan, The Uses and Abuses of History (London: Profile Books, 2009), 153-4, argues vaguely that “History can help us to be wise; it can also suggest to us what the likely outcome of our actions might be. There are no clear blueprints to be discovered in history that can help us shape the future as we wish. Each historical event is a unique congeries of factors, people or chronology. Yet by examining the past, we can get some useful lessons about how to proceed and some warning about what is or is not likely to happen.”
 John Tosh, Seán Land, The Pursuit of History: Aims, Methods and New Directions in the Study of Modern History (Harlow: Pearson Education Limited, 2006; first published 1984) 39
 Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997), 59. Evans claims that Prime Minister Neville Chamberlain became "obsessed with the desire to avoid the 'appeasement' of dictators which he had himself criticized in the later 1930s, launched an ill-advised and unsuccessful military strike against the Egyptian government under Colonel Nasser when it nationalized the Anglo-French-owned Suez Canal.” Evans does not provide evidence to support this claim, however, and this was presumably not the sole cause of the attack.
 E. H. Carr, What is History (Harmondsworth: Penguin Books Ltd, 1964; first published 1961), 63
 John Tosh, Seán Land, The Pursuit of History: Aims, Methods and New Directions in the Study of Modern History (Harlow: Pearson Education Limited, 2006; first published 1984) 39, 41, 49.
Arthur Marwick, The Nature of History (Basingstoke: Macmillan, 1989; first printed 1970), 17, 152, argued similarly that history enables us to “understand the problems of the present” and to grapple “intelligently with the future," even if it doesn’t allow us to solve problems and does not offer the “power of prediction” that “scientific laws” can.
 Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997), 59, 61.
 Postmodernist criticisms relate to a wider debate about whether “facts” are possible in history. As a brief summary, “positivist” historians have seen history as a science, drawing clear-cut conclusions from historical evidence, “idealists” stress the importance of intuition and interpretation. Postmodernists see both historical evidence and the language of historical writing as subjective and full of assumptions.
Alun Munslow, The New History (Abingdon and New York: Routledge, 2014; first published 2008), 8.3, for example, notes “working from the belief that historians ascribe meanings to the past rather than discover its inherent or given meaning, my basic pre-supposition is that the positioned historians through his/her historicised (ideologised, gendered, class-based or whatever) situation, constitutes the meaning of the past as he/she narrates it. In other words, the historical consciousness is as much tropical or narrative-linguistic as it is a synthetic empirical-analytical undertaking. This is the perspective, founded on the work of key philosophical, literary and historical thinkers like Stephen C. Pepper, Paul Rocoeur, Heyden White, and Frank Ankersmit that informs my view of history.”
Ernst Breisach, On the Future of History: The Postmodernist Challenge and its Aftermath (Chicago and London: The University of Chicago Press, 2003), 24, summarising the influence of the Linguistic Turn on postmodernist thought, notes that “in the linguistically constructed world, truth would have neither permanence nor stable foundations and hence no privileged authority. Language changed from being the neutral medium between consciousness and the outside reality to being itself the only accessible reality”
 Ernst Breisach, On the Future of History: The Postmodernist Challenge and its Aftermath (Chicago and London: The University of Chicago Press, 2003) gives many examples of compromise approaches. 204 notes one such example: “John Toews suggested guidelines for an accommodation when he defined experience as not directly available to the historian but being mediated by meaning, which was itself “constituted by and through language." However, that was no one-way process since human experience, in turn, shaped or at the least delimited the construction of meanings. Neither language nor experience could be considered reality in isolation from each other.”
 Ernst Breisach, On the Future of History: The Postmodernist Challenge and its Aftermath (Chicago and London: The University of Chicago Press, 2003), 88, summarises that “A considerable number of works have shed light on the importance language and rhetoric has had in the construction of historical accounts. This recognition has been the most important contribution of narrativism to historical inquiry. For some time to come, narrative and language will not be regarded as silent and passive partners in the creation of historical works. But as a theory of history meant to replace traditional history with its full recognition of the dimension of time and of elements that in human terms are stable and can be grasped cognitively, poststructuralist narrativism has not yet commended itself ot historians”; 116 adds that “The rejection of demands for radical change in historical inquiry came from historians, however much they differed in their views, foci and arguments. They agreed on the destructive effect of the denial of a possible, referential, authoritative truth and of the theoretical tenets derived from it. Historians also protested that, contrary to postmodernist contentions, they have never claimed to have attained the final truth about the past but only that sucha truth must remain the ultimate although probably never attainable ideal. Good historians always have been aware of the fallibility of their findings - due to error, insufficient data, risks taken in interpretation, and the experiential background of the historians. They have affirmed, however, that the cumulative results of proper perception, cognition, and imagination, expressed in the symbolic forms of language, could lead to a sufficiently proximate knowledge of the past as it had once been lived.”
Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997) is one example of many books which (I find, convincingly) refutes the more radical proposals of postmodernism. 252-3 concludes that “it is right and proper that postmodernist critics should force historians to rethink the categories and assumptions with which they work, and to justify the manner in which they practise their discipline. But… for my own part, I remain optimistic that objective historical knowledge is both desirable and attainable… I will look humbly at the past and say despite them all: it really happened, and we really can, if we are very scrupulous and careful and self-critical, find out how it happened and reach some tenable though always less than final conclusions about what it all meant.”
 Alun Munslow, The New History (Abingdon and New York: Routledge, 2014; first published 2008), 8.8, 8.32
 E. H. Carr, What is History (Harmondsworth: Penguin Books Ltd, 1964; first published 1961) has 4381 citations on Google Scholar.
Richard J. Evans, In Defence of History (London: Granta, 2000; first published 1997) has 1543.
John Tosh, Seán Land, The Pursuit of History: Aims, Methods and New Directions in the Study of Modern History (Harlow: Pearson Education Limited, 2006; first published 1984) has 1459.
John Lewis Gaddis, The Landscape of History: How Historians Map the Past (Oxford and New York: Oxford University Press, 2002) has 974.
Arthur Marwick, The Nature of History (Basingstoke: Macmillan, 1989; first printed 1970) has 683.
 William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975), v
 William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975), 14-15 and 42-3. In testing both “acceptance” and “new advantages” across groups, the p value was < 0.001. The definitions of the “target of influence” of the group is “that set of individuals, groups, or social institutions that must alter their decisions or policies in order for a challenging group to correct a situation to which it objects. Such a target is the object of actual or planned influence attempts by the group… In the case of the Steel Workers Organizing Committee, for example, the various steel companies were the antagonist. In the case of the Social Revolutionary Clubs (Anarcho-Communists), the national regime was the antagonist.” On pages 323, Gamson notes that 55% of the groups “had predominantly governmental targets.”
See the section “Empirical data on the success rates of social movements” in “How tractable is changing the course of history?” for details on Gamson’s methodology. All the categorizations are subjective
 William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975), 51, notes, for example, that “There is only a negligible difference between groups of different size in gaining new advantages. Larger groups are not much more likely to achieve the changes they desire than smaller ones.” 54% of groups over 10,000 gained new advantages, compared to 44% of groups under 10,000.
 Some of the criticisms are included in some editions of the book itself. See “Appendix a., Further Analysis and Disputation,” in William Gamson, Strategy of Social Protest (Belmont: Wadsworth Publishing Company, 1990; first published 1975), 181-276. I have not checked through all criticisms of the book to see if the findings relating to the effects of attempting to displace antagonists has been challenged by reanalysis or further evidence.
Marco Giugni, Social Protest and Policy Change: Ecology, Antinuclear, and Peace Movements in Comparative Perspective (Lanham, MD: Rowan and Littlefield, 2004), 3, summarizes that “Other authors share [Gamson’s] view about the effectiveness of disruptive movement tactics [and Giugni cites 5 studies]... however, a number of studies have shown that the use of violence does not pay off and sometimes is even detrimental to the movement [and Giugni cites 3 studies].”
 Jeff Goodwin and James M. Jasper, “Caught in a Winding, Snarling Vine: The Structural Bias of Political Process Theory," 27, argues that “The search for universally valid propositions and models, at least for anything so complex as social movements, is bound to fail… Even when they do not intend to, process theorists appear to propose invariant models because of the structural models they deploy; greater attention to strategic choice, cultural meanings, and emotions would highlight the complex, open-ended quality of social conflict… The explanation of empirical variation will likely require considerable conceptual and theoretical variation as well. Some kinds of movements require political opportunities, while others do not; some recruit through pre-existing social networks, while others do not; some require powerful grievances or collective identities, while others do not.”
Charles Kurzman, “The Poststructuralist Consensus in Social Movement Theory," in Jeff Goodwin and James M. Jasper (eds.), Rethinking Social Movements (Lanham, Md.: Rowman and Littlefield, 2004), 111-120, argues that invariant “structuralist” models have lost popularity, suggests a newer “constructionist” consensus, and that “Constructionist norms about the importance of process imply that the outcome of movements cannot be foreseen in advance." On page 117, Kurzman argues that “constructionist norms about the importance of process imply that the outcome of movements cannot be foreseen in advance.”
 Marco Giugni, Social Protest and Policy Change: Ecology, Antinuclear, and Peace Movements in Comparative Perspective, (Lanham, MD: Rowan and Littlefield, 2004), 215, in a summary of the joint effect of unconventional mobilization, political alliances, and public opinion on public policy in three countries, Giugni notes that “like the previous analyses, these final ones do not yield results that hold consistently across countries and across movements.” On page 222, Giugni adds that “when comparing across countries, one finds it difficult to draw reliable conclusions. However, we do observe a difference consisting in a strong joint effect of the movements in the United States; an intermediate one in Switzerland; and a weak one in Italy, especially if we look at the third variant of the joint-effect model. In this case, the two federal countries with more open political opportunity structures seem to be more conducive to a policy impact of social movements than the more centralized country. However, this might simply be a result of the lack of data on public opinion available for the Italian case concerning the ecology and antinuclear movements.”
 Marco Giugni, Social Protest and Policy Change: Ecology, Antinuclear, and Peace Movements in Comparative Perspective, (Lanham, MD: Rowan and Littlefield, 2004), 222, for example, in spite of the methodological difficulties above, ties findings together, noting that “In sum, the analyses presented in the second part of the book support the general hypothesis that the policy impact of social movements is facilitated by the presence of allies within the institutional arenas and a favorable public opinion. This facilitating action, however, occurs only if these two factors are accompanied by a sustained mobilization of the movements."