In the last reflection article on my dissertation, I answered the most recent question that has been posed to me. Today, I want to revisit one of the oldest questions that someone has asked me.
I remember it vividly: It was December 2020, the pandemic still raging globally, and we were sitting in the traditional go-to bar of IAS, Ölstugan Tullen in Norrköping.1 It was me, a few coworkers, and the question-asker: the first PhD graduate of the Institute for Analytical Sociology, Alex Giménez de la Prada. Back then, he was almost finished with his thesis and set to successfully defend a few months later. We were sitting with a beer, and I was asking what the overarching goal of a PhD at the IAS would be. He answered: “What is analytical sociology? This is the big question that you should be answering with your PhD.”
I already suspected that this harmless sounding question would probably turn out to be an entire barrel of worms. And why wouldn’t it? It’s usually the innocent questions that turn out to be the monster under your bed. And since he told me that my entire PhD (that is, four entire years) should be dedicated to answer this, I took it with the appropriate amount of respect.
Over the next years, I have had many discussions with colleagues and friends about analytical sociology. I am especially thankful to Rodrigo Martínez Peña, with whom I discussed this question for years until he also successfully defended his PhD two years ago. It felt almost like a treasure hunt. But a treasure hunt without an end goal.
So, what exactly is analytical sociology? Do I have an answer, after five years?
The Core Tenets of Analytical Sociology
Having observed the field and the research emanating from there over the past five years, I believe there are five “core tenets” of analytical sociology, if you so will. Five things that mark something as a piece in analytical sociology. These stem largely from the four central theoretical tomes on analytical sociology, released in 1998,2 2005,3 2009,4 and 2021.5 But they are also based on recent developments starting only in the late 2010s such as the emergence of computational social science, and my own observations.
The central hallmark of analytical sociology is certainly its focus on methodological individualism, “the striving for explaining social phenomena (mostly) without reference to structures and institutions.”6 Essentially, methodological individualism posits that, in order to understand why society works the way it does, we must look at the individuals comprising it, and not some macro-phenomenon we may imbue with meaning.
At the core of methodological individualism sits a specific tool that offers a visual icon like no other to identify a methodological individualist: the Coleman boat, or (in the German context), Coleman’s bathtub.7 The Coleman boat describes a way to think about social mechanisms, in which one can explain macro-phenomena through their micro-foundations.
It is clear that the Coleman boat is merely a tool to think about social phenomena, not the one and only description. Because it is possible to define a Coleman boat for any time period from minutes to centuries, and for a variety of use-cases it is a great analytical tool. But this also comes at the expense of ease of understanding.
This is one of the biggest issues that plague any new student of analytical sociology, including me. Where should one start? And what should I include in my theory, and what should I exclude? These are all very hard analytical questions, and part of what make AS “analytical” in the first place.
The Coleman boat seamlessly leads into another core feature of what people globally associate with “analytical sociology”: agent-based models. Once you accept the premise that society only works via micro-foundations, and that it is the combination of individuals and their actions that lead to the emergence of macro-phenomena, you enter the realm of ABMs.
Some believe ABMs to be a core feature of analytical sociology because of the way they map onto the epistemological foundations of the Coleman boat.8 An ABM is a simulation in which the researcher defines a set of actors and a set of actions they can perform, and observes what macro-patterns emerge from there. The most famous and “OG” ABM is probably Thomas’ Schelling’s segregation model.9 Indeed, it is one of the first visual demonstrations of individual-level behavior leading to emergent macro-phenomena any student of analytical sociology is exposed to.
The appeal of ABMs for analytical sociology is that they allow for testing highly specific mechanisms for how society works. They require a Coleman boat and a definition for each of the four nodes in it, and then one can just run a simulation to see if the theoretical hypothesis holds.10
This leads to a fourth tenet of analytical sociology: its use of middle-range theories. In a short chapter in the 2009 book, Peter Hedström and Lars Udehn explain what they understand as “middle-range.” Essentially, they occupy a middle space between the “grand theories” of the mid-twentieth century and very particular “mini-theories.” The idea is that sociology should only aim to explain a single phenomenon, using a Coleman boat approach and possibly an ABM, and neither try to limit its theory to a particular instantiation of the phenomenon, nor try to make general claims about a wide range of different instantiations of this and similar phenomena.
The fifth, and final, core tenet is relatively new, and it might be debatable in what way it actually constitutes a core feature of analytical sociology. In 2009, a paper by David Lazer et al. made headlines.11 Published in Nature, it was one of the first social scientific papers that recognized and embraced the “age of big data.” The authors argue that, with the new surge of available data, and increasing capabilities of computers, social scientists could tap into a data source of a never-before known resolution. Instead of relying on large-scale surveys, (field) experiments, or qualitative efforts, social scientists could just take a look at the massive amounts of data generated by human interactions globally to understand how society works.
This revolution has led to an increase in the computational requirements of analytical sociology. Today, I know only few colleagues who perform analytical sociology without the need for high-performance computing clusters (HPC). Almost all of my colleagues, including me, know of at least one period where we had to leave our computers running for days waiting for an analysis to conclude. Even our master students at the institute (who, coincidentally, are formally studying computational social science) all know of this. And once you realize that students usually don’t have the money for powerful laptops, you have an understanding for the frustration they sometimes have to go through.
But What is Analytical Sociology, Really?
But none of this really explains what analytical sociology is. Methodological individualism and the Coleman Boat were around way before the first book on analytical sociology. The same holds for agent-based models and computationally heavy workloads. Middle-Range Theories even stem from the times of Robert K. Merton. None of this really tells us what makes analytical sociology … well, analytical.
I believe that an answer to this question requires an indirect approach. There is no singular definition of what analytical sociology is. Indeed, if you take a look at the literature, many scholars have attempted to define analytical sociology, and there is still no consensus. Some have called it a “superfluous revolution,” others obsess over ontological problems in the practical application of it, and even the godfather of the field, Peter Hedström, continuously tweaks the definition of analytical sociology.
Instead of providing a direct answer, I will follow the lead of my university’s dean, who has — in my opinion — asked in the correct way: Does analytical sociology solve the problem of lawmaking?
Essentially, therefore, this article cannot answer the question generally, but by means of my dissertation. By doing so, this entire article is necessarily incomplete and only one piece to the puzzle. It is based on my personal experiences, having spent five years at the IAS. Due to the simple fact that any observer might see it differently,12 there cannot be a final answer to the question of what analytical sociology is or is not.
Does Analytical Sociology Solve the Issue of Lawmaking?
I personally believe that having access to the methodological toolkit of analytical sociology has been immensely beneficial. I cannot recount how often I have sketched a Coleman Boat in the past five years when trying to understand a particular phenomenon related to lawmaking. Also, simply being forced to take an individual-level perspective has made it much simpler to reason about lawmaking processes. It has prevented me from falling into the trap of assuming the entire U.S. legislature as one large “black box,” as some scholars tend to do.13
But I didn’t make use of the entirety of the toolbox of analytical sociology. I did not run an ABM, I did not perform simulations, and as such I was unable to perform proper in-depth causal analyses over the data. This also had to do with the fact that text data does not lend itself easily to analytical sociology. Analytical sociology has been developed in an age where all social scientists had available to them were simple, straight-forward behavioral datasets. Text data is not simply behavioral data. And this really is an issue.
I think that my dissertation is a testament to how difficult it can be to transform textual data in such a way that standard approaches to sociology become possible. And I further believe that much of the current deluge of text analysis that emerges in the field of computational social science is neat, but it also severely lacks the analytical depth required for rigorous research. Many of the papers that currently float on the hype wave will likely prove to be unreproducible. Also, if you think about them more deeply, you will realize that many of these papers cannot provide us with any form of mechanism for what makes society work. They have a certain form of … theoretical shallowness to them.
Turning back to the core question, I also believe that analytical sociology sometimes does not help us in understanding the issue of lawmaking. I have found that, in order to understand the discursive dynamics of U.S. Congress, I had to take recourse to more traditional political science frameworks. And this includes something analytical sociology tries to avoid at all costs: referencing macrostructures as part of the explanation.
You see, analytical sociology sees individuals as the atomic movers of society, the fundamental force of the universe. For analytical sociology, at least in a strict sense, there cannot be any macrostructure that isn’t explainable as the emergent outcome of dozens or thousands of individual interactions. And if you think about what this presumed ontology of society from the perspective of analytical sociology means in practice, one cannot escape a very particular quote manifesting in one’s head: “There is no such thing as society. Only individuals” (and families).
And maybe this is wholly true. Maybe it was a mistake of Hedström to renounce the necessity for accessing the mental states of individuals in order to arrive at theoretically solid causal mechanisms.14 But I would argue that, in practice, it is simply impossible to fulfill the requirements of a strict, narrow, causal form of analytical sociology.
And the primary reason is that we simply cannot have all data available to us. This is one of the – in my opinion – primary insights I gained over the past five years. I always struggled to place my work in the framework of analytical sociology, and only this year did I realize why this is the case. Politics, and lawmaking more specifically, always misses information. We cannot ever have full information on these processes. This insight emerged from my inability to include parties in my work. As I write in my dissertation introduction:
Indeed, the epistemologically hazy foundations of party power in U.S. Congress is one of the primary reasons why this thesis must subscribe to the “weak” form of methodological individualism: Parties absolutely do wield influence over the lawmaking process, but most of this influence happens via backroom-politics, that is: off the record. Conversely, this thesis has only access to sources that were deliberately put on the record. Thus, in order to account for party power, the essays of this thesis have to utilize proxy measurements wherever necessary. This includes controlling for party affiliation, committee assignments, or seniority. However, all of these controls need to measure the party as a single macro-institution, without being able to track individual behavior. None of these controls will accurately measure the influence parties have over the representatives. This becomes evident in the final essay of this thesis, which resorts to measuring party pressure via the standard error of an OLS regression model (Nokken & Poole, 2004). It serves as a reminder that parties can remain hidden in plain sight.
So, ultima ratio, I have to conclude that analytical sociology does not, in fact, solve the issue of lawmaking. It provides a plethora of tools, and it makes it a thousand times easier to think about lawmaking. It has helped me tremendously to make sense of the data that I have, and figure out how lawmaking processes work. But ultimately, analytical sociology fell short of helping me cross the final threshold. Because, in the end, proper analytical sociology requires access to a perfect database.
Final Thoughts
I would call myself an analytical sociologist. I believe that analytical sociology can help us understand society to a degree that is impossible with many other approaches. It offers unprecedented resolution. And for this reason, I believe it is the single-most suitable approach to work on lawmaking processes.
But analytical sociology is no silver bullet. Strong methodological individualism poses strict requirements to both data and analytical approach. It requires both removing a lot of the social grit that makes society interesting to study in the first place, and at the same time almost perfect data availability. And lawmaking can fulfill neither. Lawmaking is influenced by many factors, all of which can become quite important, including mere cohabitation. At the same time, a lot of lawmaking happens behind closed doors, where one of the decisive factors is that no data is being produced.
While I believe that analytical sociology is an incredibly powerful approach to society, lawmaking, and text analysis more specifically, demonstrates its limits. Not everything can be answered by analytical sociology. And methodological and theoretical pluralism in the social sciences are really the only answer if one aims to deliver a holistic understanding society. Sometimes, analytical sociology fails. But I see this as an opportunity.
So, what is analytical sociology? It is many things. And for different scholars, it means different things. The only thing I know after five years is that it offers amazing resolution that I would never want to miss.
1 Reminder: Sweden implemented the “YOLO” pandemic protocol.
2 Hedström, P., & Swedberg, R. (1998). Social Mechanisms: An Analytical Approach to Social Theory. Cambridge University Press.
3 Hedström, P. (2005). Dissecting the Social: On the Principles of Analytical Sociology. Cambridge University Press.
4 Hedström, P., & Bearman, P. S. (Eds.). (2009). The Oxford Handbook of Analytical Sociology. Oxford University Press.
5 Manzo, G. (2021). Research Handbook on Analytical Sociology. Edward Elgar Publishing. https://doi.org/10.4337/9781789906851
6 Erz, H. (2025). On the Record: Understanding a Century of Congressional Lawmaking through Speech and Vote Behavior [PhD Thesis, Linköping University]. https://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-217773, page 7.
7 To quote my supervisor Jacob Habinek: “It’s either the Coleman boat or the Coleman bathtub, but we don’t know because James never drew any water.”
8 See Hedström 2005, Coda.
9 Schelling, T. C. (1978). Micromotives and macrobehavior (1st ed). Norton.
10 Disclaimer, before anyone gets mad: I have never run an actual ABM to test a mechanism-based hypothesis yet, and this is merely based on close contact with colleagues who did. It’s probably more involved than I make it appear here.
11 Lazer, D., Pentland, A., Adamic, L., Aral, S., Barabasi, A. L., Brewer, D., Christakis, N., Contractor, N., Fowler, J., Gutmann, M., Jebara, T., King, G., Macy, M., Roy, D., & van Alstyne, M. (2009). Life in the network: The coming age of computational social science. Science, 323(5915), 721–723. https://doi.org/10.1126/science.1167742
12 Luhmann, N. (1992). Beobachtungen der Moderne. Westdeutscher Verlag.
13 Looking at you, international relations.
14 Opp, K.-D. (2024). The recent turn in analytical sociology: The dismissal of general theories, mental states, and analytic philosophy – and the old issue of mechanism explanations. Social Science Information, 63(2), 131–154. https://doi.org/10.1177/05390184241247724