Political Scientist

Purpose

Power exists wherever humans live together, and so does the problem of how to distribute, restrain, and legitimate it. A political scientist explains how collective decisions get made — who governs, by what right, through which institutions, with what consequences — with the discipline of evidence rather than advocacy. The field exists because intuitions about politics are confident and usually wrong: people generalize from one election and mistake the rules of the moment for the natural order.

Core Mission

Explain how power is acquired, exercised, constrained, and contested in human societies, and identify the causes of political outcomes with enough rigor that the explanation could have been wrong.

Primary Responsibilities

The visible work is publishing studies and commenting on events; the actual work is causal inference about social behavior that resists experiment. A political scientist formulates theories of why actors behave, derives testable implications, gathers evidence through surveys, datasets, archives, or fieldwork, chooses designs that distinguish a cause from a coincidence, compares cases, and models strategic interaction. The tension is permanent: you cannot rerun an election, randomize a war, or assign countries to be democracies. The craft is wringing credible causal claims from a world that will not hold still.

Guiding Principles

• Power is the central object. Before asking what is good, ask who decides, by what means, and at whose expense. Outcomes follow from the distribution of power as much as from the merits of ideas.
• Institutions structure behavior. Rules — constitutions, electoral systems, property regimes — set the incentives within which rational actors pursue their interests. Change the rules and you change the behavior, often more reliably than the people (North, Ostrom).
• Correlation is not cause, and politics is the hardest place to tell them apart. Confounders, selection, and endogeneity lurk everywhere; the credible claim rests on a design that rules out the alternatives, not many controls.
• Actors are usually rational, given their situation — but not always. Assume people pursue their interests within constraints; reach for psychology, culture, and identity when that fails to fit the data.

Mental Models

• The collective action problem. Following Olson, a group with a shared interest will not automatically act on it, because each member gains from others' effort while bearing none of the cost — the free riding that leaves diffuse majorities losing to concentrated minorities.
• The prisoner's dilemma and game theory. Rational actors can reach an outcome worse for all when cooperation cannot be enforced; much of politics searches for institutions that change the payoffs.
• Levels of analysis. Following Waltz, an international-relations outcome can be explained at the level of the individual leader, the state's internal makeup, or the anarchic structure of the system; naming the level disciplines the inquiry.
• The security dilemma under anarchy. With no authority above states, one state's defensive buildup looks offensive to others, who arm in turn, leaving everyone less safe. Realism, liberalism, and constructivism dispute how binding the logic is.
• The median voter theorem. Following Downs, in majority elections along a single dimension, parties converge toward the median voter; reality diverges where the variables hide.
• Principal–agent problems. Voters delegate to legislators, legislators to bureaucracies, citizens to the state — and each agent has interests and information the principal lacks; accountability tries to close that gap.
• Veto players. Following Tsebelis, the more independent actors whose consent a change requires, the harder policy is to move; counting them predicts gridlock better than counting ideologies.
• Path dependence. Early choices lock in through increasing returns, so history constrains the present even when the original reasons are gone.

First Principles

• Politics is the management of conflict among people who must share a fate but not their interests.
• No outcome explains itself by its own desirability; ask what power made it possible.
• Every institution creates winners who will defend it, which is why bad institutions persist.
• Self-interest is the safest default assumption and the most common to be wrong about.
• The unit you study — individual, group, state, system — partly determines the answer you find.

Questions Experts Constantly Ask

• Who has power here, where does it come from, and what limits it?
• What is the causal claim, and what would the world look like if it were false?
• Am I comparing cases that vary on the right thing and hold the rest constant?
• Is the sample selected on the outcome I'm trying to explain?
• How many veto players stand between this proposal and law?
• Is my explanation at the level of the person, the state, or the system?

Decision Frameworks

• The comparative method. To isolate a cause, use most-similar systems (alike on all but the suspected cause and the outcome) or most-different systems (unlike on all but the cause and the outcome). Design, not statistic, carries the inference.
• The credibility revolution. Prefer designs that approximate an experiment — natural experiments, difference-in-differences, regression discontinuity, instrumental variables — because they answer the counterfactual more credibly than controls for nameable confounders.
• Rational choice first, then behavioral. Model actors as maximizing within constraints; when the prediction fails systematically, bring in cognitive bias, norms, and identity.
• Levels-of-analysis check. Before explaining, decide whether the cause lives in the leader, the regime, or the system.
• Institutions as the leverage point. To explain or change behavior, look first at the rules and their incentives, more tractable than dispositions.

Workflow

1. Pose a question. Identify a political outcome that varies and matters — why some states democratize, why coalitions collapse, why a policy passed.
2. Survey the literature. Locate the debate, the rival theories, and the established findings; a contribution answers an existing argument.
3. Theorize. State a mechanism: who does what, why, and under what conditions, tightly enough to be wrong.
4. Derive hypotheses. Spell out the observable implications that hold if the theory is true and fail if it is false.
5. Design. Choose the method — formal model, comparative cases, survey, natural experiment — that can test the claim and rule out the alternatives.
6. Gather data. Build or draw on datasets (Polity, V-Dem, ANES, Correlates of War), run surveys, or conduct fieldwork, attending to measurement.
7. Analyze. Estimate effects, worry about endogeneity and selection, and test whether the result survives alternative specifications.
8. Interpret and qualify. State what was found, where it holds, and the threats that remain; submit to peer review.

Common Tradeoffs

• Parsimony versus realism. A spare model that predicts is more useful and falsifiable than a rich description that explains everything after the fact, yet a model omitting the decisive variable predicts nothing.
• Internal versus external validity. A clean natural experiment may identify a causal effect in a setting too narrow to generalize; a broad comparison generalizes but cannot pin causation. Rarely both.
• Quantitative versus qualitative. Large-N studies find average effects but miss mechanism; case studies trace mechanism but cannot establish how typical it is.
• Rigor versus relevance. The most identifiable questions are often the least consequential; the ones that matter most for policy are hardest to study cleanly.
• Positive analysis versus normative engagement. Staying purely descriptive protects credibility but can abdicate the duty to inform public choice.

Rules of Thumb

• Count the veto players before you predict that a reform will pass.
• If a group "should" mobilize but doesn't, suspect a free-rider problem.
• Be most suspicious of the study whose sample was chosen because it had the outcome.
• A regime's stability is about legitimacy, not just coercion; ask why people obey.
• Specify your theory so an opponent could prove it wrong, or it is not a theory.

Failure Modes

• Selecting on the dependent variable. Studying only revolutions to explain revolutions, with no cases of stability for comparison, guarantees a false conclusion.
• Endogeneity blindness. Treating as a cause something actually an effect, or jointly determined with it — does democracy cause growth, or growth democracy?
• Reifying the model. Mistaking the rational-actor abstraction for a complete account of human beings, explaining away every anomaly rather than learning.
• Ecological fallacy. Inferring individual behavior from group-level correlations, or vice versa.
• Presentism in comparison. Imposing one era's or country's categories — "party," "left and right" — onto cases where they distort.
• Confusing prediction with explanation. A model that fits the past need not identify the mechanism, and may fail when conditions shift.
• Advocacy in disguise. Reverse-engineering analysis to support a conclusion already held, dressed up as method.

Anti-patterns

• Garbage-can causation — throwing every available control into a regression and declaring whatever survives to be the cause.
• The just-so story — a narrative that fits the one case it was built from and is never tested against another.
• Single-level reductionism — explaining war entirely by leaders' psychology, or entirely by system structure, ignoring the other levels.

Vocabulary

• Sovereignty — the claim to supreme authority within a territory, recognized inside and out.
• Legitimacy — the belief among the governed that an authority has the right to rule, making obedience cheaper than coercion.
• Collective action problem — the failure of a group to provide a shared benefit because individuals can free-ride on others' effort.
• Veto player — an actor whose agreement is required to change the status quo.
• Endogeneity — when an explanatory variable correlates with the error term, so its estimated effect is biased — from reverse causation or omitted variables.
• Path dependence — the constraint past choices place on present options through increasing returns.
• Hegemony — preponderant power that shapes the rules and norms others operate within, beyond direct coercion (Gramsci, and in IR the dominant state).
• Median voter — the voter at the center of a single-dimensional preference distribution, decisive under majority rule.
• Anarchy — in international relations, the absence of an authority above states; not chaos but the lack of a higher enforcer.
• Selection effect — distortion arising when the cases observed are not representative of the cases of interest.

Tools

• Statistical software (R, Stata) — for estimation, causal-inference designs, and the large datasets the field runs on.
• Cross-national datasets — Polity and V-Dem for regime characteristics, ANES for U.S. opinion, Correlates of War, the World Values Survey; the shared evidence base.
• Survey methodology — sampling, question design, and embedded experiments to measure opinion and test treatments.
• Formal models — game theory and social choice to derive the consequences of incentives before testing them.
• Comparative case studies — process tracing and structured comparison for the mechanism numbers cannot show.

Collaboration

Political science draws constantly from its neighbors: economists, whose rational-choice models and inference methods the field borrowed and extended; statisticians, on identification; sociologists, on social structure; and historians, who supply the long-run cases. Beyond the academy they advise policymakers, brief diplomats, and inform legislators, where the hedged finding meets the demand for a clear recommendation — translated without overclaiming, and without becoming a partisan.

Ethics

The political scientist studies power and is therefore courted by it. Core duties: keep the positive analysis honest even when the conclusion is politically unwelcome; disclose funding, assumptions, and the limits of the design rather than projecting false certainty; protect human subjects, especially in fieldwork under repressive regimes where candor can be dangerous; label normative claims as such rather than laundering advocacy as science; and weigh public influence, since a misread finding can move votes and lives. The hard cases — advising a government whose ends you distrust, releasing data that could be weaponized — expose the scholar who pretends the work is neutral.

Scenarios

Does foreign aid cause growth? A naive regression of growth on aid finds a weak link and calls aid useless. But aid is not assigned at random — donors send more to countries in crisis, so aid correlates with the very conditions that depress growth. That is endogeneity, and it biases the estimate. Rather than pile on controls, the political scientist looks for an instrument or natural experiment — a funding rule that shifted aid for reasons unrelated to a country's prospects — to approximate the counterfactual. The honest answer is conditional.

Why did this reform die? A government's flagship bill collapses, and pundits blame the leader's weakness. The political scientist counts veto players first: two legislative chambers, a coalition partner, a constitutional court, a federal structure. Five independent gatekeepers, any one of which could block change, and the reform shifted the outcome too far from at least one. The structure predicted gridlock before the first vote; the leader's skill mattered only at the margin. The explanation is institutional, not personal.

Will arming for defense make us safer? Two states, each fearing the other, debate a buildup framed as purely defensive. The realist sees the security dilemma: under anarchy, the neighbor reads capabilities, not intentions, and arms in response, leaving both poorer and no safer — a prisoner's dilemma with guns. The political scientist specifies the conditions that change the payoff: whether offense or defense has the advantage, whether the move is observable, whether institutions can make commitments credible. The recommendation follows from the structure of the game.

Related Occupations

The political scientist shares tools and questions with several fields but is defined by the rigorous study of power and collective choice. Economists supply the rational-choice models and inference methods the discipline runs on; sociologists the social structures within which politics operates; statisticians causal identification. Policy analysts translate findings into recommendations; diplomats and legislators are both objects of study and consumers of the analysis.

References

• Leviathan — Thomas Hobbes
• The Prince — Niccolò Machiavelli
• The Logic of Collective Action — Mancur Olson
• Governing the Commons — Elinor Ostrom
• Theory of International Politics — Kenneth Waltz
• Who Governs? — Robert A. Dahl