Professional Documents
Culture Documents
Giacomo Pasini
Outline
1 introduction
2 course overview
3 Designing research
introduction course overview Designing research
A world of Data
Source: EU
introduction course overview Designing research
Source: VoxEU
introduction course overview Designing research
Source: SWG
introduction course overview Designing research
Outline
1 introduction
2 course overview
3 Designing research
introduction course overview Designing research
This class
Why coding
Coding is increasigly inportant, for quantitative research but also for many other
scopes
The book provides examples side by side in R, Stata and Python
You (should) know R, excellent way to learn other two languages
1/3 of my classes will be with Stata, the TA will do Python
introduction course overview Designing research
It’s easy to get data to tell us what happened, but not why. “correlation does not
equal causality”
Economists have been looking at causality for longer than most other fields. We’re
good at it!
Causal inference is often necessary to link data to models and actually learn how
the world works
We’ll be taking a special approach to causal inference, one that lets us avoid
complex mathematical models
introduction course overview Designing research
Prerequisites
Courses I think it is useful you passed or at least you know the content.
Math 1 & 2, Stats 1 & 2, Intro to coding, Economic policy, Political Economy
You can follow the course even if you did not pass them
Anyway, it’s a third year course, I will assume you know something about them...
Propedeuticity
Courses you mast have passed to sit the exam of Intro to Econometrics
Long story short: for you, NONE
introduction course overview Designing research
Grading
Two assignments (deadlines in moodle), mostly from the book’s website, in Stata
and Python
Exam. half coding, half pen & paper
final grade: exam + up to 2 points for (correctly) solved assignments
Tutor will be available to help out with coding issue in solving assignments
introduction course overview Designing research
Lecture overview
1 introduction
2 course overview
3 Designing research
Outline
1 introduction
2 course overview
3 Designing research
introduction course overview Designing research
Research Question
A research question is a question that you have that you plan to answer, or at least
try to answer, by doing research
A good research question is well-defined, answerable, and understandable
Example: Does adding an additional highway lane reduce traffic?
introduction course overview Designing research
Empricial Research
Problem with quantitative empirical research: the numbers that we observe often
don’t tell us exactly what we want to know.
we might want to study the impact of additional lanes by comparing two-lane
highways to three-lane highways.
...But we aren’t actually interested in how much traffic there is on three-lane
highways and on two-lane highways. We’re interested in whether we can make
traffic go down by turning a two-lane highway into a three-lane highway!
introduction course overview Designing research
Problem with quantitative empirical research: the numbers that we observe often
don’t tell us exactly what we want to know.
we might want to study the impact of additional lanes by comparing two-lane
highways to three-lane highways.
...But we aren’t actually interested in how much traffic there is on three-lane
highways and on two-lane highways. We’re interested in whether we can make
traffic go down by turning a two-lane highway into a three-lane highway!
The numbers we have don’t tell us that right away. All we have are two-lane
highways and three-lane highways.
We don’t have a what if highway that tells us how much traffic there would have
been if we’d made that two-lane highway one lane wider.
Quite often then you have to carefully design the analysis you want to do with
numbers in order to answer your question
introduction course overview Designing research
a research question is a question that can be answered, and for which having that
answer will improve your understanding of how the world works.
A question that can be answered: it’s possible for there to be some set of
evidence in the world that, if you found that evidence, your question would have a
believable answer.
Questions that are not research questions:
Who is the best football player alive?
Is Coca-Cola better than Pepsi or not?
What is the best James Bond movie?
Questions that are are research questions:
Which football player alive scored more goals?
Did Coca-Cola sold more or less than Pepsi in 2021?
Which James Bond movie had the highest ticket sales worldwide?
introduction course overview Designing research
So, do those questions improve your understanding of how the world works?
I.e., does answering those questions tell us something broader than just the answer
to the question?
Another way to see it, is that the answer to that question tells us something about
some underlining theory
A Theory is an explanation of why something happens.
introduction course overview Designing research
So, do those questions improve your understanding of how the world works?
I.e., does answering those questions tell us something broader than just the answer
to the question?
Another way to see it, is that the answer to that question tells us something about
some underlining theory
A Theory is an explanation of why something happens.
A good research question takes us from theory to hypothesis, where a hypothesis is
a specific statement about what we will observe in the world
Then, the answer to the research question should tell us whether the theory is likely
to be right or wrong.
If we can stick to the original theory easily even if the result is unexpected, then the
research question (or the theory, or both!) is not a good one.
introduction course overview Designing research
Why bother about research questions? Most of you will not do research in her/his
life.
why not skip the hard part of deriving a research question from a theory and
instead just see what sorts of patterns are in the data?
This is called data mining. It is good in finding patterns, but not in finding causal
relations.
Data mining delivers correlations, but correlation may have little to do with why
some variables are moving together
introduction course overview Designing research
someone using data mining to try to understand ice cream sales may well notice
that the proportion of people who wear shorts is a fantastic predictor of ice cream
sales
But shorts-wearing isn’t why people buy ice cream. They buy ice cream and wear
shorts because it’s hot
shorts can be a great way of predicting ice cream eating....
However, if what we’re really interested in isn’t predicting ice cream but explaining
why people eat ice cream, it’s tempting at that point to try to invent a story (make
up a theory) to justify why shorts might actually be the reason people eat ice cream.
In the case of ice cream and shorts we can tell that’s ridiculous, but it’s a lot harder
when we don’t actually know what’s ridiculous and what’s not ahead of time.
introduction course overview Designing research
Consider Potential Results. Then, imagine what kind of sense you’d make of that
result, or what conclusion you would draw. Can you really say something about the
theory it is related to
Consider Feasibility. Is the right data available? If answering your research
question is possible but requires following millions of people repeatedly for decades,
then it’s not a good research question
Consider Scale. What kind of resources and time can you dedicate to answering
the research question? Some things can be done with a large lab and funding, for
an Intro to Econometrics home assignment better to choose something less complex
introduction course overview Designing research
Consider Design. research question can be great on its own, but it can only be so
interesting without an answer. Figuring out whether given a research question you
do have a reasonable research design is the topic of the rest of course.
Keep It Simple! A common mistake is to bundle a bunch of research questions
into one. “What are the determinants of inequality?”. There are many determinants
of inequality. You’re unlikely to answer that question well. Instead try “Is wealth of
the parents a determinant of inequality?”