Research Advice

The purpose of this page is to provide some useful advice to students working on a research paper. The main audience are students in my Law and Economics classes, though graduate students may find it useful as well. It is based on my own workflow when writing a paper.


Literature Review

The goal of research is to add to known knowledge, and hence we need to find out what is known. This is not so easy since there is so much published material out there. Here I outline my workflow when starting in a new area:

  1. I always begin with "web of science", the core collection. This is available by going to the library website and searching for web of science, and clicking on the electronic access. Please do this BEFORE a google search. You can begin by choosing your database (I use the social science one mainly, but also medline for medical research). There are a number of reasons for this. Articles that are published in recognized journals are indexed, and the site provides links to articles cited by the article. Once an article that is relevant to ones research is found, then you can use the links to see the earlier literature. The subsequent papers that cite the current paper are also linked. This is also a useful measure of the importance of the paper - more highly cited papers are typically better and/or better written.

  2. At the top left corner of web of science this is a link to a pdf of the paper. Sometimes the link does not work, but then you can go to jstor or the journal site to find the article.

  3. Finally, one needs to record the article and make notes. Web of Science has an online link to endnote for this purpose that I often use. However, on my computer I use jabref (www.jabref.org). This program is not perfect, but it the best I have found. I put ALL my references here. From here you can insert them easily into your document. You can add notes and put references in groups. The reason for using a single file is that I never have to enter a reference twice, which reduces errors in bibliographies.

  4. After working with web of science, one should have an idea of who are the major contributors to a field. At that point I will do a google search to find recent papers. The reason for not starting with google is very bad papers can be cited on the web - I find searches in web of science yield higher quality papers (and if you click the journal in web of science it will provide a measure of quality.

Writing

I have a number of linux computers for day to day work, as well as a macbook. For writing I use exclusively lyx. It uses latex as the backend, and it is the easiest program in which to write mathematics. I also use latex and libreoffice, depending upon co-authors. This is combined with jabref mentioned above.

Analysis

For statistical analysis these days I use R exclusively, though coauthors use stata, which all good. I find the syntax of stata unorthodox, while R is more like a computer language. R is unfortunately very slow for numerical work, so currently I recommend julia for numerical work. Both R and julia have good support for machine learning, though python is the standard for deep learning these days.