Charles Jones on writing a thesis(zz)

On Writing a Thesis:

Keep a journal of ideas: whenever you have a research idea, jot it down. I
still keep one of these (online as a web page) and find it to be very helpful.


Reading papers when you are formulating a topic is important but dangerous.
DO NOT just casually read papers that are closely related to your topic. The
danger is that you will have a half-baked idea, read a paper superficially,
and decide your topic has already been fully researched.

Instead, when you read a paper close to your topic, you must read it
thoroughly: take pencil and paper in hand, derive every equation and
theoretical result. Get the data the authors are using and reproduce every
empirical result. Researchers always want to make their results looks as good
as possible, so they tend to hide or shade the big problems with their
research. If you only read a paper casually, you will miss these, but if you
derive every result yourself, the problems will jump out at you. I find this
can sometimes be a useful way to generate a topic: if you are struggling,
pick a couple of key papers related to your interests and read them this way.

Of course, remember that you do not want to write a paper that is "A Small
Variation on X's Work." However, sometimes you can find big holes that others
haven't seen by understanding X's work very well.

Finally, I personally like papers with the following structure: Here is a
stylized fact that everyone will accept as roughly true. Here is something
puzzling or interesting that one gets from this fact. Here is a model that we
can use to try to understand that fact. Here is a simulation of the model
that reproduces the fact. This is a good outline for many (but of course not
all!) good macro papers.

Read more!