Notes on Empirical Falsifiability, Theory, and Scientific Method

The nature of theories and empirical falsifiability
  • Theories are empirically falsifiable and have withstood attempts to falsify them; they have empirical support.

  • You cannot prove a theory correct with statistics; you can only fail to falsify it (or fail to disprove it).

  • Observations in most sciences are probabilistic, not deterministic.

  • The null hypothesis framework tests causation: reject H0H_0 (no relationship) to accept HAH_A (relationship).

  • Statistics assess the probability of a result occurring by chance, not ultimate proof.

  • Anecdotes are not proofs; systematic evidence is needed to falsify a theory, not isolated cases.

  • Falsification, not definitive proof, is central to scientific progress.

Determinism, realism, and the object/subject distinction
  • Objects (e.g., rocks) obey physical laws, while subjects (humans) have cognition, can resist, and react to being studied (reactivity/Hawthorne effect).

  • Reactivity complicates control of environmental and belief factors.

  • Social sciences face greater complexity due to studying subjects with cognitive constraints and social contexts.

The scientific process in criminology: causation, prediction, and theory-building
  • Criminology aims to explain and predict crime rates through plausible causes and mechanisms (e.g., low self-control leading to delinquency).

  • The process involves observing patterns, proposing explanations, testing predictions, and refining theories.

Conceptualization and the ontology-epistemology pair
  • Ontology: Questions about what is real (the nature of reality).

  • Epistemology: How we know what we know; the scientific method.

  • The scientific approach assumes realism: an external reality exists, and our knowledge is refined through definitions and observation.

Nominal, operational, and theoretical definitions; measurement and observation
  • The pathway from concept to measurement involves:

    • Concepts (e.g., low self-control, delinquency).

    • Nominal definitions: Lay terms or precise definitions of concepts.

    • Theoretical proposition: Link between concepts (e.g., low self-control leads to delinquency).

    • Operational definitions: Concrete, measurable indicators (e.g., questionnaires for self-control).

    • Observation/measurement: Collecting data; ensuring measurement validity.

    • Hypothesis: A testable statement about relationships between variables.

The empirical test cycle: replication, falsification, and robustness
  • Science progresses through replication; a single study rarely settles a question.

  • Replication across contexts and populations is crucial, as early findings may not generalize (e.g., domestic violence arrest experiment).

  • Robust science relies on ongoing testing and refinement.

The role of theory and policy implications in criminology
  • The ultimate goal is to identify influenceable factors to reduce crime and improve social welfare.

  • Policymakers should rely on robust, replicated evidence before large-scale implementation of interventions.

Practicalities of scientific modeling in criminology
  • The modeling process defines concepts, proposes theoretical links, operationalizes variables, collects data, and tests hypotheses.

  • Assumptions: Taken as given (e.g., reality, human nature).

  • Propositions: Asserted relationships between constructs.

  • Hypotheses: Testable predictions from propositions.

Social order, human nature, and the structure of society
  • Consensus model: Society organized by agreement on shared values.

  • Conflict model: Social order arises from negotiation and conflict among groups.

  • Pluralism: Society is diverse; social order emerges through negotiation and adaptation (analogy: tossed salad, not melting pot).

Key formulas and formal notations (to be used in study and exams)
  • Null and alternative hypotheses:

    • H0:ρLSC,delinquency=0H_0: \rho_{LSC, delinquency} = 0

    • HA:ρLSC,delinquency≠0 (or ρLSC,delinquency>0 if a directional test)H_A: \rho_{LSC, delinquency} \neq 0 \text{ (or } \rho_{LSC, delinquency} > 0 \text{ if a directional test)}

  • Correlation/association measure:

    • r=Cov(X,Y)σXσYr = \frac{\mathrm{Cov}(X,Y)}{\sigma_X \sigma_Y}

  • Conceptualization-to-measurement pipeline (summary): concept (\rightarrow) nominal definition (\rightarrow) theoretical proposition (\rightarrow) operational definitions (\rightarrow) observation/measurement (\rightarrow) data (\rightarrow) hypothesis (\rightarrow) test/replication

  • Determinism vs probabilistic science: science often tests probabilistic relations, with replication and counter-evidence shaping confidence in findings.