CM 2: Approaches in political science, part 2

What are the steps of the scientific cycle?

Research question → empirical analysis → empirical test/falsification → scientific debate

What does RQ → RD → RA stand for?

Research Question → Research Design → Research Answer

What is a research question (RQ)?

What do I want to know?

What is a research design (RD)?

The bridge between the question and the answer: how am I going to find out? (cases, data, variables, method)

What is a research answer (RA)?

What did I find? What the evidence suggests.

What is a theory?

A relationship between a cause and an effect (X → Y).

In a theory, what are X and Y?

X = cause / explanatory factor. Y = effect / outcome. (ex. education investment increases GDP growth)

What 3 qualities must a good research design produce?

Valid, reliable and generalizable results.

What is validity?

Does the research really allow me to draw the conclusion I draw?

What is reliability?

Does the method produce consistent, reproducible results?

What is generalizability?

Can the result be extended to similar cases that were not studied?

What is internal validity?

Are my conclusions correct for the cases I actually studied?

What is external validity?

Can I generalize my findings to similar cases I did NOT study?

What is the trade-off between many cases and few cases?

Many cases → stronger generalizability (external validity ↑) but less depth.

Few cases studied deeply → better understanding of mechanisms (internal validity) but less generalizable.

Why are experiments harder in social sciences?

Controlled experiments are difficult, so we mainly use statistical methods, comparative methods and case studies.

What is a case study design?

1 case + 1 point in time.

What is a time series design?

1 case + many points in time.

What is a closed universe design?

Few cases + few time intervals.

What is a cross-section design?

Many cases + 1 point in time.

What is a pooled analysis?

Many cases × many time intervals.

What does "rising N" mean?

N = number of cases/observations.

Small N → richer detail, more qualitative. Large N → more statistical tests, more quantitative.

What is the quantitative approach? (for the RQ)

Collecting and analysing numerical data using statistical methods.

What are the 3 sources of quantitative data?

Surveys, observations (coding info numerically), secondary data (governments, international organizations, databases).

ex. is quality of life higher in democracies or autocracies ? (compare poli. regimes)

What is a dependent variable (DV)?

What you want to explain: the outcome.

ex. why is inequality increasing ? (inequality, here, is what you want to explain)

What is an independent variable (IV)?

The factor you think explains the outcome.

In "Does education investment increase GDP growth?", what are the IV and DV?

IV = education investment. DV = GDP growth. (GDP growth DEPENDS ON the education investment)

Why is correlation ≠ causation?

Two variables moving together doesn't mean X causes Y: a third variable Z may influence both.

What are the 3 functions of comparison?

Describe, Control, Explain.

What does comparison do when it DESCRIBES?

It shows similarities and differences (patterns, regularities). E.g. Tocqueville, Democracy in America.

What does comparison do when it CONTROLS?

It holds some variables constant/eliminates them. E.g. Tunisia and Egypt 2011 (similar religion, colonial past, geography → why different outcomes?).

What does comparison do when it EXPLAINS?

It builds/tests a theoretical explanation: WHY is there a difference? E.g. why is voter turnout lower in the US than in Canada?

What is a case?

A comparable unit of observation.

What are the macro, meso and micro levels of a case?

Macro = country. Meso = regional party / institution. Micro = individual.

What is the difference between a case and a unit of observation?

Case = the broader unit studied (e.g. Belgium).

Unit of observation = the actual data points (e.g. individual voters in Belgium).

In a data matrix, what are the rows and columns?

Variables = columns. Cases = rows.

What is the aim of case study research?

Explaining causality in a single case: how does X produce Y?

What is a necessary condition?

X must be present for Y to happen. Without X, no Y. But X alone doesn't guarantee Y.

E.g. oxygen for fire.

necessary = you need it, but that isn't necessarily enough

What is a sufficient condition?

If X is present, Y will occur. But X isn't necessarily needed: other paths to Y may exist.

What is the Most Different System Design (MDSD)?

Based on Mill's Method of agreement (1875). Different cases, similar outcomes.

What is the Most Similar System Design (MSSD)?

Based on Mill's Method of difference (1875). Similar cases, different outcomes.

What is the logic of MDSD?

Maximize differences between cases, assuming that factors that differ across cases are irrelevant to explain the similar outcome.

What is the aim of MDSD? Give an example of MDSD.

Highlight a potential sufficient condition, removing all conditions that are not necessary.

Causes of revolution in Russia, France and China (Theda Skocpol).

What is the logic of MSSD?

Select the most similar cases, assuming that factors common to the cases are irrelevant to explain the difference in outcome.

What is the aim of MSSD? Give an example

Highlight a potential necessary condition, removing all conditions that are not sufficient.

Why did Tunisia and Egypt differ in their 2011 uprising outcome?

Which Mill method corresponds to MDSD and which to MSSD?

Method of agreement = MDSD. Method of difference = MSSD.

What is a causal mechanism?

A complex system of interactions between different parts that produce a result: X → M1 → M2 → ... → Y.

What is the key question to ask about a causal mechanism?

WHO DOES WHAT? (which actors, which actions)

Is a causal mechanism the same as an intervening variable?

No. A mechanism shows the process (who does what, leading to the next step), not just a statistical variable.

What is narrative history?

A chain of events without causal mechanisms. Chronology = what happened, not how/why.

What is conceptualization?

Going from theory to empirics: from an abstract concept to something observable in data.

What is the conceptualization path?

Concept (ex. democracy) → Definition → Operational definition (how can I observe/ recognize democracy empirically? : presence of civil liberties) → Indicators (Freedom of association, freedom of organization, rule of law, protection of personal liberties) → Empirical observation.

What is conceptual stretching?

Extending a concept (A → B) to more cases, at the risk of losing intension (precision).

What is intension?

The precision of the meaning of a concept.

What happens with more intension?

More precise definition → internal validity ↑, but comparative applicability ↓.

What happens with more extension?

Applicable to more cases → external validity ↑, but risk of losing meaning.

What is family resemblance?

A category defined by a network of overlapping similarities: members share some features, but not necessarily one single defining feature. E.g. fascism.

What are radial categories?

ONE essential/core attribute + optional secondary attributes. Different secondary attributes → different sub-types. E.g. democracy.