====================================================
KEYWORDS FOR DATASET: Income, Education Level, Twins
====================================================

===============================================================
ACCOMPANYING DATA PROVIDED BY: Guido Imbens, PhD 
                               UCLA, Department of Economics
===============================================================

================================
GENERAL EXPLANATION OF THE STUDY
================================
The original study that used these data sought to answer what seems
a simple question: By how much will another year of schooling most likely 
raise one's income? Attempts had been made to estimate the value of a year's 
education in previous studies, but previous estimates may have been 
imprecise for two reasons. The first, most obvious reason is the difficulty 
of extracting education's effect on income from the effect that other 
variables related to education have on income. That is, a worker's natural 
ability, his family background, and his innate intelligence are all possible 
confounding factors that must be controlled for to estimate the effect of 
education on income accurately. Thus this study interviewed twins, 
collecting information about education, income, and background. Because
monozygotic twins (twins from a single egg) are genetically identical and 
have similar family backgrounds, they provide an excellent control for 
confounding variables. 

The second difficulty in measuring the effect of income on education has
to do with the false reporting of education levels, and this study is the 
first to address it. Since people are more likely to report a higher 
education level than they have actually attained, especially in face to 
face interviews, the data will contain a number of people with lower 
education levels in the higher education categories. Thus, since education 
usually increases income, estimates for the precise amount of this 
increase will be too low. To correct for this bias the researchers 
interviewed the twins separately and recorded two entries for each 
individual's education level: his self-reported education level and the
education level reported by his twin. This allowed them to estimate the 
"measurement error" of reported education levels and correct for it. The 
result was a much higher estimate of the effect a year of education is 
likely to have on one's income. In fact, this study's estimates were 
higher than those of all previous studies, which did not correct for 
measurement error in education level.  

The full citation for the published study is below. It is also available 
on the web at JSTOR, located at "www.jstor.org".  

Ashenfelter, Orley and Krueger, Alan.  "Estimates of the Economic Return 
	to Schooling from a New Sample of Twins."  The American Economic 
	Review 84.5 (Dec. 1994) 1157-1173.
	
=============================
BRIEF DESCRIPTION OF THE DATA          
=============================
The data were collected by a team of five interviewers at the 16th Annual 
Twins Day Festival in Twinsburg, Ohio, in August 1991. A booth was set up 
at the festival's main entrance, and an ad inviting all adult twins to 
participate in the survey was placed in the festival program. In addition, 
the interviews roamed the festival grounds, approaching all adult twins for 
an interview, and almost every pair of twins accepted. 

In total, 495 individuals over the age of 18 were interviewed. 

========================
HOW TO USE THE DATA FILE
========================
The data file is comma delimited text. Each row contains information on a 
pair of twins for sixteen variables, which are explained below. 
Note that the each individual in a pair of twins was randomly assigned a 
number: twin 1 or twin 2.

Missing data are indicated by a period.

NOTE: For most data analysis purposes, the logarithm of the hourly wage is 
used instead of the hourly wage itself. 

DLHRWAGE......the difference (twin 1 minus twin 2) in the logarithm of 
              hourly wage, given in dollars.                                              
DEDUC1........the difference (twin 1 minus twin 2) in self-reported 
	      education, given in years.                                        
AGE...........Age in years of twin 1.                                                 
AGESQ.........AGE squared. 
HRWAGEH.......Hourly wage of twin 2. 
WHITEH........1 if twin 2 is white, 0 otherwise.
MALEH.........1 if twin 2 is male, 0 otherwise.
EDUCH.........Self-reported education (in years) of twin 2.
HRWAGEL.......Hourly wage of twin 1.
WHITEL........1 if twin 1 is white, 0 otherwise.
MALEL.........1 if twin 1 is male, 0 otherwise.
EDUCL.........Self-reported education (in years) of twin 1.
DEDUC2........the difference (twin 1 minus twin 2) in cross-reported 
              education. Twin 1's cross-reported education, for example,
              is the number of years of schooling completed by twin 1 as
              reported by twin 2.                                     
DTEN..........the difference (twin 1 minus twin 2) in tenure, or number of 
	      years at current job.
DMARRIED......the difference (twin 1 minus twin 2) in marital status, where
              1 signifies "married" and 0 signifies "unmarried". 
DUNCOV........the difference (twin 1 minus twin 2) in union coverage, where
              1 signifies "covered" and 0 "uncovered". 
                                                        
================================================
STATISTICAL TESTS AND ANALYSES USED IN THE STUDY
================================================
1) one-sample t-test
2) two-sample t-test
3) Linear Regression
4) Ordinary Least Squares
5) Generalized Least Squares
6) Maximum Liklihood Estimation
7) Method of Moments Estimation