Gary Goertz — Social Science Concepts: A User’s Guide

223 published claims from this book.

• A concept states which elements constitute a phenomenon, not merely what it is called.
• A concept’s definition must match the mechanism its causal theory posits
• A concept’s negative pole often has multiple candidates rather than one obvious opposite.
• A concept’s theoretical levels are distinct from its measurement level
• A control group must include only cases that could have shown the effect
• A definition states the essential necessary and sufficient attributes of a thing
• A numeric measure should mirror the structure of its concept.
• A numeric measure with a different structure than its concept measures a different concept.
• A state’s unequal dependence on trade can make it more conciliatory, not less.
• A thing’s causal powers change when its defining properties change.
• A third of conflict variables embed causal hypotheses in their measurement.
• A variable’s absence only counts as a cause when it names a real category
• Abstract concepts need intermediate theoretical dimensions before they can be measured.
• Actively drawing concept structures exposes ambiguities that reading alone hides.
• Actual semantic practice frequently violates logical canons of concept combination
• Adding a dimension to a concept reveals new causes and effects.
• Adding family resemblance dimensions can increase, not shrink, the extension of a concept.
• Adding irrelevant cases inflates significance and can flip causal signs
• Addition and correlation cannot formalize necessary-and-sufficient concept structures.
• Additive quantitative measures translate back into family resemblance concepts.
• Additive scoring embodies a family resemblance structure with no necessary features.
• Aggregating individuals into dyads is structurally identical to aggregating dimensions into concepts.
• Aggregation choice only matters when constituent units are heterogeneous.
• Aggregation choices that boost necessity findings undercut sufficiency findings
• Alliance effects on conflict more than double under a different control population
• Alliances let minor states fight wars they lack the capacity to reach.
• Alternative institutional arrangements can substitute for one another to produce the same outcome
• Ambiguous borderline cases get sorted in ways correlated with outcomes.
• An m-of-n counting rule operationally identifies family resemblance concept structures.
• Analysts often attribute necessary conditions authors may not actually intend.
• Analysts routinely present unstructured dimension lists while assuming a necessary-and-sufficient default.
• Analyzing a concept is equivalent to defining the word for it.
• AND-logic shrinks a concept’s extension while OR-logic expands it.
• Assuming substitutability raises measured constraint and correlates the measures less.
• Baking suspected causes into a concept blocks later hypothesis testing.
• Being human is defined by what humans do, not what they are made of
• Binary coding of concepts creates measurement error by treating unequal cases as equivalent.
• Cases can be excluded from study because their nonoccurrence is overdetermined
• Cases lacking both cause and outcome are irrelevant to a sufficiency hypothesis
• Cases with low values on every independent variable are irrelevant to necessity theories
• Causal and measurement hypotheses cannot be tested simultaneously.
• Choosing negative cases to test a theory lacks clear rules.
• Choosing substitutability between dimensions is the same act as weighting them
• Clustering at a scale’s endpoints signals the scale is too short.
• Comparative coding requires judging functional equivalence across cultures and eras
• Comparative theories reduce to two logical structures at the basic level
• Comparing analyses across differently defined populations reveals selection bias.
• Complex concepts decompose into dimensions that then require data-level measurement
• Composite concepts decompose into a two-level structure of necessary and substitutable dimensions
• Concept construction must explicitly decide how much each dimension weighs
• Concept dimensions constitute their concept through identity rather than causation
• Concept extension tracks how many defining dimensions a rule requires
• Concept formation is an ontological claim about what matters, not mere definition.
• Concept formation logically precedes any measurement or ranking of a variable.
• Concept formation must trace causes, not only measurable effects
• Concept logic and statistical measurement obey incompatible formal structures
• Concept scholarship is a paradox against its theoretical centrality
• Concept structure and dichotomous measurement are logically separate issues
• Concept structure choices are equivalent to case-selection decisions.
• Concept structure itself carries causal hypotheses among its dimensions.
• Concept structure should determine methodology, not the reverse.
• Concept structures differ by how far their attributes can substitute for each other.
• Concept structures form a continuum defined by how far substitutes are permitted.
• Concept-generated categories commonly lack clear boundaries
• Concept-measurement methods should vary with the substantive phenomenon under study.
• Concept-structure choices can be tested empirically by varying substitutability.
• Concepts are warranted by causal theory, not stipulated by definitional fiat.
• Concepts become multidimensional only when broken into secondary-level dimensions
• Concepts contain causal hypotheses, so theories are already two-level
• Concepts defined only by necessary conditions cannot support positive claims about cases.
• Concepts need abstract upper levels and concrete indicator levels to work cross-nationally.
• Concepts on a continuum require defining both poles plus the space between them.
• Concepts used in theoretical propositions occupy a cognitively basic level
• Concepts with direct indicator links need no special measurement theory
• Concepts, like geometric primitives, are best defined by their construction
• Conceptualization choices are ultimately governed by their usefulness in causal theories.
• Constitutive factors describe what makes up a cause but still explain outcomes
• Continuous measurement makes small coding errors flip necessity or sufficiency results
• Control cases should be chosen where the outcome nearly occurred.
• Counting women’s employment misses households excluded by male-earner-designed welfare.
• Culture-specific indicators can satisfy universal necessary conditions in measurement.
• Deep structural properties yield better concepts than observable surface qualities.
• Defining a concept by necessary conditions makes any missing attribute disqualifying
• Defining populations by independent variables smuggles in cases the outcome was impossible for
• Defining the relevant population changes causal findings before any sampling occurs
• Definitions severed from empirical analysis become arbitrary instruments of whoever holds definitional power.
• Democratic transitions and the democratic peace both require treating democracy as a gray-zone continuum.
• Deterrence logic makes power asymmetries screen which contests count as crises.
• Dichotomous and necessary-and-sufficient structures let researchers silently skip dimension weighting.
• Dichotomous variables make weighting dimensions impossible, forcing equal treatment
• Dyadic democracy is operationalized as the weaker member’s democracy score.
• Dyadic measures of democracy face an unresolved problem of aggregating two parties’ scores.
• Empty zones along a continuum signal causally interesting constraints worth investigating
• Essentialism logically entails necessary-and-sufficient condition concept structure
• Essentialist and family resemblance concept structures differ statistically, not just philosophically.
• Even natural-kind concepts shift as empirical knowledge of the object changes.
• Excluding positive cases reveals scope conditions rather than the Possibility Principle
• Explicitly specifying the positive-negative continuum’s character improves research coherence
• Faithfulness to a concept requires measures that existing indicators systematically violate.
• Family resemblance concepts are structured by OR, the set-theoretic union.
• Family resemblance concepts can gain extension by adding attributes rather than losing them.
• Family resemblance concepts contain no necessary conditions, only sufficient resemblance.
• Family resemblance concepts hold when at least one dimension is present.
• Family resemblance concepts require sufficient attributes, never necessary ones
• Flat additive measures erase the multilevel structure of the underlying concept.
• Formal logic requires substantive concepts to be scientifically useful
• Formalizing concept structure improves both theory and quantitative measurement
• Functionally equivalent indicators can be uncorrelated; low correlation may even confirm substitution.
• Gendered analyses reclassified the welfare state from effect to cause of women’s well-being.
• Generalizing a concept requires stripping away attributes until it travels across cases
• Good concept structure follows the substance of the phenomenon, not the discipline
• High inter-measure correlation can conceal systematic, non-random measurement divergence.
• How you define a concept silently biases which populations your study selects.
• Human capabilities cannot substitute for one another in any quantity
• Ideal types are defined by having zero empirical extension
• Ideal types function as standards with near-zero actual instances.
• Ideal-type concepts have near-zero empirical extension by definition
• Ideally coded dichotomous models predict all positive and negative cases
• In a dyad, the less democratic state sets the rules of engagement.
• In dyadic trade, the more dependent state is the more constrained one.
• In social science, the negative pole of a concept usually lacks independent theoretical content.
• Increasing a concept’s intension can increase, not shrink, its extension
• Increasing a term’s intension never increases the set of things it denotes.
• Indicator-concept links can run causally in either direction or not at all.
• Indicator-level factors often serve as finer-grained causes of secondary-level dimensions.
• International crisis is defined by three perceived conditions in top decision-makers
• Irrelevant added dyads numerically overwhelm the cases that actually matter
• Joint constraints on war are set by the less democratic member of a pair.
• Ladder metaphors of degree hide the negative pole concepts require.
• Latent-variable measurement assumes the underlying concept causes its indicators.
• Later researchers treat contested datasets as natural facts rather than as constructed artifacts.
• List-style concepts are routinely misread as necessary conditions across the social sciences.
• Listing a concept’s dimensions without logical structure leaves measurement entirely unguided.
• Logical connectives and statistical operators are distinct measurement choices, not interchangeable operations.
• Logically equivalent concept structures can differ sharply in theoretical meaning.
• Long-term rivalries between unequal powers are common, so rivalry need not imply parity.
• Loosening a concept’s attributes increases the number of cases it covers
• Loosening a concept’s definition buys more cases at the cost of meaning
• Many social science concepts are defined by their constituent dimensions, not caused by them.
• Measurement choices usually drive concept definition in social science, not the reverse.
• Measurement is impossible until the underlying concept has been formed.
• Measures spread rapidly when they are tied explicitly to a convincing theory.
• Measuring a concept by its own logic yields different rankings than standard indices.
• Methodology courses abandoned concept analysis, which Mill rightly treated as foundational.
• Most democracy concepts use necessary-and-sufficient structure while their measures use family resemblance.
• Most measurement assumes effect indicators, so misspecifying causal indicators yields biased estimates.
• Most scholars define democracy as necessary and sufficient conditions
• Most state pairs in world history have no realistic opportunity for war.
• Most state pairs lack the proximity and power to fight at all
• Multiple institutional paths can realize the same political goal
• Multiple means to one end are substitutes, not causes, of that end
• Naming the missing attribute lets a diminished subtype increase differentiation rather than reduce it
• Narrowing the object from class to family simultaneously generalized the study beyond one class.
• Necessary conditions can coexist with substitutable indicators at lower levels.
• Necessary-and-sufficient concepts behave cleanly under dichotomous two-valued logic
• Negative-case universes remain undefined even when core concepts are clearly specified.
• No single attribute is necessary for something to count as a game
• One family of power means unifies minimum, mean, and maximum as substitutability endpoints.
• Only changes to an object’s essential properties alter what kind of thing it is.
• Only negative cases where the outcome was possible belong in an analysis
• Ontological concept analysis can be formalized without being unscientific
• Ontological concepts ask what something is, not merely what indicates it
• Operationalization is a distinct concept level where data collection becomes possible.
• Opportunity for war is measured by great-power status or shared borders
• Parliamentary and presidential regimes differ on whether executive and legislature depend on each other
• People categorize things by similarity to prototypes rather than by shared defining features
• Philosophical interpretation reorganizes minute scientific details without exceeding their range
• Quantitative and qualitative scholars hold tacitly different models of what a concept is
• Quantitative measures routinely use mathematics mismatched to their concept’s logic.
• Rare outcomes and rare causes mean most cases occupy the irrelevant zone
• Reclassifying the same democracy data as essentialist changes results significantly.
• Redefining a concept to fit a case alters the hypotheses built on it.
• Redundant substitutable means make social and weapons systems more stable
• Rival democracy measures disagree on classification for 38 percent of country-years.
• Scholarly consensus can fix on the weaker of two causal factors
• Scientific concepts are theories about phenomena, so their definitions shift with new knowledge.
• Secondary-level variables act causally on outcomes rather than measuring concepts
• Selecting a population first creates boundary anomalies between cases
• Selecting cases with extreme values on causes biases tests against necessity
• Semantic analysis of definitions alone cannot justify a social-science concept.
• Skewed empirical distributions justify causal theories rather than concept revisions.
• Skocpol’s theory makes state breakdown and peasant revolt jointly sufficient for revolution
• Some logical structures for combining concept dimensions are much rarer than their converses.
• Specifying which attribute is missing selects a distinct set of cases
• Spending programs constitute the welfare state rather than being caused by it
• Statistical findings depend entirely on which negative cases enter the population
• Statistical latent-variable models make unmeasured concepts cause their indicators.
• Statistical representativeness and the welfare state emerged together
• Statistically significant results can mask large, decisive differences between measures.
• Substitutability assumptions can be tested by building rival aggregation measures.
• Substitutability forms the underlying continuum spanning AND and OR concept structures.
• Substitutable causes in set-theoretic analysis resolve into a more abstract shared concept.
• Summing indicators while the concept demands ANDs makes measure and concept inconsistent.
• Terminological drift signals an unstable concept and unsettled underlying continuum
• Testing two-level causal theories demands concentrated depth over superficial breadth
• The billiard-ball metaphor of causation shaped how quantitative scholars define concepts.
• The choice of fuzzy-logic operator depends on data type, not conceptual meaning.
• The cognitively central concept level is the one theoretical propositions actually use.
• The intension-extension tradeoff only holds under necessary-and-sufficient definitions.
• The inverse intension-extension relation holds only for necessary-and-sufficient concepts
• The inverse relationship between extension and intension does not hold for concepts in general.
• The same hypothesis can reverse sign depending on the case population chosen
• The theory of a concept dictates which statistical measure of it is valid.
• The weakest-link measure of dyadic democracy is a testable claim about substitutability.
• The welfare state is structured as a family resemblance concept, not a necessary-condition one.
• Theories built from conjunctions and disjunctions resist simple regression-style testing.
• Theory determines which negative cases qualify as relevant controls.
• Trivial threshold changes can slash a case population while barely losing real disputes
• Unattainable ideal types function as measuring sticks for real cases.
• Undocumented indicator weights can quietly turn dimensions into necessary conditions.
• Unstated negative-pole assumptions drive scholarly disagreements about a concept
• Using a variable to select cases shrinks its measured causal effect
• Using the same rule to pick cases versus populations yields opposite inclusion errors
• Visual concept diagrams place diminished subtypes and added-attribute subtypes at the same level.
• Weakest-link aggregation is itself a causal theory about how democracy prevents war.
• Weakest-link measures embed an untested necessary-and-sufficient hypothesis.
• Weighting in additive scores can quietly make dimensions necessary.
• Weighting indicators is meaningless under necessary-and-sufficient concept structures
• Welfare provision presupposed unpaid female care work that made benefits possible
• Welfare-state measures funnel household benefits through the male wage earner.
• Widespread weakest-link use converted strict nonsubstitutability into an untested working assumption.
• With continuous variables, logical OR equals the maximum observed value
• Zero-value cases artificially satisfy causal sufficiency unless theoretically excluded