How to Read a Paper

  1. first, read the introduction and the conclusion
  2. second, scan for tables
    1. normally one or more tables will describe data — average values and so on
    2. one or more tables will report statistical analysis.
      1. those who have had econ 203 will know how to look for “significant” results (typically asterisks, sometimes only standard errors)
      2. in some papers the primary emphasis is on the general nature of the results and not on which coefficients matter most
      3. fancy statistical methods in general do not affect how to read the tables. for example, the Berry-Waldfogel paper reports both standard OLS results and a fancy method-of-moments result and notes they’re basically the same. (so why bother? well … mainly to show you can do it, journal editors demand “best” technique. in sociology of science terms, prove you’re an “insider”.)
  3. read the first paragraph of each subsequent section
    1. a well-written paper will also have final paragraphs that are prose rather than math
    2. subsections may or may not begin with prose … if so, read those, too
  4. read the results and related discussion
    1. you likely have “basic” results
    2. then come extensions, or robustness checks: do our simplifying assumptions matter

Sept 15, 2012

Leave a Reply