ED 125 912

AUTHORTITLE

PUB ELTrNOTE

EDRS PRICEDESCRIP-TbPS

TDvNITFTvPS

DOCUMENT RESUME

SE 021 168

Berry, Richard N.NSF University Statistics Analytical Uses forEducational Researcners.Apr 7618p.; Paper presented at the Annual Meeting of theAmerican Educational Pesearch Association (SanFrancisco, California, April 19-23, 1976); Containssome light type

SF-$0.83 BC-$1.67 Plus Postage.Data Eases; *Educational Finance; *EducationalPolicy; Government Role; Graduate Students; Grants;Post Doctoral Education; PublicSupport; Research;*Science Education; Statistics; *Surveys;*Universities*National Science Foundation; NSF

ABSTRACIThis report presents examples of statistics on higher

education which are cddlected by the National Science Foundation(NSF). Datajmire presented from .the four annual university surveysconducted by NSF. These concern (1) graduate student enrollment and

support, (2) academic research and djBvelopment expenditures, (3)

professional and technical employdent, and (4) Federal obligations toinstitutions of higher education. The reporting and processingcharacteristics of NSF surveys are discussed, and the quality of thedata gathered is examined. Data analysis and presentation in the pasthave used descriptive statistics primarily, but future plans call formore analytical treatments. NSF plans to have anlintegratedcomputerized data base in operation by 1976, and to encourageeducational researchers to use the data. Small grants will beavailable to persons conducting studies of educational policy:(SD)

***********************************************************************Documents acquired by ERIC include many informal unpublished

* materials not available from other sources. ERIC makes every effort ** to obtain the best copy available. Nevertheless, items of marginal *

* reproducibility are often encountered and this affects the quality *

* of the microfiche and hardcopy reproductions ERIC makes available *

* via the ERIC Document Reproduction Service (EDRS). EDRS is not* responsible for the quality of the original document. Reproductions *'* supplied by EDRS,are the best that can be made from th original.#********************************************************A*************

-

U S OEPAIRTIAENTOF NEALTSEOUCATtON &WELFARENATIONAL INSTITUTE OF

EDUCATION

DOCUMENT MAS EEEN REPRO

...CO EAAC*Lv AS RECEIVED cROM

TNE PERsoN oRoRlothi

_TING,T po.NTs oR

STATED DO Nor NECESSARILYREPRE-

SENT Or F ICIAL NAT,OHALNSIiisjiE OF

EDUCATION 00517.04 OR POLICY

NSF UNIVETRSITY STATISTICSANALYTICAL USES FOR EDUCATIONAL RESEARCHERS'

by

Richard M. Berry*National Science Foundation

Presented at the session oat: "NSF Statistical

Indicators: Implications for Educational.ReSearchers,"at the Annual Meeting-of the American Educational

Research Association

.San Francisco, CaliforniaApiil 20,-1976

a

*Mr. Berry is Study Director of N5F's Universityand Other Nonprofit Institutions Studies Group:

2

4

Is

1.

I am pleased to behere tqday to talk with you about the NSF University

Statistics Program.-'-As part of my presentation, I will be describing a

new grant program we are corfsidering that would be aimed at educational4

"

wand "other researchers using an integrated data file of four annual surveys.

George Nozicka of Moshman Associates 44-11 talk about the technical aspects

of rho. file, and Ivars Zageris of the American Council on Education will ad-

dress the group on uses of the data at the National Institutes of Health (for

William Gopeland who could not be with us) and by a P,residential Commis --,

sion for Biomedical Research. Finally, we are pleased to have Dr: David

Drew of the RAND Corporation as discussant.,

let us turn to a discussion of examples of the higher education statistics

that NSF collects on an annual basis. I will refer later to these data...

in terms of timeliness, quality, and their relevance to some important

ptilicy issues in higher education. .Generally, the NSF university.statistics

program .consists of four annual surveys ,of graduate student enrollment and

support, academic R&D expenditures, professional and technical employment,

and Federal obligations to institutions of higher education. Together,

they provide a consistent time series of statistical data on various

4 characteristics of University science.,

SURVEY OF GRADUATE STUDENT SUPPORT AND POSTDOCTORALS

. The first survey"I will discuss, the survey of graduate student enrollment,

covers approximately 7,500 departments of science and engineering in

Ph.D. granting institutions. The data elements for fulk-time studentS-.4

are the medlanisms of Support such as research assistantships, level

3

$,

I

4 ,

2.

graduate study, Federal.agency sources of support, non=Federal eources,

and foreign students. Other'data items are part-time enrollment,

nuber of G.I. benefit.recipiepts, and postdoctorals and theft- sources

of support.

03ne verysignificitant characteristic of the survey is that it presents

a profile of the efferts_ol_ new Federal educational policy'and the

4

resulting impact on graduate education. Fat example, Char shows a

Asteady drop in the number of-full-time graduate science students who

were primarily supported by Federal funds. Up until two years ago,

the total number of stu4nts also declined. Many factors,contributed

to the declines, including shrinking, job markets for 11-.D.is'in a

number of scientific occupations, increased costs ofhigher education,

abolition of military draft deferments,,as well as curtailment of

A

Federal grants for fellowships' and traine9ships.

Beginning in 1974, we experiencel(a,two-year increasedn graduate

enrollment, even though Federalsuppo;t Continued its -declint Among

the most' important factors leading to the i ncreaSe were the recessionary,'-'

economy and unfavorable employment_ conditions. Many of the recent

college graduates were unable to find:suitable employMent in tight

job markets and to continue their .education, thd/Itudentsturned to

private sources, college scholarships and their pare?, for instance.

These shifting sources support are.shoWn in the next chart - -now.

institutional and self-support are the main sources for' more than.

Alt two-thirds of. all full-time graduate science' students (Chart

/

4

TR

EN

DS

IN F

UL

L-T

IME

GR

AD

UA

TE

pR

OL

LM

EN

TA

ND

.

FED

ER

AL

1SU

PPO

RT

IN

TH

ESC

IEN

CE

S A

ND

EN

GIN

EE

RIN

G: 1

971

-75

IND

EX

(197

1=10

0)

110

4100 90 70

1971

%

FE

DE

RA

LLY

SU

PP

OR

TE

D

aZS

TU

DE

NT

Sar

ea ft

,

FU

LLT

IME

ST

UD

EN

TS

-197

219

73

SO

UR

CE

:S

RS

/ST

1A,7

1qA

TIO

NA

C S

CIE

NC

EF

OU

ND

AT

ION

-1

1974

.,

1975

(ES

T)

SRV

STIA

-4/1

,, .r

4

)

2-D

IST

RIB

UT

ION

OF

FUL

L -

TIM

E G

RA

DU

AT

E S

CIE

NC

E S

TU

DE

NT

S,

BY

sou

gcE

OF

MA

JOR

SU

PPO

RT

-A

*19

71 A

ND

197

41. 40 30

...."

1971

1974

10

INST

ITU

TIO

NA

LSU

PPO

.RT

SO

UR

CE

: SR

S/S

TIA

, NA

TIO

NA

L S

CIE

NC

E F

OU

ND

AT

ION

. '

OT

HE

R O

UT

S I

bESU

PPO

RT

.

VL

F-SU

PPO

RT

.

1.

SR

OT

IA-4

/1

3.

The survey also produces' nalysis of trends in graduate student

enrollment in various fields of science (Chart 3): The life science

fields have shown the greatest rates of increase in recent years. These

fields of study have growing appeal among young adults in terms of their

occupational aspirations and several years ago, significant increases

were noted in junior year enrollment in the life sciences and health

professions.- *Many of these students were hoping to enter medical

school but only about one of three were successful. 'Among the

unsuccessful ones were those whp enrolled in graduate school in a

related 4eld in the life sciences.

A

SURVEY OF ACADEMIC R&D EXPENDITURES

Turning now to the next survey the academic R&D expenditure survey,.

it obtains data from about 600 colleges and universities that are

petTOrmers of sponsored research. The .expenditures survey covers

separately budgeted research by sources of funds, by type of activity Ir.

such as basic and applied research, and by fields of science.

Instruction and departmental research, and capital expenditures by

field of science are also collected.

Cha serves as an example of the kind of data we get from

acad is R&D expenditure survey. Total R&D expenditures increased

steadily since 1964 in current dollars. By 1975, they grew to almost

three-times their level a decade ago. In constant dollars, however,

they remained at roughly the same level as 1968 and in two periods,

, the rates of inflation exceeded any increases. Except for 1974,

7

%IN

. . . .

3 FULL-TIMF GRADATE i--....TiOlilliiENT BY AREA OF SPENCE:

(INDEX: 1971 :100.0) ,: ikfC,ci.DE.;i0YEARS 1 °71 -75

120.0

115.0

4197°-LIFE' SCIENCES

110.0 / PSYCHOLOGY

SCIENCES/00' . '.....105.0 L

SOCIAL 7,

'''K ot-. ' . .--..-.,.-7....le

0"." . .....

a .. .

...........

....

......//

4.0°

,°°.TO..NTA

L

,i00.0 --. / .. ""%

0,'

...,....

S.

95.0. ..... ..---......,1.-.....

.

90.0PHYSICALANDMATHEMATICAL -SCIENCES .

1971 72 g 73

SOURCE: DIVISION OF SCIENCE RESOURCES STUDIES/STIA

0.1. r8-

ENGINEERING

74 (EST,) 75

12-75

tC)

BIL

LIO

NS

OF

DO

LLA

RS

4.5

J

3.5

F

-3.0 2.0

1.5

1.0 ,.5

.

R &

DE

XP

EN

DIT

UR

ES

OF

UN

IVE

RS

ITIE

S A

ND

CO

LLE

GE

S: F

Y 1

964

-75

I

- O

RM

MO

CU

RR

AN

T D

OLL

AR

S

CO

NS

TA

NT

197

2 2/

'DO

LLA

RS

.00

.0'

woo

1.#

aso

TO

TA

L-R

&D

EX

PE

ND

ITU

RE

S

.00

OM

* O

INIO

MIN

....

......

4111

14,

Wee

eft.

%N

o1.

4111

1, W

O.

"ft-

..r m

oo m

om

0 1964

1 2/ B

AS

ED

ON

GN

P IM

PLI

CIT

PR

ICE

DE

FLA

TO

R.

S -S

OU

RC

E: S

RS

/ST

IA,

NA

TIO

NA

L S

CIE

NC

E F

OU

ND

AT

ION

,

1966

lel l

ee.1

1M.

elpi

a=

PM

0.1

00 W

OO

INN

IS O

W/

*MO

FE

DE

RA

L3.Y

FIN

AN

CE

DR

&D

EX

PE

ND

ITU

RE

S

1968

1970

1972

1473

.\19

7419

75

qtj

(ES

T.)

sRs/

s71,

,!,-.

4/1

1/4.

vf

Federal support also ncrea

The dip in 1974 was due to

1.

funds appropriated to HEW,

4,

ed in each period in current dollars.

impoundmihtby OMB of discretionary R&D

en these impounded funds were released

through court ction, a substanti 1 increase of 12 percent resulted_

in 1975. A real increase o three percent occurred also, even with

spiraling rates of inflation.

The main impactof- the release of impounded funds was felt in the

medical sciences when NIH obligated about $150 million of these dollars.

As a result, medical research in academe went up almost 20 percent

in terms of Federal Support (Charts:

SURVEY OF FEDERAL AGENCY OBLIbAT4ONS TO ACADEMIC INSTITUTIONS

Another way to look at the financial picture is from the standpoint

of Federal agency obligations. The term "ohligationliigused in

Federal accounting and means that a total award is entered on the books

0 at the point of4 the legal commitmenE, The ac uar x 5 ftditures by

the institutions occur later; perhaps, even several years later,

depending on the type of award. In this'survey, 14 Federal agencies

report to NSF on their funding to each educational, institution. The

statistics are published by institution and agency by various categories

of supporty such as R&D and R&D plant, and by field of science.

Because of the lag, the obligations data are important in projecting

year to year changes in Federal support to academic institutions.

For example, look at the slope of the curves in-the next chart (Charts,

1 0.

4

1''

t 4.

sR

&D

EX

PEN

DIT

UR

ES

AT

UN

IVE

RSI

TIE

S A

ND

CO

LL

EG

ES,

,

TO

TA

L,

AL

L F

IEL

DS'

EN

GIN

EE

RIN

G

PHY

SIC

AL

SCIE

NC

ES

EN

VIR

ON

ME

NT

AL

SCIE

NC

ES

MA

TH

EM

AT

ICA

LSc

IE

NC

ES

LIF

E S

CIE

NC

ES

PSY

CH

OL

OG

Y .

:SO

CIA

L S

CIE

NC

ES'

OT

HE

R..S

CIE

NC

ES

n.e.

c.;

BY

FIE

LD

OF

SCIE

NC

E: F

Y 1

974-

75

TO

TA

L 12

IMM

IKW

W\L

NV

FE

DE

RA

L

7

\A

5

10 10'

13 Jr,

161

ISX

.:M&

N1

12M

M.

05

10

(PE

RC

EN

T).

sT)4

14/1

)SO

UR

CE

: SR

S/S

TIA

, NA

TIO

NA

L S

CIE

NC

E F

OU

ND

AT

IPN

-,

I t?

17 17

20

.t

b_T

RE

ND

S_IN

FE

DE

RA

LO

BL

IGA

TIO

NS

TO

UN

VIV

ER

SIT

1ES

AN

D

GO

LL

EG

FS:F

Y19

63-7

6B

ILLI

ON

S O

FD

OLL

AR

S

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0 .5

-

aim

e.g.

AC

AD

EM

IC .S

CI E

NC

OB

LIG

AT

ION

S

0 1963

i,

; .C

aRE

OLL

AR

S

CO

NS

TA

NT

197

2 D

OLL

AR

S(

1i

II

II

II

I.

I.I

i

1964

1965

.1966

1967

1968

1969

197k

1971

1972

1.97

3

F 1

SCA

L: Y

EA

R1

IRO

OM

OM

OW

ei P

RE

LIM

INA

RY

ES

TIM

AT

ES

FO

R 1

975

SO

UR

CE

: SR

S/S

TIA

, NA

TIO

NA

L S

CIE

NC

EF

OU

ND

AT

ION

t t974

1975

5.

from 1972 to 1973 and 1973 to 1974. They go down in the first period

during the impoundment process and jump up in thenext year. They

match very clo,sely the corresponding trend lines that we saw 'in the

expenditure survey charts a year later. Ind using the same analysis,

because of the very small inc"rement.in,the Federal obligations in 1975,

we can-expect a leveling off.of the 1976 academic R&D expenditures,

with actual decreases in real terms. In 1977, however, the R&D budget

is slated for an 11 percent increase, compared with an average 5.5

ercept increase in other federally supported activities.

SURVEY OF EMPLOYMENT OF SCIENTIFIC AND TECHNICAL PERSONNEL

The fourth and-last survey covers scientific and technical personnel.

.The manpower survey consists of about 2,200 colleges and universities

that report information on fields'ot employment, functions in which

primarily.employed (such as research and teaching), hiOlest earned

degree, and sex.

6

The.ftext graph is an example.of theie personnel data and it shows

that 1975,,Amployment of full -time scientists and engineers in higher

education increased nearly three, percent over the 1974 level to a

new high A.Almost.240,000. Part-tim'e employmerit jumped 10 percent,

also. a newhigh, DeSpite fiscal crises, S&E employment continues

to groW, although_at a much slower rate than was reported in the 1960's.

40

These data serve to illustrate the point that the behavior of academic

'stitutions under differing conditions of growth and deceleration

s not symmetrical. As all of you know who aye faced withthe_problem

.of keeping university accounts in the black, contraction is a much

4

sloweand more difficUlt process than expansion.

13.

p

.7 C

HA

NG

E I

N E

MPL

OY

ME

NT

OF

FUL

L-

TIM

E S

CIE

NT

IST

S 'A

ND

EN

GIN

EE

RS

AT

UN

IVE

RSI

TIE

S. A

ND

CO

LL

EG

ES,

BY

FIE

LD

AN

D S

EX

, JA

NU

AR

Y 1

974

-75

TO

TA

L,-

AL

L F

IEL

DS

-E

NG

INE

ER

S

PHY

S IC

AL

ScSCIE

NT

IST

S.

-EN

V I

RO

NM

EN

TA

L

SCIE

NT

IST

S

MA

TH

EM

AT

ICA

L.

SCIE

NT

IST

S

.L

IFE

'SC

IE

NT

I ST

S

-0PS

YC

HO

LO

GI$

TS

. SOC

IA

LSC

IEN

TIS

TS

ME

N

WPA

TW

ASW

iff

Iff

eIF

ER

VIe

r/0A

//

t.

I

iret

rO A

VZ

, Off

4 soM

1.01

:MP7

0

af,0

1400

152

-2-1

01

SO

UR

CE

: $R

S /S

TIA

NA

TIO

NA

L S

CIM

CE

FO

UN

DA

TIO

N

e

tt

_j

1t

t.

23

4

4

..

67

8

(PE

RC

EN

T)

1011

.-1

213

14

SRS/

STIA

-4/1

/

p

tEmployment of women scientists aRe pagineers is increasing at ai,o'st

. . .-

. . / . .. .

triple the pace of men during 1975,'probab1y reflecting theeriphasis4./

6.

placed on affirmative action dealing wit sex descriminagon. One

,,,,

should remcmber, however, that de$eiie recent gains', the Oroportion

scientists anti engineers-comprises only 15 percen,t,Of full-time

employment in elesields.

10 women

academic

REPORTING AND PROCESSING CHARACTERISTICS.

,..-- ,/

,I would like to turn briefly to,a description'of some of the reporting

/ .

,. ..,

A--..2,

and processihg characteristUs of the'NSY'academic surveys in terms./

.-,. I': //

of broader Federal statistics collection issues. First, the NSF data4/,

are timely, the gfeduate studengdata covering the current acadeneic

semeste and the others being within one year of the survey perted.

This contrasts markedly from some other Federal surveys where, in.r

certain cases, delays between collection and publication have been as

long as three to four years.

The most substantial issue we face is quality of the data. Although

our,response rates are excellent approaching 90 to 100 percent in,

mott of the surveysmany of thj quality problems are co nceptual.

Record-keepingopractices of nrvers#ies often do not correspond

to the statistic al

l- l, the idea of accountability is still fairly new to most academic

institutions. We'haveinyeeted considerable resources into the

concepts embodied, in our questionnaires and, after

I

quality problem-7reliability and validity checks in the fiel4

advisory paneIS.ofeerts, national conferences, and plenty ofgalta

editing -but much more remains to be dond beforewe are satisfied.,.,

7.

THE MOIFFrNT TO ANALYSIS

The new, .future emphasis of the NSF academic statistics program

will be placed on analytical treatment of thedata. In the past,

with a few exceptions, most of the analytis has been descriptive.

Some of the exceptions are David'Brenenapr-, who used the graduate

student data to analyze the issue pf shifting quality in enrollment

)

in graduate departments '1/ and their use by the RAND Corporation

in a model to try to measure the impact of Federal support on research

in the biomedical sciences for a Presidential task force,

Of course, the data bases for the four surveys are in computerized

formats for large-scale analytical treatment. The statistics have

great potential of becoming a very valuable resource to researchers'

, both within NSF andZother groups, by providing an interface with the

various commonly-used statistical analysis systems, such as the

Statistical Package for the Social Sciences (SP$S). This sp-called.

"integrated data file," providing interface with the data from-the

four surveys and the statistical analysis systems, is being developed

presen'tly'for NSF by Moshman Associates.

4

Such an interface capability will be powerful, very flexible, and

will permit analysis of many issues. Otherwise, the large size of

the entire data base would cause its use for direct input into

1/ Breneman, David. Graduate School Adjustments to the. Of New DepressionII in Higher Education. February, 1975, National

Academy of Sciences, Technical Repoft No. 3.

16,

statistical analysis programs to be slow ari4 expensive. The integrated

data file will permit issues and research hypotheses to be studied that

L 1

involve only a small fraction of the la%ge7amount of statistics in tyi!

entire file.

NSF is planning to have the integrated system In operation by the end

or fiscal year 1976. The data base will be completed for four survey

years. A Data Base user Manual will describe the integrated system

contents, coding structures, and procedures for data extraction and

analysis by academic researches.!

To encourage analysis heducational researchers of he NSF academic

statistics from the ft surveys, we hope to be abl4 to launch a 116

program of_small grakisfusing data from the-integrated data file.

,

Among the various ,types 0A assistance that we might consider suppo ting. .

/under.the new gr4tr, prOgram are: (1) researckthat is releVaht to some

impOrtant policy jsues in science and higher education; (2) devel pment

"= `of.modelS for 94eistanding the dynamics of higher education; and,

(3) dissertation research by doctoral candidateS on sciencerelate

policy issues :. liW hope that.budgetary funds will become available

support some of

are worketi

these grants next fiscal year, and when the detail

we will be publishing appropriate grant uidelines.

44:--, -

I 4 .L7

to

9.

CONCLUDIWREMIRKS

As you see, the NSF university statistics program, represents many.

7

years and auch effort of compiling extensive information on important

characteristics of academic science. Four surveys of student enroll-.

ment, finances, and-manpower form the computerized data bases. Whent. 4

these files are integrated, a great potential will be created to use

these statistics in analyzing important policy issues in academic

science. Wehope that our plans materialize' for a new grant program

so that the wealth of data may be used.by educational researchers.to

empirically examine these issues. fi

t 1

Or

TO.