POLI 110: Investigating Politics

An Introduction to Scientific Political Analysis

Lecture 5: Theory building, observable implications, & hypotheses

Monday, May 27, 2024
Roadmap for today’s lecture
1. Recapping research questions
2. What is a theory?
3. What are the sources of theory?
• Inductive vs. deductive theory building
4. What makes a good theory?
5. “Good Question, Bad Answer” breakout room activity
6. What form can a theory take?
7. General vs. Specific Knowledge
8. Moving from theory to observable implications to hypotheses
Lecture Check-in
• Pop over to Canvas
• Under the Week 3 discussion - Monday, add a comment noting
you’re checking in for today’s lecture and what your favourite
colour is.
• E.g. Professor Konken is here, my favourite colour is green.
Topic à Question à Claims à Theory
• Topic: single noun expression
• Question: a sentence in an interrogative form, addressed to
someone in order to get information in response.
• Claims
• Descriptive
• Causal
• Prescriptive/Normative
• Theory: a causal explanation for an outcome.
• A series of interrelated claims leading to observable implications
• Observable implications serve as the foundation of hypotheses
Where do research questions come from?
Research is centred around an underlying question that you, as the
researcher, are motivated to learn more about, and possibly answer.

Is your motivation to describe, explain, predict, or to understand?

● Descriptive Questions
● Causal Questions
● Predictive Normative Questions
Where do research questions come from?
1. What concepts are you curious or passionate about?
• Concepts are abstractions: allow you to describe, classify,
or organize different political phenomena
• Concepts exhibit variation
2. Questions from existing research
3. Encountering a research “puzzle”
A good research question…
● Is clear and focused.
● Is concise, but nuanced.
● Is feasible, with time and resources available.
● Leaves room for debate.
● Contributes to understanding.
The role of theory
Theories seek to…
• Explain what happened
• Predict a future outcome
• Explain differences between cases (i.e.., individuals, events, countries,
legislative sessions, etc…)
• Explain changes over time
Theories seek to explain nature of relationships between concepts
• “How” are they related?
• “When” are they related?
• “Why” are they related?
Defining a causal theory
• A social scientific theory is “a reasoned and precise
speculation about the answer to a research question,
including a statement about why the proposed answer is
correct” (King, Keohane, and Verba 1994, 19).
• More simply: a theory refers to a causal explanation for an
outcome.
• It explicitly contains a something-leads-to-something-else claim.
Causal theory vs. conceptual/theoretical
frameworks
• A causal theory is different from empirical research involving
conceptual or theoretical frameworks
• Theoretical or conceptual frameworks use theoretical tools or
concepts as lenses to ‘explain’ phenomena.
o “shed light on” what is happening and defining outcomes.
o E.g. feminist theories of war; post-colonial theories of economic growth
• Causal theories seek to explain how, when, and why an
outcome occurs.
Example of causal vs. theoretical/conceptual
Example of causal vs. theoretical/conceptual
Causal question: how do legacies of wartime violence shape
the dynamics of protest in twenty-first-century Iraq?
• Causal theory:
• experiences of shared violence against civilians generate strong social
and organizational ties. These strengthened social networks endure
beyond the end of the conflict, forming important vehicles that can
facilitate the organization of protest when new grievances or
opportunities arise.
• Conceptual/theoretical framework: the “what”
• “Experiences of shared violence”
• “Strong social and organizational ties”
All theories are wrong, but some are useful
• Theories are simplifications of reality
• No theory is trying to explain every single detail and every facet of an
event or phenomenon.
• But the insights we gain from these simplifications can be useful
and guide policy prescription.
• Critical assumptions of a theory should mirror reality.
Parts of a theory
A typical theory has four parts:
1. Expectation/Prediction: what the author expects to happen.
2. Causal mechanism: It provides a specific chain of steps,
series of links, or other specific accounting of how changes in
the cause variable affect the outcome variable.
3. Assumptions: claims about how the world operates.
4. Scope conditions: identify the theory’s domain – the set of
cases over which the theory is expected to operate.
What makes a good theory
Theories should be…
• Parsimonious: A simplification of a complex process.
• Generalizable: Applicable beyond a single case.
• Falsifiable: A belief is not a theory.

Note: there are trade-offs between parsimony and

generalizability.
Theory building cycle
Sources of theory
Inductive theorizing: the potential explanation emerges from a
study of specific cases, and then is generalized outward.
• Role of common sense and personal experience?

Deductive theorizing: potential explanations emerge from

abstract analysis outside of the context of any specific case. We
step back from the data and think, in general, about ways in
which something could occur or what could cause a certain
outcome.
Theory-Building research
Inductive theory-building is a bottom- Theory
up process that makes generalizations
based on specific observations.
Tentative
Hypothesis

Pattern

Observation
Example of inductive research
RQ: What explains strong, emotional opposition to
environmental regulation in highly-polluted Louisiana?
• People don’t want pollution and suffer negative health
consequences yet they oppose government regulation.
Data: Extensive research, mixing interviews, focus groups,
participant observations and government statistics to
understand how government regulation is perceived and
understood by Tea Party supporters in Louisiana.
Findings? Widespread perception government
interventions do not benefit “people like me.”
1. Cultural values, religious belief, and strong attitudes about the role
of government
2. Coupled with perceived unfairness inherent in redistribution
policies lead many Louisiana voters to support deregulation.
Theory-testing research
Deductive theory-testing is a
Theory
top-down approach that begins
with an established theory and
draws on observations to test the
Hypothesis
theory by weighing evidence for /
against specific hypotheses.
Observation

Confirmation /
Falsification
 Example of deductive research

Research question: Why is there a gender gap in

political knowledge?
Theory: Not a question of ability, but socialization.
• Women’s political engagement tends to be lower in
countries with more pronounced gender inequality.
• Disparities in the representation of men and women
in political office creates a socialization effect that
reinforces (dis)engagement from politics.

Hypothesis: “…as the share of women elected

representatives increases, these new female
politicians will serve as role models, strengthening
women’s engagement with politics” (p. 250).
Data: “Comparative Study
of Electoral Systems”
(CSES) & 2009 “European
Election Study” (EES).

Survey data showing

estimated size of gender
gap in political knowledge,
across many established
and emerging democracies
in Europe.
Results: Descriptive
representation at a young
age is especially important.

The greater proportion of

women elected when
respondents’ are young
adults (ages 18-21), the
smaller the gender gap in
political knowledge,
Why theorizing by induction can be problematic:

Research Process with Research Process with

Inductive Theorizing Deductive Theorizing
1. Get data 1. Develop theory
2. Poke around in it, 2. Project theory to
looking for patterns cases (hypothesize)
3. Develop theory 3. Get data
4. Project theory to 4. Poke around in it,
(additional) cases looking for patterns

When might it be proper to generalize from limited cases?

In practice: dynamic theory building is iterative
• Iteration is a core aspect of research in areas for which theory
or empirical knowledge is underdeveloped and requires
reacting as the research environment evolves.
• What is most important is transparency
Everyday theorizing

“Theorizing is not a hard task in and of itself. In fact, you theorize

every day, as part of your ordinary daily behavior. You do it without
even thinking about the fact that you’re doing it. For instance,
consider your route across campus to your noon class”
(Powner, Revised Chapter 2)
What is an example of every day
theorizing that you do?

ⓘ Start presenting to display the poll results on this slide.

Good Question, Bad Answer
Question: Why do countries start wars?
Bad answer: Because they wanted to fight.

Better answers:
• because they felt threatened by their adversary
• because they wanted something their adversary had
• because they believed they could win…
 Good Question, Bad Answer
Question: Why do some countries have parliamentary systems of government while
others have presidential systems?
Bad answer: Because that’s what their Constitution says they should have.

Better answers:
• countries with multiple social cleavages may have wanted a system that allowed
for multi-party systems, which tend to work best with parliamentary systems;
• countries that formerly were monarchies may have evolved into constitutional
monarchies;
• countries that had bad experiences with presidents-turned-dictators may want to
avoid concentrating power in one person again…
 The Scientific Process
Broadly speaking : 1. Causal
theory
1. Ask a research question
6. Scientific 2.
2. Define and apply a procedure to knowledge Hypothesis
record observations
3. Analyze the data, revise the theory*
4. Communicate results
5. Evaluate 3. Empirical
5. Replicate theory test

4. Evaluate
evidence
10 Minute Break
Breakout room activity
We will randomly assign you to groups of 4-5. Within your group,
choose 1 of the following “Good Question, Bad Answer” activities.
Brainstorm a series of better theories explaining the question.
Designate one person from your group to post your answers on
Canvas. Be sure to include all of your group members names in the
post.

1. Q: Why do some incumbents lose elections? A: Because they got fewer votes
than another candidate.
2. Q: Why do states legalize marijuana? A: Because they want to.
3. Q: Why do some marginal political groups turn to violence (terrorism) and
others don’t? A: Because they think violence is the answer.
4. Q: Why do some school districts reject literacy programs based on the science
of reading? A: Because they’re anti-science.
Format of theories
• Structural causes vs. triggering events
• Necessary vs. sufficient conditions
• Causal complexity
• Conjunctural causation
• Multiple causation
• Multiple and conjunctural causation
Structural causes vs. triggering events
The triggering event stems from a broader “structural cause” of the
outcome we seek to explain
• Triggers are substitutable
• Any spark will start a blaze if there’s enough fuel
• When structural cause present, tons of things could trigger the outcome
• Some trigger may be necessary; but the specific trigger is probably
causally unimportant – a detail of how/when

Social scientists are more interested in structural causes than

triggering events
 Causes of Tunisian Regime Collapse
Spark
Bouazizi self-
immolation
Oppressive Collapse of
corrupt regime Tunisian regime
Fuel
Necessary conditions
Sometimes our causal claims take the form:

“Condition C must happen for outcome E to emerge.”

Condition C = necessary condition for E.

Necessary conditions
Necessary condition: a condition without which an
outcome cannot occur.

• A large middle class is a necessary condition for the

emergence of democracy.
• High levels of immigration are a necessary condition
for a government to adopt multiculturalism policies.
China’s rising middle class
China’s Middle
Class (2022)
What does it take to pass climate
legislation in the U.S.?
• A President committed to solving climate change

• Democratic Party majorities in both houses of Congress

• A public willing to support action on climate

• A proposal that does not threaten the fundamental interests of

well-financed interests
Many necessary conditions for climate
action in the U.S.
• A President committed to solving climate change

• Democratic Party majorities in both houses of Congress

• A public willing to support action on climate

• A proposal that does not threaten the fundamental interests of

well-financed interests
Sufficient conditions
Sufficient condition: a cause that ALWAYS produces an effect.
• A fire is sufficient to cause heat.
• Social science examples?
VERY RARE
BECAUSE
SO MUCH CAUSATION IS CONJUNCTURAL
(REQUIRES MORE THAN ONE CONDITION)
Conjunctural causation
Conjunctural causation: when an effect depends on a
combination of causes.
A specific version:
• Multiple necessary conditions: the effect only happens when
more than one condition is present
• Plant growth requires BOTH water AND sunlight.
• Revolutions require BOTH material deprivation AND arbitrary rule.
• War requires BOTH miscalculation AND strong offensive capabilities.
Conjunctural causation
Conjunctural causation: when an effect depends on a
combination of causes.
The more general form:
• Conditional effect: the effect of cause C on E depends on cause D.
• An hour of sunlight causes…
• 1 centimeter of plant growth when it is not raining
• 2cm of plant growth when it is raining
• The effect of sunlight depends on whether it’s raining
• Rain amplifies sunlight’s effect on growth
Conjunctural causation
Conjunctural causation: when an effect depends on a
combination of causes.
The more general form:
• Conditional effect: the effect of cause C on E depends on cause D.
• Poverty causes…
• Short civil war when ethnic diversity is low
• Long civil war when ethnic diversity is high
• The effect of poverty on civil war depends on how much ethnic diversity
there is
• Ethnic diversity amplifies poverty’s effect on civil war
Multiple causation
Multiple causation: when there is more than one set of causes
that can produce an effect.
A can cause E
OR
C can cause E

What can cause political leaders to lose power?

• Military defeat (Argentina , 1980s)
OR

• Economic decline (UK Labour, 2010)

Multiple & conjunctural causation
A combination of A + B can cause E
OR
A combination of C + D can cause E

Why do civil wars happen?

• A combination of Ethnic diversity AND political inequality between groups (Sri
Lanka)
OR
• A combination of Meddling by foreign powers AND deep ideological
polarization (Nicaragua)
To recap: key causal concepts so far
• Counterfactual definition of causality

• Structural causes vs. triggering events

• Necessary vs. sufficient causes

• Causal complexity
• Conjunctural causation
• Multiple causation
• Multiple and conjunctural causation
Question: does democracy affect economic
development?
• Lots of rich democracies in the world

• Lots of poor authoritarian countries

• But also poor democracies and rich authoritarian countries

• Does democracy lift a country out of poverty?

Democracy scores
The world’s rich and poor nations
Democracy

Development
Looks like a relationship
Causes of economic development
• Democracy is a necessary condition for development
• Democracy is a sufficient condition for development
• Another possibility:
Democracy makes economic development more likely.
Deterministic vs. probabilistic causal claims
• Deterministic causal claim
• A claim about what must happen or cannot happen as a result of
particular causal conditions
For instance:
• The outcome cannot happen when the cause is absent
• The outcome must happen when the cause is present
• Necessary or sufficient conditions claims are deterministic claims
• Probabilistic causal claim
• A claim that a cause makes an outcome more or less likely to occur
U.S. midterm elections: effect of the economy
Deterministic vs. probabilistic causal claims
• Deterministic causal claim (sufficient condition)
• “High income growth is sufficient to prevent a large loss of seats by the
President’s party at midterm elections.”

• Deterministic causal claim (necessary condition)

• “High income growth is necessary to prevent a large loss of seats by the
President’s party at midterm elections.”

• Probabilistic causal claim

• “High income growth growth makes it less likely that the President’s party
will lose a large number of seats at midterm elections.”
Framing probabilistic claims
• “Oppression is likely to cause a rebellion.”
• No. This is saying oppression is a probable cause of the
outcome.
• Need to say the cause changes the likelihood of the outcome.

• “A rebellion is more likely to occur if the population is oppressed”

“A country undergoing oppression is more likely to rebel.”
• No. Not clearly causal; could be just a correlation.
Framing probabilistic claims
“Oppression makes rebellion more likely to occur.”
• Yes
• Clearly causal
• Clearly probabilistic
• The cause changes the probability of the outcome.
Inequality in Canada:
The income share of the top 1%
Specific knowledge
• What caused inequality in Canada to go up since
1980?
• Chretien’s and Harper’s tax cuts caused
inequality to go up in Canada since 1980.
• The decline of the Canadian Labour Congress
caused economic inequality to go up in Canada
since 1980.
• We’ve explained a case
Inequality in liberal welfare states:
The income share of the top 1%
Specific vs. general knowledge

• What causes inequality to rise in liberal welfare

states?
• Tax cuts for upper-income groups cause inequality
to rise in liberal welfare states.
• Declining trade unions cause inequality to rise in
liberal welfare states.
• We’re explaining a class of phenomena
Specific vs. general knowledge
SPECIFIC
• Chretien’s and Harper’s tax cuts caused inequality to go up in
Canada since 1980.
• The decline of the Canadian Labour Congress caused economic
inequality to go up in Canada since 1980.
à Explains a case

GENERAL
• Tax cuts for upper-income groups cause inequality to rise in liberal
welfare states.
• Declining trade unions cause inequality to rise in liberal welfare
states.
à Explains a class of phenomena
Specific vs. general knowledge
SPECIFIC
• Chretien’s and Harper’s tax cuts caused inequality to go up in
Canada since 1980.
• The decline of the Canadian Labour Congress caused economic
inequality to go up in Canada since 1980.
à Explains a case

GENERAL
• Tax cuts for upper-income groups cause inequality to rise in
liberal welfare states.
• Declining trade unions cause inequality to rise in liberal welfare
states.
à Explains a class of phenomena
Specific vs. general knowledge
SPECIFIC
• Chretien’s and Harper’s tax cuts caused inequality to go up in
Canada since 1980.
• The decline of the Canadian Labour Congress caused economic
inequality to go up in Canada since 1980.
à Explains a case

GENERAL
• Tax cuts for upper-income groups cause inequality to rise in
liberal welfare states.
• Declining trade unions cause inequality to rise in liberal welfare
states.
à Explains a class of phenomena
Specific vs. general knowledge
SPECIFIC
• Chretien’s and Harper’s tax cuts caused inequality to go up in
Canada since 1980.
• The decline of the Canadian Labour Congress caused
economic inequality to go up in Canada since 1980.
à Explains a case

GENERAL
• Tax cuts for upper-income groups cause inequality to rise in liberal
welfare states.
• Declining trade unions cause inequality to rise in liberal welfare
states.
à Explains a class of phenomena
Specific vs. general knowledge
How do I transform a specific claim into a more general one?
• From specific people, places, events à categories and concepts
that encompass those specifics
• Canada à liberal welfare states
• Chretien tax cuts à tax cuts for the rich
• Canadian Labour Congress à trade unions
• Consider swapping out proper nouns for concepts
Vancouver’s housing crisis
Specific vs. general knowledge
• SPECIFIC
• Why is Vancouver housing so expensive?
• Rules limiting 65% of the city to single-family dwellings have
caused Vancouver housing prices to rise rapidly.
• Explains the case of Vancouver’s housing unaffordability
• GENERAL
• Why is housing so expensive in some cities?
• Zoning rules that limit density cause housing prices in cities to
rise rapidly.
• Explains a general phenomenon: housing unaffordability in
cities.
Specific vs. general knowledge
How do I transform a specific claim into a more general
one?
• From specific people, places, events à categories and
concepts that encompass those specifics
• Vancouver à cities
• Rules limiting 65% of city to single-family dwellings à zoning
policies that limit density
Mixing specific/general with probabilistic
SPECIFIC (and PROBABILISTIC)
• The wide availability of AR-15 rifles in Florida made the
Parkland school shooting more likely to occur.

How can we make this general?

GENERAL (and probabilistic):

• The wide availability of semi-automatic weapons makes mass
homicide more likely.
Specific vs. general knowledge
How do I transform a specific claim into a more general one?
• From specific people, places, events à categories and concepts that
encompass those specifics
• Wide availability of AR15s in Florida à wide availability of semi-
automatic weapons
• Parkland shooting à mass homicide
• Dropped “in Florida”
From specific to general causal claims
• SPECIFIC: The wide availability of guns in Florida made the
Parkland school shooting more likely to occur.
• What PARTS of this statement can we generalize?

• GENERAL: Wide availability of lethal weapons makes mass

homicide more likely.
or…
• GENERAL: Wide availability of lethal weapons makes terrorism
more likely.
Theories versus hypotheses
Theories are propositions about Highly
how concepts are related abstract Concept

Concrete
Hypotheses are expectations representation of Variable
about how variables are a concept
related.

The process of moving from one Very specific:

to the next is called observable instance Indicator
“operationalization.” of the variable in the
real world
Theories generate observable implications
To move from concepts to variables (operationalization), our theory
should provide clear observable implications; i.e. what evidence
needs to exist to demonstrate that our causal theory is correct.
Theory:
The wide availability of semi-automatic weapons makes mass
homicide more likely.
Observable implications:
In areas where semi-automatic weapons are more available,
incidents of mass homicide are more likely.
Hypothesis testing
Theories motivate hypotheses: A specific expectation about some
relationship derived from theory.
○ Hypotheses must be falsifiable to be empirically testable (i.e., are not
normative statements).
○ Hypotheses must specify the variables included in the relationship.
○ Hypotheses must also specify how the variables are related (the direction of
the relationship; positive or negative).
Every hypothesized relationship has a complementary null hypothesis.
○ How strong is the evidence in favour of the null hypothesis?
○ How strong is the evidence in favour of the hypothesized relationship?
Hypothesis testing
A good hypothesis should be unambiguous, otherwise it is difficult to
test and falsify.

A good hypothesis should be general and not too specific, otherwise

we may not have the variation to examine the relationship fully.

• Poorly formulated: “The United States has more murders per

capita because so many people own guns.

• Better: “Countries with less restrictive gun ownership laws will

experience more murders per capita.”
To recap
1. What is a theory?
2. What are the sources of theory?
• Inductive vs. deductive theory building
3. What form can a theory take?
• Structural causes vs. triggering events
• Necessary vs. sufficient conditions
• Causal complexity
• Conjunctural causation
• Multiple causation
• Multiple and conjunctural causation
4. General vs. Specific knowledge
5. Moving from theory to observable implications to hypotheses
Your to do list
• Readings for Wednesday
• Evaluate areas you have questions
• Midterm is Monday June 3:
• June 3, 2:00-3:30PM PT: review session & Q&A with TA Camila
• June 3, 3:30-5PM PT: midterm exam (via Canvas)
• If you are in a time zone outside of North America, please email
Professor Konken if you would like to take the midterm at a more
appropriate time (within 24 hours of it opening on Canvas)