My research topics and methods are pretty diverse.  These matrices show the links between them.  Blank cells mean "no" or "none."
Matrix 1: Topics, Citations, & Approach
Cites outside Brand New
Papers Micro (excluding Relates to Law Topic (T) or Mainstream
(in order of competion) Role of Theory Cites to Labor statistics) & Economics Application (A) or Crazy
Alcoa S M, Me Y M
Voting I S P M
Union Wages D Y M M
Race / Cesarean Y H A M
Implicit Contracts D Y M M
Information & Sorting T M, H Y T A,T,P
Bicycle Helmets H Y A M
Malpractice / Cesarean D M, H Y M
Source of Payment Y H M
Grades as Information D Y S T T
Physician Discipline D H Y M
Union Voting S Y P A
Zero Tolerance T Y H Y M
Multiproduct Pricing D M A
Optimal Penalty Structure S H Y T A,P
Good Economy T M T M
forthcoming D=discursive M=macro M=mainstream
working paper T=new theoretical development P=political science T=crazy topic
S=utilize structural equation H=health (clinical or policy) A=crazy approach to topic
that I derived from theory S=sociology P=prospective,
Me=metals industry not retrospective *
* prospective means deliberately designing a method that can be used to assess policy before implementing it
Matrix 2: Empirical Methods and Findings
Fix Others'
Papers Data Type of Mistakes or 2-D Non-standard
(in order of competion) Size of Effect Type Estimation Misinterpretation Testing * Estimation
Alcoa S TS S Y Y nonlinear
Voting I S CS S nonlinear
Union Wages S P D Y method of moments
Race / Cesarean S P D
Implicit Contracts S P H Y method of moments
Information & Sorting S CS H
Bicycle Helmets L P D
Malpractice / Cesarean S P D Y
Source of Payment S P D Y
Grades as Information S CS H don't know its name
Physician Discipline L P none
Union Voting N CS S Y semiparametric
Zero Tolerance N P H Y Y quantile regression
Multiproduct Pricing L P H Y principal components
Optimal Penalty Structure N/A CS none
Good Economy N/A TS D
S=small (~5%) TS=time series D=descriptive (linear specification just for simplicity)
L=large CS=cross section S=structural (explicitly derived from theory)
N=none P=panel H=heuristic (simplifcation of more elaborate theory)
N/A=not applicable
* 2-D testing means testing implications on two completely different dependent variables or the equivalent
There are a few patterns.  First, my citations tend to be hybrid--many key themes run across fields.  Second, I am increasingly
moving toward non-standard topics or methodological approaches, and less bound by tradition.  Third, most of my research is
ultimately rooted, or strongly influenced by, the labor and law & economics literatures, and generally relies on panel data.