POLS1009 Lecture Notes - Lecture 9: Collective Action, Big Data, Tute
L9 INSTITUTIONAL THEORY: HISTORY, RULES AND NORMS
1. What is an institution?
2. What is “new institutionalism”?
3. What is “punctuated equilibrium” and a “critical juncture?”
A desire for more general explanations
- Political studies moved into political science. Find explanations for political
phenomena.
Scientific method
-Falsification is by far the most important one here and it really what it means is, can the
question I ask be proved wrong.
•Classic example : Did god create the world? – maybe? It’s impossible to prove
-Replication refers to the study being repeatable – that is someone else can go and so
the same study and come up with the same results
•When using qualitative methods, this means will they be broadly the same.
Because qualitative methods are inherently interpretive, its difficult to achieve
the same level of repeatability as you would with statistical methods. (However,
it should be noted that this is a serious and live issue in Quantitative research
today).
-Generalisability – this means, can you make broad claims about the phenomena you
studied that can be applied to other cases generally. This one is often a bigger problem
for qualitiative studies, but its not insurmountable, and most studies can make some
level of general claims. Don’t get too caught up about this.
Behavioural Revolution
- Before the ‘behavioural revolution’ political science (and other social sciences) were
largely focused on describing institutions and studying legal frameworks.
- Researchers increasingly became interested in the ideas of trying to find ‘universal laws’
of human behaviour, just as the sciences had found ‘universal laws’ in relation to physics
etc.
- Moreover, as people developed better statistical tools and got computers it became
possible to examine big political phenomena.
find more resources at oneclass.com
find more resources at oneclass.com
- This was virtually impossible before, because it was simply too hard to know detailed
information about 1000s of cases.
- By looking at a large number of cases (that is, a Large N study), researchers could see
whether or not certain political behaviour applied to most people rather than just in
specific cases.
- Specifically, what regression is aiming to do is to examine the variation on the data –
how far apart are people from each other.
•Where there are clear patterns in the data, we can say that there is some kind of
relationship!
- The effect of this was that there was a general rejection of ‘narratives’ as evidence
•“the numbers don’t lie”
- And there was a general suspicion of iterative approaches to knowing about the world –
of which many qualitative methods are
Causation in Quantitative analysis
An analysis based on a theory about how the world works. This theory does assume some
kind of casual processes
- Uses a large number of observations to make inferences/ searching for statistically
significant relationships
- The way they did this was to come up with a theory (a hunch) about how the world
worked and formulated a hypothesis
- Eg. If parental influence is important for shaping how people vote, then most people
should vote just like their parents
- By setting up this hypothesis, they were in effect setting up a model for how they
thought the world worked.
- In this case, their model took the form of an equation.
- In other parts of political science (game theory, rational choice insititutonalism) they
would just set up a model – like a flow chart or a diagram about how people would likely
behave.
- They then collected a lot of data on a sample of the population you are interested in –
usually in the form of surveys – where they asked about people’s demographic factors
(age, income, education, gender, religion etc) and ask them their opinions on things
(partisan choice etc). These are called variables
find more resources at oneclass.com
find more resources at oneclass.com
- They then used regression to test whether or not there were significant relationships
between any of these variables as their original hypothesis suggested their should be
- The power of multiple varieate (variable) regression was that allowed you to test the
effect of multiple possible independent variables on the dependent variable (the thing
you are trying to explain)
- Today we still use these statistical techniques, and have access to ‘big data’ (often the
whole population!)
- We have also developed far more sophisticated statistical techniques that are much
better at accurately measuring phenomena in the world.
- In fact, today we try to conduct experiments like in a science lab wherever we can.
What about Context? What about the Past?
- Asking in general e.g. why people vote in one way
-
Institutional Response
- By the 1980s, some political scientists were starting to think that the behavioural
approaches could not explain other important aspects of political behaviour
- How institutions interact with individual behaviour.
How can time/temporality matter?
- History and context matter!
- Past decisions often constrain future choices
- Some reforms, programs take a long time:
•Generational Change;
•Cultural change;
•Organisational restructuring;
•Building new political coalitions;
•High-tech programmes, projects and systems
•Complex international negotiations on security/ trade
•Legislative processes
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
L9 institutional theory: history, rules and norms: what is an institution, what is new institutionalism , what is punctuated equilibrium and a critical juncture? . Falsification is by far the most important one here and it really what it means is, can the question i ask be proved wrong. Replication refers to the study being repeatable that is someone else can go and so the same study and come up with the same results: when using qualitative methods, this means will they be broadly the same. Because qualitative methods are inherently interpretive, its difficult to achieve the same level of repeatability as you would with statistical methods. (however, it should be noted that this is a serious and live issue in quantitative research today). Generalisability this means, can you make broad claims about the phenomena you studied that can be applied to other cases generally.
