how to think. todays puzzle world is complicated lots of potential interrelations between phenomena...
TRANSCRIPT
How To Think
Today’s puzzle
• World is complicated
• Lots of potential interrelations between phenomena
• How do we explain politics?
Our goal
• Arguments– Link between cause and effect– Shows that one thing follows from another– X Y (or more complicated versions)
• Can always draw arrow diagrams
• Also called causal inference
• Some arguments better than others– More truthful, correspond to way world
actually works
Types of Arguments
• Normative Argument– Ethical or moral argument – values– What “should” or “ought” to be
• Positive Argument– About actual state of world– Shows the way things are– Answers question “Why?” or “How?”
Which are positive and normative?
• Incumbents win when the economy is growing• We should reduce the amount of money in
politics• America needs to spend more on foreign aid• Lobbying doesn’t change legislative votes• Educating women will lead to more democracy• An Obama victory is better for America• The US needs to intervene in Syria because the
Syrians have used poison gas.
One key normative argument
• Normal standards of ethics don’t apply to politics• Because politicians are responsible for people’s
lives, they need to weigh expected consequences of their actions
• Can’t act just morally and let chips fall where they may
• To achieve best outcomes may need to compromise, double-deal, lie, etc.
• Ethic of responsibility versus ethic of ultimate ends
Positive argument has two parts
• Dependent variable– Phenomenon we want to explain– Effect
• Independent variable(s)– Phenomenon that explains it– Cause– Systematically affects dependent variable
Types of Positive Arguments
• Deductive – from general theory to specific case– but only as good as your assumptions
• Empirical/Inductive – from specific facts to general theory– search for patterns in the world– but world is complex
Where to start?
• A good question– What causes countries to become
democratic?
• An interesting puzzle– Why is India democratic?– Seems unexpected b/c so poor and so
diverse
Causality = counterfactual
• To say that X causes Y means that if we take away X, we don’t get Y
• “Not X” is the counterfactual
• The problem is that we can’t observe both X and not X at the same time– You either get the drug or you don’t
Gold standard
• Ideally an experiment– Assign subjects randomly to two groups– One groups receives treatment, other doesn’t– The control group is the counterfactual
• But how to do in comparative politics?– Give democracy drug to some undemocratic
countries but not others– Ethical & practical issues
Inductive approach
• Look at countries and see what they have in common and where they differ: search for patterns
• Which countries?• Might start with democracies• Why also look at non-democracies?
• Which independent variables?• Is this correlation enough?
Correlations
• Positive: as one variable rises, second rises
• Negative: as one variable rises, second falls
• Significant: unlikely to be result of random chance
• Insignificant: likely to be result of random chance
Inductivist approach
Country Democracy Wealth Ethnic divisions
Parliamen-tary
US Yes Yes No No
UK Yes Yes No Yes
Belgium Yes Yes Yes Yes
Saudi Arabia
No Yes No No
Inductivist turkey
• Too many potential causes– Which ones are important?
• Never enough cases
• We can’t isolate the counterfactual
Deductive approach
• Start with theory• Assumptions about human nature plus set of
constraints – then logically reason to outcome• What would lead political elites to allow
democracy or citizens to demand it?• If this is true, what implications should we see in
the world?
Usually go together
• Need a theory to prevent inductivist turkey
• Need data to keep theory tied to reality
• Often go back and forth
Not as easy as it looks
• Say we find a correlation, a pattern– Richer countries tend to be democratic
• How do we know if it is a truthful argument?
• Does the pattern indicate genuine causation or is it spurious
Does wealth distinguish democracies and non-democracies?
Pitfalls in causal inference (1)
• Case selection – how did you choose your examples?– Selecting on the dependent variable– i.e., choosing only “successes” or only
“failures”
Pitfalls in causal inference (2)
• Confounding factors, omitted variables– Does another factor cause both X & Y– Correlation is not causation
Obesity and friends
Pitfalls in causal inference (3)
• Endogeneity (reverse causality)• Does Y cause X
• Going to Harvard and getting rich
Pitfalls in causal inference (4)
• Causal mechanism– Can I tell a reasonable story connecting
cause and effect including all the intermediate steps?
– Helps to have evidence for these steps– Correlation between distance from Berlin and
postcommunist democracy/economic reform
Pitfalls in causal inference (5)
• Falsifiability– How would I know if it was false
Conspiracy theories
Pitfalls in causal inference (6)
• Measurement– Have I measured everything well?
Does wealth cause democracy?
• Correlation: Yes, but not perfect• Omitted variables: Possibilities?• Case selection: Middle East? India?
Population? Time period• Endogeneity: democracy => wealth• Causal mechanism• Measurement: How have we defined
democracy? Wealth?
How to read a political science article
• What is the author trying to explain? (dependent variable)
• What is the cause? (independent variable)
• What mechanism connects cause and effect?
How to critique an article
• How were cases selected?
• Omitted variables
• Endogeneity
• Shows all the connections
• Measurement of concepts
• What evidence might disprove
Being a good political thinker
• Are you becoming angry at politics?• Do you have strong opinions before you
look at the evidence?• Do your opinions change as you gather
evidence?• Do you seek info only from sources you
agree with?• Do you think those who disagree with you
are evil?