macro-to-macro causation

Consider the following model of causation. cross-level-causation

Most readers of this blog will be familiar with the idea that social facts create the conditions of individual action (e.g., Durkheim’s theory of suicide). This proposition is fundamental to sociology and (if you’re looking for coherence Fabio) it may be the most distinctive theoretical claim the discipline makes.  Arrow 2 makes a lot of sense too. The conditions of individual action shape how individuals behave, which then leads naturally to arrow 3 – that individual action shapes social or collective outcomes. But what about arrow 4 – the macro to macro link?

Specifically, there are no conceivable causal mechanisms in the social world that operate solely on the macro-level. There are no macro-level entities on the social domain that somehow possess capacities or dispositions to act (Cartwright, 1989) that make them capable of directly producing macro-level outcomes, and there are no processes of interaction between macro-entities that take place on this level. In short, there is no macro-level causal mechanism that can be theoretically represented in terms of arrow 4.



Written by brayden king

March 3, 2009 at 3:03 pm

34 Responses

Subscribe to comments with RSS.

  1. BTW, the full paper can be downloaded here.



    March 3, 2009 at 3:04 pm

  2. Maybe the micro mediators of macro-to-macro events are trivial, but they’re never absent.


    David Chen

    March 3, 2009 at 3:17 pm

  3. The problem I always tend to see is that individuals are macro phenomena of some other micro phenomena. I’m not suggesting a runaway reductionism – I would counter it. I would say that “causation” is a useful cognitive tool to understand mechanism. If it is useful to so that macro phenomena can cause macro phenomena, then do it. I think not allowing macro -> macro would ignore extremely important emergent phenomena (“macro”) that use individuals as nothing more than a substrate to move through.

    Observe traffic shockwave : a traffic wave is an emergent phenomena that moves through individual action. I would say that none of the drivers intend on creating a wave (as opposed to “the wave” at stadiums), but the wave maintains conditions that recreate and propagate itself. I would say the wave causing the wave is a macro -> macro causality.



    March 3, 2009 at 4:25 pm

  4. Hmm, I will abstain from discussion. (This type of posting should not be allowed—you are forcing me into the internets/blogosphere.) Jack Vromen cordially says we’re wrong—here’s an old version of his piece, forthcoming in Erkenntnis —–



    March 3, 2009 at 5:11 pm

  5. Arrow 4 is pointing the wrong direction.

    Social outcomes constitute institutions. A vivid example from today is the way that buying or selling of stocks en masse leads to declines in the publicly listed price of companies, which in turn leads to new social facts for the companies. An historical example is the way that new social norms formed after the invention of the printing press, which in turn led to reorganization of religious institutions.

    All of this gets more obvious by the day as the loop accelerates thanks to the Internet…


    Michael F. Martin

    March 3, 2009 at 6:01 pm

  6. Too late Teppo. You’re already in the internets. You’re fair game. :)



    March 3, 2009 at 6:37 pm

  7. OK, non-sociologist here who doesn’t know what he’s talking about and hasn’t bothered to read any papers, but Jesse’s comment above makes sense to me. Whether or not arrow 4 makes sense is an empirical matter, not a theoretical one. Are there useful macro-macro relationships? I don’t think it’s inconceivable, but it may be pretty unlikely.

    If your job is to fully specify causal mechanisms, then even arrow 2 can’t be right, because individual behavior depends on cognitive mechanisms, which in turn may depend on biological mechanisms, and so on. You have to keep going down. On the other hand, if your job is to specify useful mechanisms, then it’s conceivable that arrow 4 may come in handy.

    I get the feeling that people are sometimes irritated by how macro-macro explanations (arrow 4) are unconstrained and hand-wavy, and so it can be reassuring to replace arrow 4 with arrows 1, 2, and 3, because that seems more like a real explanation.



    March 3, 2009 at 6:48 pm

  8. Aren’t we always to some extent dealing with a choice on the level of abstraction when explaining a phenomena and the underlying mechanism of causation? So for example, could not the link between individuals conditions of action and Individual action similarly be said only to operate through some even more mico-mechanism… cognitively biological or chemical to a responses, etc?

    However, as some of the other responders seem to reply, this does seem to lead to more and more reductionism. Perhaps the better question is when it is appropriate to abstract to the macro to macro link, even when there are still underlying micro mechanisms at work. Perhaps in the interest of parsimony, we would use macro to macro explanations only when the underlying mico mechanisms are so stable and (ideally) non-contingent on other factors that we get little explanatory power through their explication.



    Peter B

    March 3, 2009 at 7:35 pm

  9. I like Peter’s suggestion. The question should be: “when it is appropriate to abstract to the macro to macro link?”

    I wrote a neat little blog reply to this including an expanded image done with Word and paint… oooooh!



    March 3, 2009 at 11:20 pm

  10. I can’t remember if this has been posted here, but Steve Strogatz presented at TED on the math and physics behind emergent phenomena that could provide a macro-macro link.


    Michael F. Martin

    March 3, 2009 at 11:48 pm

  11. Some mannheimian thoughts:

    Jesse (post #3) says “a traffic wave is an emergent phenomena that moves through individual action”. I agree and think that this sentence gives the reason for link #2 (and thus for 1 and 3). Note further that this is the explanation given on the site Jesse linked to:

    “[shockwave jams are] typically triggered by a single driver slowing down. After that first vehicle brakes, the driver behind must also slow, and a shockwave jam of bunching cars appears, travelling backwards through the traffic.”

    Sounds like counterfactual causality using a link to the micro level: if this first single driver had not slown down, the driver behind had not and no shockwave would have started to appear. It’s hard to explain this macro phenomenon by ignoring the drivers (the micro level), i.e. links #1-3 in the graph. (I admit it’s hard to explain the drivers’ behavior without looking in their brains… see further below.)

    Jesse’s sentence “I would say that none of the drivers intend on creating a wave” sounds as if he defines causality/causation (at least on the micro level) as an *intended* act—I don’t and I think we shouldn’t.

    Michael (post #5) says “Arrow 4 is pointing the wrong direction”.
    It is not, if you allow a second boat/bathtub attached to the first to start with the point “social outcome”. The social outcomes of the first bathtub are then “social facts” for the second. This game is supposed to be played infinity.

    Finally, let me add that I like the following two comments very much:
    (1) “If your job is to fully specify causal mechanisms, then even arrow 2 can’t be right, because individual behavior depends on cognitive mechanisms, which in turn may depend on biological mechanisms, and so on. You have to keep going down.” (Barbar)

    (2) The whole post by Peter B except the idea that macro-macro explanations will be useful a number of times worth mentioning.

    In conclusion: Coleman (1986) has no macro-macro link, Coleman (1990) a solid line for the macro-macro link. I like Esser (1993)’s version better, where the macro-macro link is represented by a dashed line: (I know it’s German, sorry.) Ah yes: and I like Jesse’s “expanded bathtub” (see his blog; I drew something very similar some time ago). My argument why for most sociologist explanations it suffices to look at the macro-micro-macro bathtub, is that I don’t expect too much additional explanatory power through an approach going further—but in the end that’s an empirical question…



    Sebastian E. Wenz

    March 4, 2009 at 2:24 am

  12. […] …you’ll find here with some mannheimian thoughts (scroll down to the […]


  13. Comeptiitive markets (social facts) generate prices (social outcomes). In a competitive market, individual actors are inconsequential.

    The invisible hand is, perhaps ironically in light of our rational choice friends, a macro- to -macro mechanism. Of course this effect is not “solely” macro. Nothing is. Nothing is “solely” micro either.

    Markets and institutions are themselves emergent processes. The figure misses an arrow from individual action to emergent social facts.


    Willie Ocasio

    March 4, 2009 at 3:21 am

  14. My very quick two cents.

    Arrow four still, for me, is about high-level correlations rather than meaningful explanation and theoretical understanding. Sure, we might say (a la Milton Friedman) that its about prediction at this macro level and who cares about explanation and understanding—-but, I think the latter is absolutely critical. And, to meaningfully explain and understand, we need to delve into the motivations, interests, beliefs and actions of the actors themselves rather than simply hover at the variable level.

    Yes, arrow three definitely is all about aggregation and indeed precisely about (the often all-too-vague) “emergence,” etc—-about how collective facts, structures, order etc comes about (think Hayek, Schelling, Coleman). And, yes, arrow three can lead to social facts that then constrain the choice set in subsequent social activity. To this point, we might meaningfully, then, think about this figure being preceded by another figure exactly like it (so a continual dynamic of interaction between the micro and the macro and vice versa—-PLEASE, no one say structuration!).

    I don’t know about stacking the figure as is suggested above by Jesse. I think there is a “natural stopping point” (see Coleman, chapter 1), where we are interested in the level just below the social, and interested in explaining the social in terms of how it gets instantiated and how it emerges over time through interactions. Going to the genetic level creates an all-too deterministic conception in my mind, while actors are actually pursuing choices with sets of interests and beliefs that may have nothing to do with genes.



    March 4, 2009 at 6:30 am

  15. Agree with tf. I think Coleman (and others) has convincingly argued that we should *not* look for direct macro-macro links, but always explain via the microlevel.
    @Willie Ocasio: isn’t the relation between a competitive market and equilibirum prices a classic example of a macro-macro relation being explained by behavior at the individual level, in this case, producers’ and consumers’ behavior? After all, this stuff is taught in mircoeconomics courses for a reason ;-).



    March 4, 2009 at 8:31 am

  16. Just to add someone who has not been mentioned but who seems relevant: DeLanda, in his “New Philosophy of Society”, actually defines an assemblage (well, its half of his definition anyway) as an actor that is able to interact with other actors at the same level. He has an explicitly flat ontology — meaning, simply put, that this assertion applies to ‘macro’ actors (institutions, cities, states) just as much as ‘micro’ actors (individuals, slime molds, hydrogen atoms). He obviously committed to saying, for example, that institutions collide with each other, that institutions have macro-order effects.

    For me its simply a question of representation: if an actor is able to maintain its boundaries, to sustain its tensions, if its able to arrange other actors in a stiff enough formation, negotiate a meaningful truce or alliance with them, then its fair enough to ‘attribute’ causation to said actor not matter what ‘size’ it is (or, better, how long it is, or how far it is able to project itself). And even if this move is just a helpful short-cut that creates an unhelpful black box that does not mean that we can’t and shouldn’t open it up.

    I agree with what many have already said, though: its always going to be an empirical question.


    limitations and shortcomings

    March 4, 2009 at 11:43 am

  17. To Rense:

    “Microeconomics” does not necessarily imply individual level at all. Microeconomics is the study of markets. So here there is a confusion about what the word micro in microeconomics means. As I tell my students what counts for micro in economics would be called macro in organization theroy.

    Market pricing and equilibrium cannot be reduced to individual behavior of consumers and producers. Now individual consumer behavior creates market demand and individual producer behavior creates supply, hence my call for a link between individual behavior and social facts. But it is market-level demand and market-level supply that generates prices.

    One interesting point is that you do not need rational choice theory to generate the law of demand and supply.

    Note also that all of classical economics is based on macro-to-macro mechanisms with no attempts to show the microfoundations. For example Ricardian rent theory, which is the basis of the modern resource-based view of the firm is a macro-to-macro theory. As is international trade theory, the theory of comparative advantage.

    Neoclassical economics added microfoundations to classical economics. But for a lot of the theorems in the theory the mechanisms that do the work are still macro-to-macro.


    Willie Ocasio

    March 4, 2009 at 1:07 pm

  18. Willie: And, from my perspective, that is why some central assumptions (for example, specifying rationality as omniscient) of the neoclassical model are hard to buy—-the model does not dip down and specify the micro factors (local information, rationality in context, interests, etc) appropriately. For me, various sociological theories, ironically, suffer from the same problem (though, for example, rather than assume omniscience, they essentially assume rather strong structural determinism—or at least never really get to the arrow 3 issues at all).



    March 4, 2009 at 3:50 pm

  19. @Willie Ocasio

    “The figure misses an arrow from individual action to emergent social facts.”

    If arrow 4 is reversed, there is a connection, albeit indirect


    Friedman and his neoclassical followers (especially Becker) have done very well with prediction divorced from explanation. I have no problem with that strategy in general.

    But to get better at predicting things when neoclassical theory fails, it helps to have an explanation of the underlying mechanisms, if for no other reason than to eliminate predictive theories that have no physical meaning. Economic or sociological theories are subject to instrumental constraints that don’t apply as strongly to other disciplines.

    To clarify, the sync mechanism is macro-macro in the sense that two large groups in different locations can sync up through a weak coupling. Market price signals are certainly one example of this phenomenon, but so are many smaller scale institutions setup within firms.


    Michael F. Martin

    March 4, 2009 at 4:59 pm

  20. I’ve been enjoying the discussion. I feel like I should throw in my two cents too.

    Like Teppo, I think it’s really important that we study arrows 1 and 2, especially in connection with organizations. We know surprisingly little about how organizations affect individual behavior. A lot of organizational psychology is actually done outside of organizations altogether. The unique context of the organization is ignored, in part because it would be difficult to simulate an organizational context in an experimental lab. We can simulate decision-making processes and to some extent team work, but trying to apply an organizational experimental condition is a little bit of a stretch. At best you can only isolate one or two aspects of an organization, e.g., centralization of power. And so I’m strongly supportive of the idea that we should try to find innovative ways to study how organizational contexts influence individual behavior.

    That said, I’m also a big fan of arrow 4. Much of my research focuses on this level of phenomena. I don’t think you can reduce organizations to their constituent parts without destroying what is organizational about the entity. The same thing could be said for sports teams or social movements. All three phenomena have groupiness. Individuals within the groups are interdependent, such that the performance of any one individual affects the performance of someone else. The whole really is greater than the sum of its parts.

    But organizations have additional characteristics that make them even more distinct. Organizations, for one, can control their boundaries, shaping who enters and leaves. The sovereignty of organizations has a powerful psychological effect on members that influence the way they think and behave when acting as a member-agent of the organization. Thus, between-organization variance in individual behavior is likely to be greater than within-organization variance for at least certain kinds of behaviors. In addition, organizational goals, beliefs, or some emergent properties of organizations are not reducible to the beliefs of individuals within the organization. Executive teams make decisions that don’t always align with the individual interests of the members of that team. Strategies are path dependent and don’t reflect the preferences of one member of the organization.

    As soon as you accept the idea that organizations have some autonomy from their constituent members and that their influence is not reducible to the aggregated influence of their members, then you can begin to see how organizations influence their environments and shape collective outcomes. A corporation influences the regulatory environment through lobbying. A social movement organization organizes a protest, which changes the way the public thinks about about a certain issue. You can then move to an even higher level – multiple protests coorganized by many SMOs have a greater influence than one SMO protesting alone. One protest is probably not sufficient to affect a policy outcome, but multiple protests related to the same issue may cause Congress to pay attention and hold a hearing on that issue. The point is that sometimes collective outcomes are precisely the result of some macro phenomenon that can’t be reduced to individuals acting alone.



    March 5, 2009 at 5:27 am

  21. I taught my doctoral seminar on Org Theory yesterday and the topic was routines, knowledge, and capabilities.

    This is a topic where micro to macro, macro to micro and macro to macro issues are critical and the subject of much controversy and debate.

    I brought up in class the diagram and the debate we were having in this blog. Several students responded in ways consistent with a methodological individualism view, and somewhat to my surprise even with a rational choice perspective.

    What became clear as the discussion ensued was the key point about macro-level social facts, whether markets, institutions, organizations, or routines is that they are more than just the aggregation of their members.

    Now not reducible does not mean not influenced. For example there is a lot of interest in institutional entrepreneurship in exploring how agents exert their influence on institutions. And in the top management team theory, individuals do exert their power and influence.

    So Brayden hits the nail on the head, in my mind. Macro phenomena are not reducible to the aggregation of atomistic individuals. And neither are macro outcomes.


    Willie Ocasio

    March 5, 2009 at 12:53 pm

  22. Right, the question indeed is what we mean by “not reducible” (implies “emergence” which is tricky to capture). Is the macro not reducible because there is a past history and it simply is too complex to tease out all the micro relationships and decisions that led to the macro outcome? That’s not completely satisfactory for me since it seems precisely the job of the social scientist to explain the emergence of the macro rather than take the macro for granted. But, naturally, depending on the problem and phenomenon one is wrestling with, then one might start with some macro premises (given institutional context, market dynamics, environment) and then work from there, but it seems that one has to still then “dip down” to explain all this in terms of the actors involved. Just sticking with the logic that routines lead to organizational performance (arrow four)—feels a bit unsatisfactory.

    The micro, as Brayden’s discussion suggests (organization influences environment), might also be the organization itself as an actor, the organization in a macro environment or population. I think the intuition readily transfers upward. The Ingram and Clay paper (ARS, 2000), “choice-within-constraints,” that summarizes institutional intuition from both econ and soc, reflects some of this.



    March 5, 2009 at 5:31 pm

  23. I don’t think it’s necessary to tease out emergence at the micro or meso levels to explain macro outcomes. Once something has emerged, it has its own crystallized structure or form that is irreducible. This is not to say that we shouldn’t study emergence, but I don’t think that it’s necessary to link emergence to arrow 4 all of the time.



    March 5, 2009 at 5:43 pm

  24. Agree, very roughly.



    March 5, 2009 at 5:47 pm

  25. The reason why methodological individualism fails to describe some macro effects is because methodological individualism is predicated on linearity — i.e., that two people acting a certain way will have EXACTLY the same impact within or without an organization as twice the impact of one acting the same way. That assumption fails in many circumstances. That doesn’t mean that methodological individualism is useless, just that it is an approximation, and should be handled as such.

    Here’s another paper that the folks here might not have seen:—J—Seder–P—–Kesebir–S-


    Michael F. Martin

    March 5, 2009 at 5:49 pm

  26. Again, agreed (thanks for the paper link). Part of the problem, though, is that one assumes that one can simply, by stating the linearity/aggregation assumption, then move directly to the macro level without explicating the process and dynamics between the micro and macro—this is where we might often just fuse the micro and macro together, but that does not feel very satisfactory, at least from an analytical perspective.



    March 5, 2009 at 7:07 pm

  27. Michael, methodological individualism does not presuppose linear effects everywhere.


    Michael Bishop

    March 6, 2009 at 7:34 pm

  28. Say more please. I am not a specialist in this field. Thank you.


    Michael F. Martin

    March 6, 2009 at 7:40 pm

  29. Coupla things:

    1. I think arrow 4 should be double-headed.
    The design of any organization is goal-oriented, therefore social facts (organizational structure) lead to (forwards) and are organized around (backwards) social outcomes. *note that forwards and backwards are in reference to arrow directions and not some ontological opinion, but could be understood in terms of the metaphor engineering and reverse engineering.*

    2. I think there should be an arrow from individual action to social facts.
    When i behave in a certain matter, other people view that behavior and integrate that into the fact that someone else in their social-circle did this. This could be opening an umbrella on leaving a building or moving out of town.

    3. I think there should be an arrow from social outcomes to conditions of individual action.
    The formal and informal intended social outcome of a group is a context for individual action, as when a salesman calls a prospect because he want to sell some of his product.

    – I know i’m late to the game, but i just read Schelling’s Micro-motives and Macro-behavior (got to get it back to the library today).
    – I also want to recommend philosopher John Searle’s work on Intentionality where he elucidates the creation of social facts (money) and social structures through shared internal representations.

    ps: i think it unfair that blayden’s posts are grey :)


    Byron Woodson

    March 11, 2009 at 8:31 pm

  30. […] macro-to-macro causation […]


  31. […] Archivado en: antropología, estudios, investigación — Esteban S @ 2:31 pm Hace unos días se preguntaban en orgtheory por la existencia de una causación macro-macro. El razonamiento es el siguiente: los hechos […]


  32. I agree with Peter Bearman’s argument that the question needs to be rephrased.

    I will extend his argument to make the point that we can account for macro to macro links when we study institutions and organizations in which the individual has little opportunity to shape the social outcome, such as a citizen in a dictatorship or an infant in a family. In these cases we may argue that social facts explain social outcomes above and beyond individual action. However, I would also argue that individual actions in these conditions can turn significant to the model in the long-run (history). Here is an example:

    If we take the concentration camp as an example for an institution and the social outcome as the mass death of Jews, then we might argue that the social fact determined the social outcome above and beyond individual actions (resistance).

    However, we know that even in this very extreme case, individual action, the act of defiance, was very limited, but yet it happened. Although individual actions may not have changed the social outcome of Jews at that time, they contributed to the overall arching social outcome of Jews in the long-run . In other words, an act of defiance by individuals in the Warsaw Ghetto may have not changed the overall outcome of the Holocaust, but it may have contributed to the social outcome of Jews many years later(such as uniting/uplifting the collective consciousness of Jews …).


    Shira Mor

    March 19, 2009 at 6:31 am

  33. Shira, great comment. Thanks for that insight.



    March 19, 2009 at 1:35 pm

Comments are closed.

%d bloggers like this: