Graduate Students and Postdoctorates in Science and Engineering Survey: 2001

Data Collection Practices 

The Survey of Graduate Students and Postdoctorates in Science and Engineering (GSS), conducted each fall by NSF in cooperation with the National Institutes of Health, compiles a factual base for assessing shifts in graduate enrollment and patterns of financial support for students. It provides the only nationally representative database on sources of support of graduate science and engineering (S&E) students and their enrollment characteristics, and on S&E postdoctoral appointees in academic institutions.

From 1965 through 1971 NSF collected data on the support of graduate science students and postdoctoral appointees through institutional applications for NSF traineeships. The graduate student survey was introduced in 1972 to continue the collection of similar data from a larger universe of graduate departments in all S&E doctorate-granting institutions.

Population Size and Structure 

The survey universe has been expanded in three ways since 1972: first, by including all medical schools with graduate programs; second, by broadening the scope of the survey to include additional doctorate-granting institutions and a wider range of academic departments; and third, by including all S&E departments in institutions that grant a master's degree as the highest degree in the sciences and engineering.

Although medical schools with graduate programs are included in the survey population, the student counts exclude M.D., D.V.M., D.D.S., and D.O. candidates unless they are concurrently working for a master's degree or Ph.D. in a science or engineering field or are enrolled in a joint M.D./Ph.D. program. Since many medical school departments/programs included in the survey grant only health-professional degrees, a number of departments are reported as nondegree-granting. These departments may, however, report postdoctoral appointees or other nonfaculty doctoral research staff. Interns and residents primarily involved in patient care are excluded.

The graduate student survey department populations since 1972 are shown in the following table. Counts may differ from those shown in previous user guides, however, because revisions to current and prior year data are processed during each survey cycle. In addition, the survey population has been revised periodically, and departments not considered science, engineering, or health fields were deleted from the survey population.

In 1976 the survey was expanded to include S&E master's-granting institutions; the expansion was initiated as a parallel survey using an abbreviated form. For comparison purposes, selected 1975 data variables were collected for these institutions at the same time. The expanded population (doctorate- and master's-granting) was again surveyed in parallel for 1977.

In 1978 NSF conducted an abbreviated version of the survey, restricted to doctorate-granting institutions. The abbreviated form did not collect data for mechanisms of support for full-time graduate students; sex or citizenship for all graduate students; or any data on postdoctoral appointees. Data for departments for schools not sampled in 1978, which included departments at all master's-granting institutions, were imputed based on data received from those departments in the preceding and following survey years. The survey returned to a full population survey in 1979, with a revised full-scale survey form sent to both doctorate- and master's-granting institutions.

The scope of the survey remained the same until 1984, when NSF converted from a full population survey of 618 graduate institutions to a sample survey of 401 institutions. All S&E doctorate-granting institutions, all land-grant institutions, and all historically black colleges and universities with graduate S&E programs were surveyed; master's-granting institutions (other than historically black colleges and universities and land-grant institutions) were surveyed on a sample basis. The 1985-1987 samples were enlarged by the addition of 18 institutions. Data for departments for schools not sampled for 1984-1987 were imputed based on data received from departments in the preceding and following survey years. The survey returned to a full population survey in 1988.

Survey Populations Since 1972 

• before 1975, data were collected for the doctorate level only
• from 1975 to 1977, master's-level survey data were collected in a parallel survey
• in 1978, data were collected from an abbreviated survey of only doctorate-granting institutions
• from 1979 to 1983, data were collected from a complete survey universe (doctorate- and master's-granting)
• from 1984 to 1987, data were collected based on a combined sample survey (universe for doctorate-granting and sample for master's granting)
• since 1988, data have been collected from a complete survey universe

Data Items 

Prospective data users should note that data items have varied over the years of the survey. Not all variables were collected for both doctorate-granting and master's-granting institutions during the 1975–1978 period. Hence data for those years for doctorate-granting and master's-granting institutions cannot be combined for some variables. In the 1976 survey, for example, data on women part-time students were obtained only from master's-granting institutions. Attributing the sum of this cell to all institutions, doctorate granting as well as master's granting, would be a major error.

The data variables collected over the years have consisted primarily of full- and part-time graduate students and postdoctoral appointees, with detailed information on full-time student sources and mechanisms of major support (for example, fellowships, traineeships, research assistantships, or teaching assistantships), enrollment status, sex, level of study, and citizenship. The collection of full- and part-time student racial/ethnic background data was introduced as an optional item in the 1979 survey and became a standard survey item in 1980. Data on the sex of part-time students has been collected since 1977. In 1994, full- and part-time student racial/ethnic background data separated by sex were requested for the first time.

NSF Discipline Classification Scheme 

Survey data are collected at the departmental level by field of science and engineering. The discipline classification scheme used by NSF is as follows:

Changes to the Survey 

From 1965 to 1971 NSF collected graduate student and postdoctoral appointee data through traineeship applications. These applications collected the following data: number of graduate degrees granted; types and sources of full-time graduate student support; number of students enrolled (full- or part-time) in nondegree programs; number of part-time graduate students enrolled; number of faculty members by function; and number of postdoctoral appointees.

Postdoctoral appointees and other nonfaculty research staff have been included in the Graduate Student Survey since its introduction in 1972. This item has been expanded over the years to include the following data: source and mechanism of support (1973); citizenship (1977); sex (1979); separate counts of postdoctoral appointees and nonfaculty doctoral research staff (1979); and medical professional-degree status (1982).

Faculty variables, including rank and tenure status, appointment, and departure data, were added to the survey as optional items in 1982 and were included as standard items in the 1983 and 1984 surveys. These items were dropped from the survey in 1985.

Here is a summary of the major changes that have occurred as the survey evolved to its present status:

• In contrast to the early doctorate-granting institution survey, source and mechanism data are no longer collected separately for first-time students. Whereas there were once about 50 data cells dealing with first-time students, there have been only 2 since 1980.
  
  
• Fellowships and traineeships have been separately identified for both full-time students and postdoctoral appointees since 1979, whereas previously they were combined.
  
  
• Source of support and citizenship information has been collected for women postdoctoral appointees as well as total postdoctoral appointees since 1979. In the pre-1979 doctorate-granting institution surveys, the total number of women postdoctoral appointees was collected only in 1977 and 1978.
  
  
• Pre-1978 master's-level forms had 11 data cells per department. These earlier forms omitted all information on mechanisms of support, some of the information on women, and all information on first-time students. Several of the source-of-support categories were combined on these forms.
  
  
• The 1978 form for doctorate-granting institutions only, like the early master's forms, contained 11 data cells per department and omitted all information on women and postdoctoral appointees, while including first-time counts and more detailed source of support identification.
  
  
• Beginning in 1979, separate data were requested for postdoctoral appointees and other nonfaculty employees holding doctorates.
  
  
• Questions on racial/ethnic background were first added to the 1979 survey as an optional section and became a standard part of the questionnaire in 1980.
  
  
• Beginning with the 1982 survey, the racial/ethnic breakdown divided the former "foreign, other, or unknown" data cell into two cells: one comprising U.S. citizens of other or unknown background and a second comprising non-U.S. citizens only.
  
  
• The 1982 survey expanded the item on postdoctoral appointees and other nonfaculty doctoral research staff to include individuals holding a medical degree (that is, M.D., D.D.S., D.V.M., D.O.).
  
  
• In 1983 the survey items requesting data on full-time S&E faculty (which formed an optional insert in the 1982 survey) were added to the questionnaire. Faculty data were collected with respect to rank and tenure status, appointments, and departures. Departments that reported no students or postdoctoral appointees (typically clinical medicine fields) were not required to complete these items. The faculty items were dropped from the survey in 1985.
  
  
• Beginning with the 1985 survey, separate data were requested on the number of students supported by the Department of Agriculture. Previously, separate source-of-support data were collected for students supported by NSF, the National Institutes of Health, other subdivisions of the Department of Health and Human Services, the Department of Defense, other Federal funds, institutional funds, foreign funds, other U.S. funds, and personal funds.
  
  
• In fall 1992 the definitions of foreign students and U.S. citizens were modified to match those used by the Department of Education's National Center for Education Statistics and the Institute of International Education. Hence, in previous years permanent residents were counted with foreign students, but in fall 1992, permanent residents were included with U.S. citizens.
  
  
• Beginning in 1994, racial/ethnic data separated by sex were requested for the first time.
  
  
• Beginning with the 1996 survey, separate data were requested on the number of students supported by NASA. Previously, separate source-of-support data were collected for students supported by NSF, the National Institutes of Health, other subdivisions of the Department of Health and Human Services, the Department of Agriculture, the Department of Defense, other Federal funds, institutional funds, foreign funds, other U.S. funds, and personal funds.
  
  
• In fall 1998 a new Web-Based Data Reporting System was made available to school coordinators and departmental respondents which allowed respondents to report their data directly, via the Web at http://www.qrc.com/gssweb. The web survey offers several advantages. Coordinators do not have to collect the individual department's data and then mail it in. Instead, departmental respondents can log onto the Web and enter their data directly. In addition, coordinators can track the progress of the departments using a specially developed monitoring system. Also, entering the data on the Web allows respondents to receive immediate feedback on their responses and make any corrections; this helps ensure more accurate and timely reporting.
  
  
• Beginning with the 1999 survey, separate data were requested on the number of students supported by the Department of Energy. Also in 1999, separate data were collected for first time, full-time students based on their racial/ethnic background for total as well as women students. This expanded the first-time data collected from 2 to 22 cells.
  
  
• Categories were added to the race/ethnicity breakdown for full- and part-time students that expanded the data collected from 54 to 66 cells. The Asian/Pacific Islander category from earlier surveys was divided into Asian American and Native Hawaiian/Pacific Islander categories. The Hispanic category was divided and now includes One Race Only Hispanic/Latino and More Than One Race Hispanic/Latino categories. The Other category is now More Than One Race Non-Hispanic/Latino.
  
  

Data Element Availability by Year 

The lists that follow indicate the valid line and column numbers for each segment of the survey population in each survey year. This information covers only those data cells that appear in the file. Availability is indicated by the appropriate symbol:

      D = doctorate institutions
      M = master's institutions only
      A = all doctorate and master's institutions
      S = sampled master's institutions
      - = not available (in most cases, not collected)

These lists should be used in conjunction with the questionnaire facsimile.

The lists are meant to be guides only. There are numerous ways in which a particular institution might deviate from the normal pattern. There are master's-granting institutions that later became doctorate-granting, master's-granting institutions that at one time were considered doctorate-granting, and institutions that could not respond to certain data cells and for whom imputation was not possible. In all cases, the individual status codes on each data cell should be the primary indicator of cell validity.

Full-Time Graduate Students by Mechanism and Source of Support 

Racial/Ethnic Background by Sex ^*  

First-Year Full-Time Graduate Students by Sex and Racial/Ethnic Background  

Postdoctorates and Nonfaculty Doctoral Research Staff 

Data File Record Structures 

Unlike the other NSF academic surveys, the Graduate Student Survey is conducted on a departmental rather than an institutional basis. Each department is uniquely identified by a 12-digit number that consists of

• a seven-digit School Identification Number (SIN), which consists of a six-digit Federal Interagency Committee on Education (FICE) code plus a one-digit school sequence code identifying a reporting unit within the institution (values of 0-4 usually designate graduate schools, and values of 5-9 usually designate medical schools);
  
  
• a five-digit Departmental Identification Number (DIN), which consists of a three-digit discipline code and a two-digit departmental sequence number; the latter is used to distinguish between two or more departments in the same discipline within a given school. A listing of the discipline codes used is provided later in this section.

Six different record types are used for the data files produced from the 1972 and subsequent Graduate Student Surveys at both doctorate-granting and master's-granting institutions. These six record types are institution header records, school header records, school survey coordinator records, departmental header records, departmental respondent records, and departmental data records. The data records for each year are sequenced in ascending order by SIN, DIN, and record type; the record type is in column 17 of every line of the data file. Please note that the lines and columns listed below are guides to the data file and do not correspond directly to the survey questionnaire.

Institution Header Records 

The institution header record provides basic identifying data for a given institution. One record of this type is present for each institution surveyed; it is the first record for that institution for any given year.

The record format is as follows:

COLUMN	      CONTENTS
1-6	      Institution FICE Code
7-12          '000000'
13-16         Year
17	      Record Type = 1
18-58	      Institution Name
59-64	      Not Used
65	      Institution Highest Degree
		   1 = Doctorate-granting
		       (at least one S&E doctorate department)
		   2 = Master's-granting
66	      Institution Type of Control
	  	   1 = Public
		   2 = Private
68-70	      Not Used
71-72	      Institution State Code (A two-character state abbreviation)
		AL = Alabama		MT = Montana	
		AK = Alaska		NE = Nebraska
		AZ = Arizona	     	NV = Nevada
		AR = Arkansas		NH = New Hampshire
		CA = California		NJ = New Jersey
		CO = Colorado		NM = New Mexico
		CT = Connecticut	NY = New York
		DE = Delaware		NC = North Carolina
		DC = D.C.		ND = North Dakota
		FL = Florida		OH = Ohio
		GA = Georgia		OK = Oklahoma
		GU = Guam		OR = Oregon
		HI = Hawaii		PA = Pennsylvania
		ID = Idaho		PR = Puerto Rico
		IL = Illinois		RI = Rhode Island
		IN = Indiana		SC = South Carolina
		IA = Iowa		SD = South Dakota
		KS = Kansas		TN = Tennessee
		KY = Kentucky		TX = Texas
		LA = Louisiana		UT = Utah
		ME = Maine		VT = Vermont
		MD = Maryland		VI = Virgin Islands
		MA = Massachusetts	VA = Virginia
		MI = Michigan		WA = Washington	
		MN = Minnesota		WV = West Virginia
		MS = Mississippi	WI = Wisconsin
		MO = Missouri		WY = Wyoming
		
73	      Historically Black College Flag
	  	   0 = no
		   1 = yes
74	      Land Grant Flag
		   0 = no
		   1 = yes
75-77         Carnegie Code (1994)
		   R1 - Research Universities I 
		   R2 - Research Universities II 
		   D1 - Doctoral Universities I 
		   D2 - Doctoral Universities II 
		   C1 - Master's (Comprehensive) Universities and Colleges I 
	  	   C2 - Master's (Comprehensive) Universities and Colleges II 
		   LA1 - Baccalaureate (Liberal Arts) Colleges I 
		   LA2 - Baccalaureate (Liberal Arts) Colleges II 
		   2YR - Associate of Arts Colleges
		   ART - Schools of Art, Music, and Design 
		   BUS - Schools of Business and Management 
		   ENG - Schools of Engineering and Technology 
		   HLT - Other Separate Health Profession Schools 
		   LAW - Schools of Law 
		   MED - Medical Schools and Medical Centers 
		   REL - Theological Seminaries, Bible Colleges, and Other 
			   Institutions Offering Degrees in Religion 
		   TEA - Teachers Colleges 
		   TRI - Tribal Colleges and Universities 
		   OTH - Other Specialized Institutions
		   N/A - Not Classified
78-80         Not Used	

School Header Records 

The school header record provides basic identifying data for a given school or reporting unit within an institution. One record of this type is present for each school surveyed in a given institution; it is the first record for that school.

The record format is as follows:

COLUMN	    CONTENTS
1-6	    Institution FICE Code
7           School Sequence Number
8-12        '00000'
13-16       Year
17	    Record Type = 2
18-58       School Name
59-79       Not Used
80	    School Type:
  	        G(Graduate)   = Graduate Schools
	        M(Medical)    = AAMC Member Medical Schools
	        N(Nursing)    = Schools of Nursing
	        O(Osteopathic)= Schools of Osteopathic Medicine
	        P(Public)     = Schools of Public Health
	        D(Dentistry)  = Schools of Dentistry
	        V(Veterinary) = Schools of Veterinary Medicine
	        H(Hospital)   = Hospital-Affiliated Medical Research Centers
	        A(Allied)     = Schools of Professional or Allied Health
	        B(Both)       = Schools With Programs From Both a Graduate 
  	  		            School and an AAMC Member Medical School
   	        U(Unique)     = Other Health Sciences Schools

School Survey Coordinator Records 

Each set of school survey coordinator records provides identifying information on the person who most recently coordinated the collection of questionnaires at a given school. Seven records of this type are provided for each school surveyed. The survey-coordinator line numbers distinguish different records in a set. These records are the second through the eighth records for each school. Their format is as follows:

COLUMN	     CONTENTS
1-6	     Institution FICE Code
7            School Sequence Number
8-12         '00000'
13-16        Year
17	     Record Type = 3
18	     Survey Coordinator Line Number
		   1 = Name
		   2 = Title
		   3 = Address, Institution
		   4 = Address, School or Office
		   5 = Address, Building or Street
		   6 = Address, City, State, and ZIP
		   7 = Telephone in the format "(999)999-9999ext.9999"
		   8 = E-mail Address
19-79	     Survey Coordinator Information
80	     School Type

Departmental Header Records 

The departmental header record provides basic identifying data for a given science or engineering department. Included among the variables are the five-digit Departmental Identification Number (DIN), which consists of a three-digit code for area and discipline, described in the following list, and a two-digit departmental sequence number.

One departmental header record is present for each department in a given school for each year surveyed; it is the first record for that department each year. Its format is as follows:

COLUMN	      CONTENTS
1-6	      Institution FICE Code
7	      School Sequence Number
8-12	      DIN
	      The DIN is composed of the following two parts:
			8-10	Discipline Code
			11-12	Departmental Sequence Number
13-16         Year 
17	      Record Type = 4
18-57	      Department Name
58	      Highest Degree Program Offered by Department
                    (may change from one year to another)
			1 = Doctorate or Ph.D. or equivalent
			2 = Master's
			3 = No Ph.D. or Master's equivalent
			    (usually postdoctorate department)
59-60	      Not Used
61-65	      Parent DIN (used primarily in medical school units)
66-79	      Not Used
80	      School Type


Area and Discipline
Discipline Name
Code

Engineering
	101	Aerospace Engineering
	102	Agricultural Engineering
	103	Biomedical Engineering
	104	Chemical Engineering
	105	Civil Engineering
	106	Electrical Engineering
	107	Engineering Science and Engineering Physics
	108	Industrial/Manufacturing Engineering
	109	Mechanical Engineering
	110	Metallurgical and Materials Engineering
	111	Mining Engineering
	112	Nuclear Engineering
	113	Petroleum Engineering
	114	Engineering, not elsewhere classified (n.e.c.)

Physical Sciences
	201	Astronomy
	202	Chemistry
	203	Physics
	204	Physical Sciences, n.e.c.

Earth, Atmospheric, and Ocean Sciences
	301	Atmospheric Sciences
	302	Geosciences
	303	Ocean Sciences
	304	Earth, Atmospheric,
		and Ocean Sciences, n.e.c.

Computer Sciences
	401	Computer Science

Mathematical Sciences
	402	Mathematics and Applied Mathematics
	403	Statistics

Agricultural Sciences
	501	Agricultural Sciences

Biological Sciences
	601	Anatomy
	602	Biochemistry
	603	Biology
	604	Biometry and Epidemiology
	605	Biophysics
	606	Botany
	607	Cell and Molecular Biology
	608	Ecology
	609	Entomology and Parasitology
	610	Genetics
	611	Microbiology, Immunology, and Virology
	612	Nutrition
	613	Pathology
	614	Pharmacology
	615	Physiology
	616	Zoology
	617	Biosciences, n.e.c.

Health Sciences
	701	Anesthesiology
	702	Cardiology
	703	Oncology/Cancer Research
	704	Endocrinology
	705	Gastroenterology
	706	Hematology
	707	Neurology
	708	Obstetrics and Gynecology
	709	Ophthalmology
	710	Otorhinolaryngology
	711	Pediatrics
	712	Preventive Medicine and Community Health
	713	Psychiatry
	714	Pulmonary Disease
	715	Radiology
	716	Surgery
	717	Clinical Medicine, n.e.c.
	718	Dental Sciences
	719	Nursing
	720	Pharmaceutical Sciences
	721	Veterinary Sciences
	722	Health Related, n.e.c.
	723	Communication Disorders Sciences

Psychology
	801	Psychology, Combined
	802	Psychology, Except Clinical
	803	Clinical Psychology

Social Sciences
	901	Agricultural Economics
	902	Anthropology (Cultural and Social)
	903	Economics
	904	Geography
	905	History and Philosophy of Science
	906	Linguistics
	907	Political Science/Public Administration
	908	Sociology
	909	Sociology /Anthropology
	910	Social Sciences, n.e.c.
	

In the case of certain medical school departments, such as surgery and internal medicine, data are collected at the specialty level rather than at the departmental level. In such cases, a "Parent DIN" is specified on the departmental header record to permit the aggregation of the detailed specialty data into the parent departmental unit.

Departmental Respondent Records 

The departmental respondent record provides identifying information on the person, if known, who last prepared the survey questionnaire. This record is the second record for each department each year. Its format is as follows:

COLUMN	   CONTENTS
1-6	   Institution FICE Code
7	   School Sequence Number
8-12	   DIN
13-16	   Year
17	   Record Type = 5
18	   Departmental Respondent Line Number
		1 = Name
		2 = Telephone in the format "(999)999-9999ext.9999"
		3 = E-mail Address
19-79	   Departmental Respondent Information
80	   School Type

Departmental Data Records 

Departmental data records contain the numerical responses to the questionnaire items. The data records are identified by line numbers, similar to the line numbers on the questionnaire. Since most questionnaire lines can have up to 13 columns of data, provision is made on the data record for 13 columns of responses. A data record is present for each data line on the questionnaire to which a nonzero response has been received. For line 6 (full-time graduate S&E enrollment), however, a data record is present regardless of whether a zero or nonzero response has been received. An unused numeric field will contain a right-justified single zero.

Data for 1972-2000 are maintained in the line and column format of the 2001 questionnaire to facilitate inter-year comparisons. Associated with each data column is a Data Status Code, which indicates the source and/or special treatment of the data. The format of the data records that include the 2001 survey is as follows:

COLUMN	   CONTENTS
1-6	   Institution FICE code
7	   School Sequence Number
8-12	   DIN
13-16	   Year
17	   Record Type = 6
18-19	   Line Number (1-18) 
	      (Not necessarily the same as on the questionnaire)
20-23	   Data Column 1
24	   Status Code for Data Column 1 (Blank = Normal Response)
		N = Data Not Available*
		E = Estimated
		I = Imputed for Nonresponse
25-28	   Data Column 2
29	   Status Code for Data Column 2
30-33	   Data Column 3
34	   Status Code for Data Column 3
35-38	   Data Column 4
39	   Status Code for Data Column 4
40-43	   Data Column 5
44	   Status Code for Data Column 5
45-48	   Data Column 6
49	   Status Code for Data Column 6
50-53	   Data Column 7
54	   Status Code for Data Column 7
55-58	   Data Column 8
59	   Status Code for Data Column 8
60-63	   Data Column 9
64	   Status Code for Data Column 9
65-68	   Data Column 10
69	   Status Code for Data Column 10
70-73	   Data Column 11
74	   Status Code for Data Column 11
75-78	   Data Column 12
79	   Status Code for Data Column 12
80-83      Data Column 13
84	   Status Code for Data Column 13
85	   School Type

Data Availability by Year 

Graduate student data are available annually for the years 1972 through 2001.

Data Imputation 

Data imputation is an automated procedure that estimates data for totally and partially nonrespondent institutions. The imputation logic involves the determination of inflator/deflator factors for a series of key variables, including total full-time students, total part-time students, total postdoctoral appointees, and total nonfaculty doctoral researchers. These factors are calculated on the basis of fully responding departments in both the current and previous years' surveys for each combination of degree level and discipline. For nonrespondent schools, these key variables are estimated by applying the appropriate inflator/deflator to the previous year's data values for that school. The key variables are then distributed among the various subtotal and detail fields by using the same relative percentages that were last reported by that institution.

Imputation rates for all survey data cells are calculated annually for various combinations of institutions and departments as determined by discipline, highest degree granted, and level of department. At the grand-total level (that is, all departments at all graduate institutions), the 2001 imputation rates for the key survey variables ranged from 0.6 percent for total full-time students to 1.8 percent for total part-time students.

The Data Status Code indicates whether the data for each item were provided by the institution (blank), were not available (N), were estimated by NSF (E), or were imputed by computer (I).

Confidentiality 

Data from the Graduate Student Survey are not confidential.

Questionnaire Facsimiles 

Copies of survey questionnaires for fall 1992 through fall 2001 are available. Data users requiring questionnaires for years in which different data elements were collected (fall 1989, and fall 1981 and earlier) can request copies.

In addition to reviewing the instructions and definitions provided in the questionnaire, data analysts are advised that during each survey cycle, institutions are given the opportunity to revise data that they had provided in prior years. The most recent versions of the data for all years should be used in historical analyses.

National Science Foundation, Division of Science Resources Statistics
Survey of Graduate Students and Postdoctorates in Science and Engineering 2001
Guide to the Data Files
Arlington, VA [June 2003]