ED 469 645 AUTHOR TITLE INSTITUTION REPORT NO PUB DATE NOTE AVAILABLE FROM PUB TYPE EDRS PRICE DESCRIPTORS IDENTIFIERS ABSTRACT DOCUMENT RESUME HE 035 362 Siegel, Peter H.; Whitmore, Roy W.; Johnson, Ruby E.; Yu, Di National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000), CATI Nonresponse Bias Analysis Report. Working Paper Series. National Center for Education Statistics (ED), Washington, DC. NCES-WP-2002-03 2002-03-00 59p.; Andrew G. Malizio, Project Officer. For full text: http://www.nces.ed.gov/pubsearch/ . Reports Evaluative (142) EDRS Price MF01/PC03 Plus Postage. Data Collection; Higher Education; National Surveys; *Responses; *Statistical Bias; Telephone Surveys *National Postsecondary Student Aid Study; *Nonresponse Bias; Weighting (Statistical) Unit nonresponse causes bias in survey estimates when the outcomes of respondents and nonrespondents are different. In the National Postsecondary Student Aid Study of 1999-2000 (NPSAS:2000) there were three levels of response, one of which was computer-assisted telephone interview (CATI) response. Because the response rates were less than 70% in some sectors or overall, an analysis was conducted to determine if CATI estimates were significantly biased due to CATI nonresponse. Through other databases, considerable information was available about CATI nonrespondents to this survey, and these data were used to analyze and reduce bias. The distribution of several variables using the design-based, adjusted weights for study respondents were found to be biased before CATI nonresponse adjustments. The CATI nonresponse and poststratification procedures, however, reduced the bias for these variables, and when the weighting was completed, no variables available for most respondent and nonrespondents had significant bias for all students combined. The bias was significantly reduced, and the remaining bias is small. Section 2 discusses the characterization of the bias before CATI nonresponse adjustment, and section 3 describes the weight adjustments used to reduce bias. Section 4 describes the bias for CATI variables, and section 5 discusses the bias remaining after weight adjustments. Section 6 discusses the overall predictive ability of the three nonresponse models, and section 7 presents conclusions. (SLD) Reproductions supplied by EDRS are the best that can be made from the original document.
Transcript
ED 469 645
AUTHOR
TITLE
INSTITUTION
REPORT NOPUB DATENOTE
AVAILABLE FROMPUB TYPEEDRS PRICE
DESCRIPTORS
IDENTIFIERS
ABSTRACT
DOCUMENT RESUME
HE 035 362
Siegel, Peter H.; Whitmore, Roy W.; Johnson, Ruby E.; Yu, DiNational Postsecondary Student Aid Study 1999-2000(NPSAS:2000), CATI Nonresponse Bias Analysis Report. WorkingPaper Series.
National Center for Education Statistics (ED), Washington,DC.
NCES-WP-2002-032002-03-0059p.; Andrew G. Malizio, Project Officer.For full text: http://www.nces.ed.gov/pubsearch/ .
Reports Evaluative (142)
EDRS Price MF01/PC03 Plus Postage.Data Collection; Higher Education; National Surveys;*Responses; *Statistical Bias; Telephone Surveys*National Postsecondary Student Aid Study; *Nonresponse Bias;Weighting (Statistical)
Unit nonresponse causes bias in survey estimates when theoutcomes of respondents and nonrespondents are different. In the NationalPostsecondary Student Aid Study of 1999-2000 (NPSAS:2000) there were threelevels of response, one of which was computer-assisted telephone interview(CATI) response. Because the response rates were less than 70% in somesectors or overall, an analysis was conducted to determine if CATI estimateswere significantly biased due to CATI nonresponse. Through other databases,considerable information was available about CATI nonrespondents to thissurvey, and these data were used to analyze and reduce bias. The distributionof several variables using the design-based, adjusted weights for studyrespondents were found to be biased before CATI nonresponse adjustments. TheCATI nonresponse and poststratification procedures, however, reduced the biasfor these variables, and when the weighting was completed, no variablesavailable for most respondent and nonrespondents had significant bias for allstudents combined. The bias was significantly reduced, and the remaining biasis small. Section 2 discusses the characterization of the bias before CATInonresponse adjustment, and section 3 describes the weight adjustments usedto reduce bias. Section 4 describes the bias for CATI variables, and section5 discusses the bias remaining after weight adjustments. Section 6 discussesthe overall predictive ability of the three nonresponse models, and section 7presents conclusions. (SLD)
Reproductions supplied by EDRS are the best that can be madefrom the original document.
NATIONAL CENTER FOR EDUCATION STATISTICS
U.S. DEPARTMENT OF EDUCATIONOffice of Educational Research and Improvement
EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)
is document has been reproduced asreceived from the person or organizationoriginating it.
Minor changes have been made toimprove reproduction quality.
Points of view or opinions stated in thisdocument do not necessarily representofficial OERI position or policy.
Working Paper Series
National Postsecondary Student Aid Study 1999-2000(NPSAS:2000), CATI Nonresponse Bias Analysis Report
U. S. Department of EducationOffice of Educational Research and Improvement
2
EST COPY AVAILABLE
NATIONAL CENTER FOR EDUCATION STATISTICS
Working Paper Series
The Working Paper Series was initiated to promote the sharing of thevaluable work experience and knowledge reflected in these preliminaryreports. These reports are viewed as works in progress, and have notundergone a rigorous review for consistency with NCES StatisticalStandards prior to inclusion in the Working Paper Series.
U. S. Department of EducationOffice of Educational Research and Improvement
3
U.S. Department of EducationRod PaigeSecretary
Office of Educational Research and ImprovementGrover J. WhitehurstAssistant Secretary
National Center for Education StatisticsGary W. PhillipsDeputy Commissioner
The National Center for Education Statistics (NCES) is the primary federal entity for collecting, analyzing,and reporting data related to education in the United States and other nations. It fulfills a congressionalmandate to collect, collate, analyze, and report full and complete statistics on the condition of education inthe United States; conduct and publish reports and specialized analyses of the meaning and significance ofsuch statistics; assist state and local education agencies in improving their statistical systems; and reviewand report on education activities in foreign countries.
NCES activities are designed to address high priority education data needs; provide consistent, reliable,complete, and accurate indicators of education status and trends; and report timely, useful, and high qualitydata to the U.S. Department of Education, the Congress, the states, other education policymakers,practitioners, data users, and the general public.
We strive to make our products available in a variety of formats and in language that is appropriate to avariety of audiences. You, as our customer, are the best judge of our success in communicating informationeffectively. If you have any comments or suggestions about this or any other NCES product or report, wewould like to hear from you. Please direct your comments to:
National Center for Education StatisticsOffice of Educational Research and ImprovementU.S. Department of Education1990 K Street NWWashington, DC 20006
March 2002
The NCES World Wide Web Home Page ishttp://nces.edrov
Suggested Citation
U.S. Department of Education, National Center for Education Statistics. National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000), CATI Nonresponse Bias Analysis Report, NCES 2002-03, by Peter H. Siegel, Roy W.Whitmore, Ruby E. Johnson, and Di Yu. Andrew G. Malizio, project officer. Washington, DC: 2000.
4
Foreword
In addition to official NCES publications, NCES staff and individuals commissioned byNCES produce preliminary research reports that include analyses of survey results, andpresentations of technical, methodological, and statistical evaluation issues.
The Working Paper Series was initiated to promote the sharing of the valuable workexperience and knowledge reflected in these preliminary reports. These reports are viewed asworks in progress, and have not undergone a rigorous review for consistency with NCESStatistical Standards prior to inclusion in the Working Paper Series.
Copies of Working Papers can be downloaded as pdf files from the NCES ElectronicCatalog (http://nces.ed.gov/pubsearch/), or contact Sheilah Jupiter at (202) 502-7444,e-mail: sheilah [email protected], or mail: U.S. Department of Education, Office of EducationalResearch and Improvement, National Center for Education Statistics, 1990 K Street NW, Room9048, Washington, DC 20006.
Marilyn M. SeastromChief Mathematical StatisticianStatistical Standards Program
5
Ralph LeeMathematical StatisticianStatistical Standards Program
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000), CATINonresponse Bias Analysis Report
Prepared by:
Peter H. SiegelRoy W. WhitmoreRuby E. Johnson
Di Yu
Prepared for:
U.S. Department of EducationOffice of Educational Research and Improvement
National Center for Education Statistics
March 2002
6
Acknowledgments
The authors gratefully acknowledge the assistance of staff members of the National Center forEducation Statistics (NCES) and the Office of Education Research and Improvement (OEM) fortheir advice, guidance, and review in conducting the analyses and in preparing this document.We are particularly grateful to C. Dennis Carroll, Associate Commissioner, PostsecondaryStudies Division, Paula R. Knepper, Senior Technical Advisor, Andrew G. Malizio, ProjectOfficer for NPSAS and Program Director for Postsecondary Longitudinal Studies and SampleSurveys.
Particular thanks are also extended to the project staff members of the principal contractor,Research Triangle Institute (RTI), including Dr. James Chromy, Dr. Avinash Singh, Dr. PaulBiemer, and Dr. John Riccobono for their guidance in conducting these analyses and preparingthis document. We are especially indebted to Ms. Pat Parker, Ms. Brenda Gurley, and Ms. LilClark, who prepared the graphics, integrated the text, and prepared the drafts and final version ofthis report.
7
Table of ContentsAcknowledgements V
1 Introduction 1
2 Bias Before CATI Nonresponse Adjustment 2
3 Weight Adjustments 20
4 Bias for CATI Variables 23
5 Bias After Weight Adjustments 25
6 ROC Curve 32
7 Conclusions 32
8
List of Tablesr
Table 1. Comparison of NPSAS:2000 CATI respondents and nonrespondents for allstudents 4
Table 2. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for all students 7
Table 3. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for students sampled as baccalaureate recipients.10
Table 4. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for undergraduate students 13
Table 5. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for graduates/first-professional students 16
Table 6. Summary of significant nonresponse bias before CATI nonresponse adjustment bystudent type 19
Table 7. Variables used in final NPSAS:2000 CATI nonresponse models 22
Table 8. Nonresponse bias for CATI variables for all students 24
Table 9. Summary of significant nonresponse bias after weight adjustments by student type 31
xii
List of FiguresFigure 1. Nonresponse bias before CATI nonresponse adjustment and after weight
adjustments for selected variables for all students 27
Figure 2. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for students sampled as baccalaureaterecipients 28
Figure 3. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for undergraduate students 29
Figure 4. Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for graduates/first professional students 30
Figure 5. ROC curve for overall response propensity 32
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
1. Introduction
Unit nonresponse causes bias in survey estimates when the outcomes of respondents andnonrespondents are different. For NPSAS:2000, there were three levels of response: institutionresponse defined as the institution providing an enrollment list for sampling, computer-assisteddata entry (CADE) response, and computer-assisted telephone interview (CATI) response. ACATI respondent was defined as any sample member who completed at least Section A of theCATI interview, an abbreviated interview, or paper-copy of the interview.
CADE:Additionally, a CADE respondent was defined as any sample member for whom the
financial aid gate question was answered, ANDenrollment section had some enrollment data provided, ANDstudent characteristics section had at least one valid response for the set of items:date-of-birth; marital status; race; and sex. If the case matched to the Departmentof Education's Central Processing System (CPS), it was considered to havesuccessfully met this criterion.
A study respondent was defined as any sample member who was either a CATI respondent, aCADE respondent, or both.
The following weighted response rates were obtained:
Because the response rates were less than 70 percent in some sectors or overall, ananalysis was conducted to determine if CATI estimates were significantly biased due to CATInonresponse. For NPSAS:2000, data were collected not only from students using CATI andfrom institutions using CADE but also from databases such as the Department of Education'sfinancial aid Central Processing System and National Student Loan Data System (NSLDS)..Therefore, considerable information was known for CATI nonrespondents and these data wereused to analyze and reduce the bias. The distributions of several variables using the design-based, adjusted weights for study respondents (study weights) were found to be biased beforeCATI nonresponse adjustments. The CATI nonresponse and poststratification procedures,however, reduced the bias for these variables. When the weighting was completed, no variablesavailable for most respondents and nonrespondents had significant bias for all studentscombined. The bias was significantly reduced, and the remaining bias is small. Section 2discusses the characterization of bias before CATI nonresponse adjustment, section 3 describesthe weight adjustments used to reduce bias, section 4 describes the bias for CATI variables,
11
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
section 5 discusses the bias remaining after weight adjustments, section 6 assesses the overallpredictive ability of the three nonresponse models, and section 7 presents conclusions.
2. Bias Before CATI Non response Adjustment
CATI respondents and nonrespondents were characterized by comparing the weighted'percentage of CATI respondents with the weighted percentage of CATI nonrespondents for eachcategory of important characteristics known for both respondents and nonrespondents. T-testswere performed to determine if the difference between respondents and nonrespondents wassignificant at the five percent level.
Table 1 compares demographic characteristics of CATI respondents and nonrespondentsfor all students combined and also shows the full sample distribution. This table shows that thedistributions of many student demographic characteristics, such as age, race, ethnicity, sex,student type, fall enrollment status, and receipt of aid are significantly different for CATIrespondents and nonrespondents. Some institution characteristics, such as level, control, andregion, are also are significantly different for CATI respondents and nonrespondents. Some ofthe statistically significant differences are not large differences, but aid recipients are clearlymore likely to be respondents. When the differences between CATI respondents andnonrespondents are significant, the bias is also significant, as described below. Note that manyof the variables in this table are derived from multiple sources that could influence the results ifadditional information obtained in CATI could be the reason for a difference betweenrespondents and nonrespondents. Footnotes to table 1 indicate the primary data sources.
The nonresponse bias was estimated for variables known for both respondents andnonrespondents. The bias in an estimated mean based on CATI respondents, .3-1R, is the
difference between this mean and the target parameter, i.e., the mean that would be estimatedif a complete census of the target population was conducted. This bias can be expressed asfollows:
B(57R)=37,. -ir.
The estimated mean based on CATI nonrespondents, YNR , can be computed if data for
the particular variable for most of the nonrespondents is available. The estimation of it is asfollows:
it = (1 77) -.TR ± /15NR
where T1 is the weighted unit nonresponse rate. Therefore, the bias can be estimated as follows:
E(YR)= .1112-k
or equivalently
' The study weights and imputed data were used.
212
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
h(T,R)=77(.TR-T,NR)
This formula shows that the estimate of the nonresponse bias is the difference between the meanfor CATI respondents and nonrespondents multiplied by the weighted nonresponse rate. Thevariance of the bias was then computed using Taylor Series estimation in RTI's softwarepackage SUDAAN.
Tables 2, 3, 4, and 5 show the nonresponse bias before and after weight adjustments forselected variables for all students, baccalaureate recipients, all undergraduate students, andgraduate/first-professional students, respectively. The first set of columns in tables 2 through 5shows the estimated bias before CATI nonresponse adjustment and imputation for the variablesavailable for most responding and nonresponding students. The respondent and nonrespondentcounts and means do not match those in table 1 because table 1 included imputed data and tables2 through 5 did not include imputed data for the before CATI nonresponse adjustment estimates.Also, no categories for missing data were included in tables 2 through 5. A few variables haveno before-adjustment results because they had high levels of missing data. T-tests were used totest each level of the variables for significance of the bias at the 0.05/(c-1) significance level,where c is the number of categories within the primary variable. Below and in table 6 aresummaries of the before-adjustment significant bias across the four tables:
at least one level of most of the variables is biased for at least one student type
Pell grant amount categories are biased only for all students combined andStafford loan categories are biased only for undergraduate students
two variables are biased for two student types; five variables are biased for threestudent types; and twelve variables are biased for all four student types
Pell grant amount and Stafford loan amount are not biased for any of the studenttypes
20 variables are biased for all students combined; 17 variables are biased forbaccalaureate recipients, 18 variables are biased for undergraduate students, and14 variables are biased for graduate/first-professional students
significant biases are usually small and sometimes are due to small sample sizes.
Weighting adjustments reduced bias to the extent possible as described in sections 3 and 5.
13
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Table I.-Comparison of NPSAS:2000 CATI respondents and nonrespondents for all students
Variable
CATI respondents CATI nonrespondents Full sample
Sample sizePercentestimate' Sample size
Percentestimate'
Samplesize
Percentestimate'
Agee
19 or younger 6,480 19.5 2,560 19.0 9,030 19.3
20 to 23 16,140 31.2 6,290 32.2 22,420 31.5
24 to 29 9,380 19.3 4,140 21.8* 13,510 20.1
30 to 39 6,910 16.1 2,540 14.9* 9,440 15.8
40 or older 5,600 13.9 1,760 12.1* 7,360 13.4
Race3
White 4,980 77.7 12,840 74.2* 47,820 76.7
Black or African American 4,960 12.1 2,290 13.5 7,250 12.5
Asian 2,540 5.3 1,540 8.6* 4,080 6.3
American Indian or Alaska 280 0.7 180 1.2* 460 0.9Native
Native Hawaiian or other Pacific 140 0.4 150 1.0* 290 0.5Islander
Number of phone numbers obtained5 150 0.3 860 4.7* 1,010 1.6
0 21,080 52.4 7,960 50.1* 29,030 51.7
1 13,810 29.2 4,770 26.4* 18,580 28.4
2 9,460 18.1 3,690 18.8 13,150 18.3
3 or moreReceipt of any aid3 18,240 48.4 8,320 56.5* 26,560 50.8
No 26,250 51.6 8,950 43.5* 35,200 49.3
Yes
Receipt of federal aid3 24,140 60.4 10,320 66.9* 34,460 62.3
No 20,350 39.6 6,960 33.1* 27,300 37.7
Yes
Receipt of state aid3 37,920 85.2 15,230 87.8* 53,140 85.9
No 6,580 14.8 2,050 12.2* 8,630 14.1
YesReceipt of institution aid3 34,040 82.8 14,070 86.8* 48,110 84.0
No 10,450 17.2 3,210 13.2* 13,660 16.0
YesApplied for federal aid6
No 21,000 51.9 9,270 59.1* 30,270 54.0
Yes 23,500 48.2 8,010 40.9* 31,500 46.0
See footnotes at end of table.
5
15
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Table I.-Comparison of NPSAS:2000 CATI respondents and nonrespondents for all students-Continued
VariableCATI respondents CATI nonrespondents Full sample
Sample sizePercent
estimate' Sample sizePercent
estimate'Sample
sizePercent
estimate'Receipt of Pell grant
No 34,760 79.9 13,460 81.7* 48,220 80.4
Yes 9,730 20.1 3,820 18.3* 13,550 19.6
Pell grant amount receivedLess than or equal to $1,183 2,480 29.5 910 28.9 3,390 29.3
$1,184 to $1,953 2,400 23.2 1,020 24.5 3,420 23.6
Greater than $1,953 4,860 47.3 1,880 46.6 6,740 47.1
Receipt of Stafford loadNo 28,310 70.5 12,050 76.3* 40,360 72.2
Yes 16,180 29.5 5,230 23.7* 21,410 27.8
Stafford loan amount receivedUndergraduate students
Less than or equal to $2,625 3,710 32.7 1,340 33.1 5,060 32.8
$2,626 to $4,425 3,000 22.4 1,020 23.2 4,020 22.6
$4,426 to $5,500 3,860 22.2 1,080 20.0* 4,940 21.7
Greater than $5,500 3,080 22.8 1,060 23.7 4,140 23.0
Graduate/first-professionalstudents
Less than or equal to $8,000 640 23.4 190 23.4 830 23.4
$8,001 to $12,521 620 23.3 180 23.7 800 23.4
$12,522 to $18,500 950 39.9 260 37.5 1,210 39.4
Greater than $18,500 320 13.4 110 15.5 430 13.9
'Using the final study weights and imputed data.
2Primary data sources are CADE and CPS.
3Primary data source is CADE.
4Primary data source is sampling frame.
5Primary data source is CATI control system. The CATI respondents with "zero phone numbers obtained" had called-in to the telephonecenter to complete the interview, or completed a self-administered paper version.
6Primary data source is CPS.
7Primary data source is NSLDS.
*Difference between CATI respondents and nonrespondents is significant at the 0.05/(c-1) level, where c is the number of categorieswithin the primary variable.
NOTE: Some percentages may not sum to 100 percent for a variable due to rounding. To protect confidentiality of the data some numbershave been rounded.
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study, 1999-2000 (NPSAS:2000).
6
16
Tab
le 2
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
all s
tude
nts
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nde
nt m
ean,
stud
yw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
tsE
stim
ated
bia
s
Stu
dent
's a
ge44
,430
17,0
0027
.427
.00.
1140
*27
.327
.20.
0319
Stu
dent
age
gro
ups
19 o
r yo
unge
r6,
470
2,51
019
.518
.90.
2000
19.4
19.3
0.06
5020
to 2
316
,120
6,16
031
.232
.0-0
.200
031
.331
.5-0
.147
0
24 to
29
9,36
04,
100
19.3
22.0
-0.8
000*
20.1
20.1
0.02
60
30 to
39
6,89
02,
500
16.1
14.9
0.40
00*
15.6
15.8
-0.1
820
40 o
r ol
der
5,59
01,
730
13.9
12.2
0.50
00*
13.6
13.4
0.23
70
Has
stu
dent
rec
eive
d an
y ty
pe o
f aid
?Y
es26
,250
8,95
051
.643
.52.
3000
*49
.349
.30.
0060
No
18,2
408,
320
48.4
56.5
-2.3
000*
50.8
50.8
-0.0
060
Did
stu
dent
atte
nd in
stitu
tion
in th
e fa
ll?Y
es, f
ull t
ime
27,6
108,
640
53.7
42.0
3.30
00*
50.4
50.5
-0.0
740
Yes
, hal
f tim
e5,
670
2,72
015
.818
.8-0
.800
016
.616
.7-0
.056
0Y
es, l
ess
than
hal
f tim
e4,
000
1,90
012
.216
.0-1
.100
0*13
.313
.3-0
.029
0N
o7,
020
3,52
018
.323
.2-1
.400
0*19
.719
.50.
1590
Atte
ndan
ceF
ull t
ime
$$
$$
$36
.937
.4-0
.472
0'H
alf t
ime
$$
$$
$16
.516
.50.
0050
Less
than
hal
f tim
e$
$$
$$
21.1
21.3
-0.2
740
Mix
ed$
$$
$$
25.5
24.8
0.74
10*
Citi
zens
hip
stat
usU
.S. c
itize
n39
,660
14,5
5093
.090
.30.
8000
92.2
92.1
0.08
60R
esid
ent
1,68
088
04.
45.
1-0
.200
04.
64.
6-0
.012
0V
isa
1,49
01,
100
2.6
4.6
-0.6
000*
3.2
3.3
-0.0
740
CP
S m
atch
Yes
23,5
008,
010
48.2
40.9
2.10
00*
46.1
46.0
0.05
60N
o21
,000
9,27
051
.959
.1-2
.100
0*53
.954
.0-0
.056
0
Dep
ende
ncy
stat
us -
two-
leve
lD
epen
dent
$$
$$
$44
.342
.81.
5170
°
Inde
pend
ent
$$
$$
$55
.757
.2-1
.517
0*
See
foot
note
s at
end
of t
able
.
Tab
le 2
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
all
stud
ents
-Con
tinue
d
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nde
nt m
ean,
stud
yw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
tsE
stim
ated
bia
s
Dep
ende
ncy
stat
us -
thre
e-le
vel
Dep
ende
nt$
$$
$I
44.3
42.8
1.51
7eIn
depe
nden
t w/o
ut d
epen
dent
s$
$$
I$
27.2
29.4
-2.2
180*
Inde
pend
ent w
/dep
ende
nts
$$
:$
$28
.527
.80.
7010
*
Enr
ollm
ent t
otal
at t
he s
tude
nt's
inst
itutio
n44
,490
17,2
8016
423.
517
296.
3-2
53.1
520*
1667
3.9
1667
6.7
-2.7
413
enro
llmen
t cat
egor
ies'
Enr
ollm
ent<
=3,
267
10,6
904,
250
17.2
15.3
0.50
00*
16.6
16.6
-0.0
530
3,26
7<en
rollm
ent<
=11
,096
11,5
704,
180
28.1
26.6
0.50
0027
.927
.70.
1890
11,0
96<
enro
llmen
t<24
,120
11,0
604,
490
28.8
30.4
-0.4
600
29.1
29.3
-0.1
320
24,1
20<
=en
rollm
ent
11,1
704,
350
25.9
27.8
-0.5
300*
26.5
26.5
-0.0
040
Was
the
stud
ent e
nrol
led
in in
stitu
tion
in th
efa
ll?Y
es, a
t a N
PSA
S in
stitu
tion
36,4
1013
,520
79.7
76.2
1.02
70*
78.6
78.7
-0.1
110
Yes
, not
at a
NPS
AS
inst
itutio
n1,
060
240
2.1
1.1
0.28
20*
1.8
1.8
-0.0
480
No
7,02
03,
520
18.2
22.7
-1.3
100*
19.7
19.5
0.15
90
Did
the
stud
ent r
ecei
ve a
ny f
eder
al f
inan
cial
aid?
Yes
20,3
506,
960
39.6
33.1
1.89
30*
37.8
37.7
0.02
80
No
24,1
4010
,320
60.4
66.9
-1.8
930*
62.2
62.3
-0.0
280
Stud
ent's
sex
Mal
e17
,870
7,75
042
.246
.9-1
.398
0*43
.543
.6-0
.031
0Fe
mal
e25
,780
9,42
057
.853
.11.
3980
*56
.556
.40.
0310
Did
the
stud
ent r
ecei
ve a
nyY
es10
,450
3,21
017
.213
.21.
1610
*16
.016
.00.
0200
Inst
itutio
n fi
nanc
ial a
id?
No
34,0
4014
,070
82.8
86.8
-1.1
610*
84.0
84.0
-0.0
200
Inst
itutio
n re
gion
New
Eng
land
2,54
01,
040
5.2
5.4
-0.0
520
5.3
5.2
0.04
70M
id E
ast
7,33
02,
730
15.2
14.3
0.26
1014
.914
.9-0
.003
0G
reat
Lak
es7,
360
2,64
015
.814
.70.
2900
15.7
15.5
0.25
00Pl
ains
3,52
01,
150
7.2
6.0
0.35
00*
7.0
6.9
0.15
90So
uthe
ast
10,0
103,
440
23.0
19.4
1.03
00*
22.1
21.9
0.10
80So
uthw
est
4,65
02,
140
11.1
13.7
-0.7
500*
11.9
11.9
0.04
10R
ocky
Mou
ntai
n1,
850
610
3.9
3.7
0.06
003.
93.
90.
0040
Far
Wes
t6,
440
3,08
017
.421
.1-1
.070
0*17
.818
.5-0
.626
0*O
utly
ing
area
800
460
1.3
1.7
-0.1
100
1.5
1.4
0.01
90
See
foot
note
s at
end
of
tabl
e.
Tab
le 2
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
all s
tude
nts-
Con
tinue
d
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nde
nt m
ean,
stud
yw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
tsE
stim
ated
bia
s
Did
the
stud
ent r
ecei
ve a
ny P
ell g
rant
s?Y
es9,
730
3,82
020
.118
.30.
5400
*19
.619
.60.
0000
No
34,7
6013
,460
79.9
81.7
-0.5
400*
80.4
80.4
0.00
00
Pell
cate
gori
es f
or a
ll Pe
ll re
cipi
ents
Pell
amou
nt <
= $
1,18
32,
480
910
29.5
28.9
0.15
0029
.529
.30.
1880
$1,1
83 <
Pel
l am
ount
<=
$1,
953
2,40
01,
020
23.2
24.5
-0.3
400
23.2
23.6
-0.3
300
$1,9
53 <
Pel
l am
ount
4,86
01,
880
47.3
46.6
0.19
00*
47.2
47.1
0.14
10W
hat w
as th
e am
ount
of
the
Pell
gran
tre
ceiv
ed?
9,73
03,
820
1911
.219
09.3
0.50
9819
10.7
1910
.70.
0000
Inst
itutio
n se
ctor
Publ
ic le
ss-t
han-
2-ye
ar74
032
00.
60.
60.
0000
0.6
0.6
0.00
00Pu
blic
2-y
ear
5,95
02,
980
37.6
43.8
-1.8
000*
39.4
39.4
0.00
00Pu
blic
4-y
ear
non-
doct
orat
e-gr
antin
g6,
730
2,23
012
.710
.40.
6800
*12
.012
.00.
0000
Publ
ic 4
-yea
r do
ctor
ate-
gran
ting
14,6
405,
090
25.0
22.4
0.75
00*
24.3
24.3
0.00
00Pr
ivat
e no
t-fo
r-pr
ofit
2-ye
ar o
r le
ss98
053
00.
70.
8-0
.040
00.
70.
70.
0000
Priv
ate
not-
for-
prof
it 4-
year
non
-do
ctor
ate-
gran
ting
5,41
01,
780
9.4
8.2
0.36
00*
9.1
9.1
0.00
00
Priv
ate
not-
for-
prof
it 4-
year
doc
tora
te-
gran
ting
6,15
02,
260
9.5
8.7
0.24
009.
39.
30.
0000
Priv
ate
for-
prof
it le
ss-t
han-
2-ye
ar2,
350
1,29
01.
62.
0-0
.100
01.
71.
70.
0000
Priv
ate
for-
prof
it 2-
year
780
390
1.6
1.7
-0.0
300
1.7
1.7
0.00
00Pr
ivat
e fo
r-pr
ofit
4-ye
ar76
041
01.
21.
4-0
.060
01.
31.
30.
0000
Stud
ent's
mar
ital s
tatu
sSi
ngle
$$
$$
:73
.074
.0-1
.001
0*I
Mar
ried
$$
$$
$25
.724
.61.
0590
*Se
para
ted
$$
$$
$1.
31.
4-0
.058
0
Staf
ford
cat
egor
ies
for
all
UG
and
Sta
ffor
d am
ount
<=
$2,
625
3,71
01,
340
27.8
28.7
-0.2
200
28.2
28.0
0.19
70
See
foot
note
s at
end
of
tabl
e.
Tab
le 2
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
all s
tude
nts-
Con
tinue
d
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI n
onre
spon
se a
djus
tmen
t-un
impu
ted
data
Afte
r w
eigh
t adj
ustm
ents
-impu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
Staf
ford
rec
ipie
nts4
Am
ount
of
Staf
ford
loan
rec
eive
dD
id th
e st
uden
t rec
eive
a S
taff
ord
loan
?
Did
the
stud
ent r
ecei
ve a
ny s
tate
fin
anci
al a
id?
Stud
ent t
ype
sam
pled
Stud
ent t
ype
- C
AD
E
UG
and
$2,
625
< S
taff
ord
amou
nt <
=$4
,425
UG
and
$4,
425
< S
taff
ord
amou
nt <
=$5
,500
UG
and
$5,
500
< S
taff
ord
amou
ntG
R a
nd S
taff
ord
amou
nt <
= $
8,00
0G
R a
nd $
8,00
0< S
taff
ord
amou
nt <
=$1
2,52
1G
R a
nd $
12,5
21 <
Sta
ffor
d am
ount
<=
$18,
500
GR
and
$18
,500
< S
taff
ord
amou
nt
Yes
No
Yes
No
Bac
cala
urea
te r
ecip
ient
Oth
er u
nder
grad
uate
stu
dent
Gra
duat
e st
uden
tFi
rst-
prof
essi
onal
stu
dent
Und
ergr
adua
te s
tude
ntG
radu
ate
stud
ent
Firs
t-pr
ofes
sion
al s
tude
nt
3,00
0
3,86
0
3,08
064
062
0
950
320
16,1
8016
,180
28,3
10
6,58
037
,920
11,3
40
24,6
207,
610
920
35,5
408,
040
920
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nde
nt m
ean,
stud
yw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
tsE
stim
ated
bia
s1,
020
19.0
20.1
-0.2
700
19.1
19.3
-0.2
630
1,08
018
.917
.40.
3800
18.8
18.5
0.29
70
1,06
019
.420
.6-0
.300
019
.619
.7-0
.050
019
03.
53.
10.
0900
3.3
3.4
-0.1
320
180
3.5
3.1
0.08
003.
33.
4-0
.111
0
260
5.9
5.0
0.24
005.
75.
70.
0330
110
2.0
2.0
-0.0
100
2.0
2.0
0.03
00
5,23
060
14.3
5839
.643
.147
359
90.5
5971
.219
.286
15,
230
29.5
23.7
1.69
00*
27.7
27.8
-0.0
890
12,0
5070
.576
.3-1
.690
0*72
.372
.20.
0890
2,05
014
.812
.20.
7500
*14
.114
.10.
0180
15,2
3085
.287
.8-0
.750
0*85
.985
.9-0
.018
0
3,70
06.
95.
70.
3400
*6.
46.
5-0
.151
0 *2
10,8
9078
.883
.3-1
.300
0*80
.280
.10.
0830
2,40
012
.49.
50.
8300
*11
.711
.60.
1120
280
1.9
1.5
0.12
00*
1.7
1.8
-0.0
430
14,4
0085
.288
.5-0
.970
0*86
.286
.20.
0000
2,60
013
.010
.10.
8400
*12
.212
.20.
0000
280
1.8
1.4
0.14
00*
1.7
1.7
0.00
00
*Bia
s is
sig
nifi
cant
at t
he 0
.051
(c-1
) le
vel,
whe
re c
is th
e nu
mbe
r of
cat
egor
ies
with
in th
e pr
imar
y va
riab
le.
:Suf
fici
ent d
ata
from
oth
er n
on-C
AT
I so
urce
s w
ere
not a
vaila
ble
prio
r to
impu
tatio
n.
'The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed o
n th
e st
udy
wei
ghts
at t
he 0
.05
leve
l, an
d th
ere
wer
e no
t suf
fici
ent d
ata
avai
labl
e fr
om o
ther
non
-C
AT
I so
urce
s to
incl
ude
the
vari
able
in th
e no
nres
pons
e m
odel
s.
2The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l. Sa
mpl
ed s
tude
nt ty
pe w
as n
ot in
clud
ed in
the
nonr
espo
nse
mod
els
beca
use
it is
not
an
actu
al s
tude
nt c
hara
cter
istic
and
may
not
ref
lect
true
stu
dent
type
.
3Enr
ollm
ent c
ateg
orie
s w
ere
defi
ned
by q
uart
iles
base
d on
tota
l enr
ollm
ent f
or th
e 19
97-1
998
scho
olye
ar.
*UG
= u
nder
grad
uate
stu
dent
, GR
= g
radu
ate
stud
ent,
and
FP =
fir
st-p
rofe
ssio
nal s
tude
nt.
NO
TE
: Est
imat
ed b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t is
the
diff
eren
ce b
etw
een
the
mea
n fo
r C
AT
I re
spon
dent
s an
d no
nres
pond
ents
mul
tiplie
d by
the
wei
ghte
d no
nres
pons
e ra
te. A
fter
wei
ght a
djus
tmen
ts, e
stim
ated
bia
s is
the
diff
eren
ce b
etw
een
mea
ns b
ased
on
the
CA
TI
wei
ghts
and
the
stud
y w
eigh
ts. T
o pr
otec
t con
fide
ntia
lity
of th
e da
ta, s
ome n
umbe
rs h
ave
been
rou
nded
.
SOU
RC
E: U
.S. D
epar
tmen
t of
Edu
catio
n, N
atio
nal C
ente
r fo
r E
duca
tion
Stat
istic
s, N
atio
nal P
osts
econ
dary
Stu
dent
Aid
Stu
dy, 1
999-
2000
(N
PSA
S:20
00).
Tab
le 3
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
stud
ents
sam
pled
as
bacc
alau
reat
e re
cipi
ents
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
ts
data
Est
imat
edbi
as
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
non
resp
onde
ntm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Stud
ent's
age
11,3
40
nnon
resp
onde
nts
3,65
025
.925
.10.
1850
*25
.825
.70.
0800
*St
uden
t age
gro
ups
19 o
r yo
unge
r40
200.
40.
6-0
.100
00.
40.
4-0
.058
0120
to 2
36,
920
2,15
060
.258
.50.
4000
60.2
59.8
0.33
4024
to 2
92,
320
970
20.8
26.8
-1.5
000*
21.7
22.2
-0.5
210
30 to
39
1,15
032
010
.39.
00.
3000
9.8
10.0
-0.1
890
40 o
r ol
der
920
180
8.4
5.1
0.80
00*
8.0
7.6
0.43
30*
Has
stu
dent
rec
eive
d an
y ty
pe o
f ai
d?Y
es7,
260
2,09
063
.256
.51.
7000
*61
.161
.5-0
.394
0N
o4,
080
1,61
036
.843
.5-1
.700
0*38
.938
.50.
3940
Did
stu
dent
atte
nd in
stitu
tion
in th
e fa
ll?Y
es, f
ull t
ime
8,72
02,
490
76.4
69.3
1.80
00*
73.8
74.6
-0.8
200
Yes
, hal
f tim
e1,
090
470
10.0
13.0
-0.7
000*
11.1
10.8
0.29
30Y
es, l
ess
than
hal
f tim
e45
018
04.
04.
9-0
.200
04.
44.
30.
1440
No
1,06
047
09.
612
.9-0
.800
0*10
.710
.40.
3830
Atte
ndan
ceFu
ll tim
e$
$$
$$
49.8
50.7
-0.8
3402
Hal
f tim
e$
11
:1
11.8
11.1
0.73
30*
Les
s th
an h
alf
time
$$
$$
$7.
07.
00.
0370
Mix
ed$
$$
$$
31.4
31.3
0.06
30
Citi
zens
hip
stat
usU
.S. c
itize
n10
,550
3,23
094
.489
.81.
2000
*93
.893
.30.
5630
*1R
esid
ent
320
130
3.4
4.2
-0.2
000
3.4
3.6
-0.1
220
Vis
a21
023
02.
26.
0-1
.000
0*2.
73.
2-0
.440
0*
CPS
mat
chY
es6,
400
1,78
055
.348
.51.
7000
*53
.353
.6-0
.267
0N
o4,
940
1,92
044
.751
.5-1
.700
0*46
.746
.40.
2670
Dep
ende
ncy
stat
ustw
o-le
vel
Dep
ende
nt$
$$
$$
55.3
53.5
1.78
20*2
Inde
pend
ent
$$
$$
$44
.746
.5-1
.782
0*
Dep
ende
ncy
stat
us -
thre
e-le
vel
Dep
ende
nt$
$$
$$
55.3
53.5
1.78
20*2
Inde
pend
ent w
/out
dep
ende
nts
$$
$$
$27
.428
.7-1
.295
0*In
depe
nden
t w/d
epen
dent
s$
$$
$$
17.3
17.8
-0.4
880
Enr
ollm
ent t
otal
at t
he s
tude
nt's
inst
itutio
n11
,340
3,70
016
883.
018
442.
3-3
94.6
140*
1715
7.3
1727
7.6
-120
.322
7
Enr
ollm
ent c
ateg
orie
s3E
nrol
lmen
t<=
3,26
71,
960
520
16.8
12.9
1.00
00*
16.0
15.8
0.21
203,
267<
enro
llmen
t<=
11,0
963,
320
980
27.8
25.0
0.70
0027
.727
.10.
5720
11,0
96<
enro
llmen
t<24
,120
2,85
01,
040
25.7
29.0
-0.8
410*
25.9
26.5
-0.6
300
24,1
20<
enro
llmen
t3,
210
1,15
029
.833
.1-0
.846
0*30
.430
.6-0
.154
0
Was
the
stud
ent e
nrol
led
in in
stitu
tion
in th
e fa
ll?Y
es, a
t a N
PSA
S in
stitu
tion
10,2
103,
220
90.0
87.2
0.72
60*
88.9
89.3
-0.3
710
Yes
, not
at a
NPS
AS
inst
itutio
n80
100.
40.
20.
0390
0.3
0.3
-0.0
120
No
1,06
047
09.
612
.6-0
.765
0*10
.710
.40.
3830
Did
the
stud
ent r
ecei
ve a
ny f
eder
al f
inan
cial
aid
?Y
es5,
800
1,66
050
.645
.91.
1890
*49
.049
.4-0
.450
0N
o5,
550
2,04
049
.454
.1-1
.189
0*51
.150
.60.
4500
Stud
ent's
sex
Mal
e4,
290
1,61
040
.645
.6-1
.269
0*41
.641
.8-0
.245
0Fe
mal
e6,
920
2,08
059
.454
.41.
2690
*58
.558
.20.
2450
See
foot
note
s at
end
of
tabl
e.
Tab
le 3
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
stud
ents
sam
pled
as
bacc
alau
reat
e re
cipi
ents
-C
ontin
ued
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
ts
data
Est
imat
edbi
as
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
non
resp
onde
nts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Did
the
stud
ent r
ecei
ve a
ny in
stitu
tion
fina
ncia
l aid
?Y
es3,
540
990
30.1
26.2
1.00
20*
28.8
29.1
-0.3
210
No
7,81
02,
710
69.9
73.8
-1.0
020*
71.2
70.9
0.32
10
Inst
itutio
n re
gion
New
Eng
land
680
280
6.4
7.3
-0.2
430
6.6
6.6
0.02
90M
id E
ast
2,00
068
017
.717
.9-0
.066
017
.417
.7-0
.336
0G
reat
Lak
es2,
020
600
17.2
15.9
0.32
0017
.216
.80.
3720
Plai
ns96
024
08.
86.
90.
4730
8.6
8.3
0.30
00So
uthe
ast
2,67
083
022
.321
.30.
2450
21.7
22.1
-0.3
330
Sout
hwes
t1,
140
420
9.8
12.0
-0.5
440*
10.4
10.4
0.06
20R
ocky
Mou
ntai
n44
010
03.
72.
60.
2760
3.6
3.4
0.21
40Fa
r W
est
1,32
048
013
.414
.9-0
.392
013
.513
.8-0
.322
0O
utly
ing
area
120
600.
91.
2-0
.069
01.
01.
00.
0140
Did
the
stud
ent r
ecei
ve a
ny P
ell g
rant
s?Y
es2,
590
790
21.2
20.5
0.16
5020
.621
.0-0
.442
0N
o8,
750
2,91
078
.879
.5-0
.165
079
.479
.00.
4420
Pell
cate
gori
es f
or a
ll Pe
ll re
cipi
ents
Pell
amou
nt <
= $
1,13
867
018
028
.626
.00.
6370
28.2
27.9
0.31
60$1
,138
< P
ell a
mou
nt <
= $
1,77
567
020
025
.727
.0-0
.323
025
.326
.1-0
.737
0$1
,775
< P
ell a
mou
nt<
=$2
,975
630
190
23.8
24.2
-0.1
000
24.2
23.9
0.34
10$2
,975
< P
ell a
mou
nt63
021
021
.922
.8-0
.213
022
.222
.10.
0800
Wha
t was
the
amou
nt o
f th
e Pe
ll gr
ant r
ecei
ved?
2,59
079
018
20.7
1853
.8-8
.168
418
32.9
1828
.93.
9669
Inst
itutio
n se
ctor
Publ
ic le
ss-t
han-
2-ye
ar0
00.
00.
00.
0000
0.0
0.0
0.00
00Pu
blic
2-y
ear
00
0.0
0.0
0.00
000.
00.
00.
0000
Publ
ic 4
-yea
r no
n-do
ctor
ate-
gran
ting
2,48
068
021
.416
.11.
3590
*20
.920
.10.
7780
*Pu
blic
4-y
ear
doct
orat
e-gr
antin
g4,
900
1,68
043
.948
.8-1
.230
0*44
.545
.1-0
.649
0Pr
ivat
e no
t-fo
r-pr
ofit
2-ye
ar o
r le
ss0
00.
00.
00.
0000
0.0
0.0
0.00
00Pr
ivat
e no
t-fo
r-pr
ofit
4-ye
ar n
on-d
octo
rate
-gr
antin
g2,
140
580
20.3
17.7
0.64
8019
.819
.60.
1620
Priv
ate
not-
for-
prof
it 4-
year
doc
tora
te-
gran
ting
1,69
067
013
.315
.5-0
.542
013
.713
.9-0
.189
0
Priv
ate
for-
prof
it le
ss-t
han-
2-ye
ar0
00.
00.
00.
0000
0.0
0.0
0.00
00Pr
ivat
e fo
r-pr
ofit
2-ye
ar0
00.
00.
00.
0000
0.0
0.0
0.00
00Pr
ivat
e fo
r-pr
ofit
4-ye
ar14
090
1.1
2.0
-0.2
350*
1.2
1.3
-0.1
020
Stud
ent's
mar
ital s
tatu
sSi
ngle
$$
$$
$80
.581
.1-0
.573
0*M
arri
ed$
:$
I:
18.7
18.1
0.56
20*
Sepa
rate
d$
:$
$$
0.8
0.8
0.01
10
Staf
ford
cat
egor
ies
for
Staf
ford
am
ount
<=
$3,
500
1,27
038
023
.926
.6-0
.633
023
.724
.6-0
.840
0
Staf
ford
rec
ipie
nts
$3,5
00 <
Sta
ffor
d am
ount
<=
$5,
500
2,61
070
052
.149
.20.
6700
52.1
51.4
0.67
20$5
,500
< S
taff
ord
amou
nt1,
170
360
24.0
24.2
-0.0
380
24.2
24.1
0.16
80
Am
ount
of
Staf
ford
loan
rec
eive
d5,
050
1,45
056
96.0
5695
.20.
1816
5715
.656
95.8
19.7
161
Did
the
stud
ent r
ecei
ve a
Sta
ffor
d lo
an?
Yes
5,05
01,
450
44.6
40.5
1.04
00*
43.1
43.5
-0.4
370
No
6,29
02,
250
55.4
59.5
-1.0
400*
56.9
56.5
0.43
70
See
foot
note
s at
end
of
tabl
e.
Tab
le 3
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
stud
ents
sam
pled
as
bacc
alau
reat
e re
cipi
ents
-C
ontin
ued
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI n
onre
spon
se a
djus
tmen
t-un
impu
ted
data
Afte
r w
eigh
t ad
'ust
men
ts-im
pute
d da
ta
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
tsM
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Did
the
stud
ent r
ecei
ve a
ny s
tate
fina
ncia
l aid
?
Stu
dent
type
- C
AD
E
Yes
No
Und
ergr
adua
te s
tude
ntG
radu
ate
stud
ent
Firs
t-pr
ofes
sion
al s
tude
nt
2,26
09,
090
10,9
0041
0 40
590
3,12
0
3,52
016
0 20
19.1
80.9
96.2 3.5
0.3
15.8
84.2
94.9 4.5
0.6
0.84
90*
-0.8
490*
0.32
10*
-0.2
560
-0.0
650
18.3
81.7
96.2 3.5
0.3
18.3
81.7
95.9 3.8
0.4
-0.0
110
0.01
10
0.32
40*
-0.2
580
-0.0
660
* B
ias
is s
igni
fica
nt a
t the
0.0
5/(c
-1)
leve
l, w
here
c is
the
num
ber
of c
ateg
orie
s w
ithin
the
prim
ary
vari
able
.
:Suf
fici
ent d
ata
from
oth
er n
on-C
AT
I so
urce
s w
ere
not a
vaila
ble
prio
r to
impu
tatio
n.
'The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l. T
he in
tera
ctio
n te
rmof
this
var
iabl
e cr
osse
d w
ithst
uden
t typ
e w
as n
ot in
clud
ed in
the
nonr
espo
nse
mod
els
beca
use
the
wei
ghtin
g w
as d
one
at th
e al
l-st
uden
t lev
el a
nd n
ot s
epar
atel
y by
stu
dent
type
.
2The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l, an
d th
ere
wer
eno
t suf
fici
ent d
ata
avai
labl
e fr
omot
her
non-
CA
TI
sour
ces
to in
clud
e th
e va
riab
le in
the
nonr
espo
nse
mod
els.
;Enr
ollm
ent c
ateg
orie
s w
ere
defi
ned
by q
uart
iles
base
d on
tota
l enr
ollm
ent f
or th
e 19
97-1
998
scho
ol y
ear.
NO
TE
: Est
imat
ed b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t is
the
diff
eren
ce b
etw
een
the
mea
n fo
r C
AT
I re
spon
dent
s an
d no
nres
pond
ents
mul
tiplie
d by
the
wei
ghte
d no
nres
pons
e ra
te.
Aft
er w
eigh
t adj
ustm
ents
, est
imat
ed b
ias
is th
e di
ffer
ence
bet
wee
n m
eans
bas
ed o
n th
e C
AT
I w
eigh
ts a
nd th
e st
udy
wei
ghts
. To
prot
ect c
onfi
dent
ialit
y of
the
data
,som
e nu
mbe
rs h
ave
been
roun
ded.
SOU
RC
E: U
.S. D
epar
tmen
t of
Edu
catio
n, N
atio
nal C
ente
r fo
r E
duca
tion
Stat
istic
s, N
atio
nal P
osts
econ
dary
Stu
dent
Aid
Stu
dy, 1
999-
2000
(N
PSA
S:20
00).
Tab
le 4
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
unde
rgra
duat
e st
uden
ts
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
ts
Mea
n,st
udy
wei
ghts
Est
imat
edbi
as
35,4
9014
,220
26.4
26.3
0.01
8026
.426
.40.
0346
Stud
ent's
age
Stud
ent a
ge g
roup
s19
or
youn
ger
6,43
02,
500
22.7
21.3
0.40
0022
.422
.30.
0520
20 to
23
15,3
105,
880
34.9
34.7
0.00
0034
.734
.9-0
.161
024
to 2
95,
980
2,94
016
.119
.3-0
.900
0*17
.017
.0-0
.018
030
to 3
94,
340
1,71
014
.113
.50.
2000
13.8
13.9
-0.1
310
40 o
r ol
der
3,44
01,
180
12.2
11.2
0.30
0012
.211
.90.
2570
Has
stu
dent
rec
eive
d an
y ty
pe o
f ai
d?Y
es21
,920
7,65
052
.743
.62.
7000
*50
.150
.00.
1250
No
13,6
106,
750
47.3
56.4
-2.7
000*
49.9
50.0
-0.1
250
Did
stu
dent
atte
nd in
stitu
tion
in th
e fa
ll?Y
es, f
ull t
ime
23,1
907,
620
55.4
43.2
3.60
00*
51.8
52.0
-0.1
510
Yes
, hal
f tim
e4,
170
2,02
015
.418
.0-0
.700
0*16
.216
.20.
0320
Yes
, les
s th
an h
alf
time
2,41
01,
320
11.0
15.4
-1.3
000*
12.2
12.3
-0.0
660
No
5,61
03,
020
18.2
23.5
-1.6
000*
19.8
19.6
0.18
50A
ttend
ance
Full
time
##
#I
#38
.538
.9-
0.39
40'
Hal
f tim
e#
:#
##
16.1
16.1
-0.0
520
Les
s th
an h
alf
time
##
##
#19
.920
.3-0
.392
0M
ixed
##
##
$25
.524
.70.
8380
*C
itize
nshi
p st
atus
U.S
. citi
zen
32,4
1012
,500
93.7
91.5
0.70
00*
93.0
93.0
-0.0
180
Res
iden
t1,
440
750
4.6
5.2
-0.2
000
4.8
4.8
0.04
20V
isa
590
600
1.7
3.3
-0.5
000*
2.2
2.2
-0.0
250
CPS
mat
chY
es20
,600
7,19
050
.742
.22.
5000
*48
.348
.20.
1550
No
14,9
407,
210
49.3
57.8
-2.5
000*
51.7
51.8
-0.1
550
Dep
ende
ncy
stat
us -
two-
leve
lD
epen
dent
##
::
:50
.749
.11.
5600
*1
Inde
pend
ent
I:
:#
:49
.350
.9-1
.560
0*D
epen
denc
y st
atus
- th
ree-
leve
lD
epen
dent
1:
:#
:50
.749
.11.
5600
*'
Inde
pend
ent w
/out
dep
ende
nts
##
##
#21
.924
.0-2
.081
0*
Inde
pend
ent w
/dep
ende
nts
1#
:#
#27
.426
.90.
5210
*E
nrol
lmen
t tot
al a
t the
stu
dent
's in
stitu
tion
35,5
4014
,400
1620
7.4
1712
9.2
-274
.770
0*16
499.
416
482.
217
.249
2
Enr
ollm
ent c
ateg
orie
s3E
nrol
lmen
t<=
3,26
79,
280
3,86
017
.715
.70.
6000
*17
.117
.1-0
.027
03,
267<
enro
llmen
t<=
11,0
969,
410
3,54
028
.627
.00.
5000
28.2
28.1
0.10
40
11,0
96<
enro
llmen
t<24
,120
8,56
03,
640
28.5
30.3
-0.5
334
28.9
29.1
-0.1
690
24,1
20<
=en
rollm
ent
8,28
03,
350
25.2
27.0
-0.5
507
25.8
25.7
0.09
20W
as th
e st
uden
t enr
olle
d in
inst
itutio
n in
Yes
, at a
NPS
AS
inst
itutio
n28
,960
11,1
5079
.675
.81.
1298
*78
.378
.4-0
.125
0th
e fa
ll?Y
es, n
ot a
t a N
PSA
S in
stitu
tion
970
230
2.3
1.2
0.32
36*
1.9
2.0
-0.0
600
No
5,61
03,
020
18.1
23.0
-1.4
534*
19.8
19.6
0.18
50D
id th
e st
uden
t rec
eive
any
fed
eral
fin
anci
al a
id?
Yes
17,7
406,
210
41.3
33.8
2.21
95*
39.1
39.0
0.09
70N
o17
,800
8,19
058
.866
.2-2
.219
5*60
.961
.0-0
.097
0St
uden
t's s
exM
ale
14,0
806,
430
42.2
47.4
-1.5
688*
43.6
43.7
-0.1
010
Fem
ale
20,8
707,
890
57.8
52.6
1.56
88*
56.4
56.3
0.10
10
See
foot
note
s at
end
of
tabl
e.
Tab
le 4
.-N
onre
spon
se b
ias
befo
re C
AT
! no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
unde
rgra
duat
e st
uden
ts -
Con
tinue
d
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI n
onre
spon
se a
djus
tmen
t-un
impu
ted
data
Afte
r w
eigh
t adj
ustm
ents
-impu
ted
data
CA
TI
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
ts
Mea
n,st
udy
wei
ghts
Est
imat
edbi
as
Did
the
stud
ent r
ecei
ve a
nyY
es8,
030
2,45
016
.312
.11.
2542
*15
.215
.00.
1100
inst
itutio
n fi
nanc
ial a
id?
No
27,5
1011
,950
83.7
87.9
-1.2
542*
84.9
85.0
-0.1
100
Inst
itutio
n re
gion
New
Eng
land
1,92
080
05.
05.
1-0
.042
35.
15.
00.
0630
Mid
Eas
t5,
670
2,15
014
.513
.50.
2972
14.2
14.2
-0.0
070
Gre
at L
akes
5,85
02,
150
15.4
14.3
0.33
0015
.415
.10.
2820
Plai
ns2,
770
940
7.0
5.9
0.35
00*
6.8
6.7
0.14
80So
uthe
ast
8,20
02,
930
23.4
19.6
1.13
00*
22.4
22.3
0.16
30So
uthw
est
3,74
01,
810
11.3
14.0
-0.7
900*
12.1
12.1
0.02
70R
ocky
Mou
ntai
n1,
560
560
4.1
3.9
0.03
004.
04.
0-0
.025
0Fa
r W
est
5,10
02,
640
17.9
21.9
-1.1
900*
18.4
19.1
-0.6
670
Out
lyin
g ar
ea74
042
01.
41.
8-0
.120
01.
61.
50.
0170
Did
the
stud
ent r
ecei
ve a
ny P
ell
Yes
9,69
03,
800
23.5
20.6
0.87
00*
22.6
22.6
-0.0
010
gran
ts?
No
25,8
5010
,600
76.5
79.4
-0.8
700*
77.4
77.4
0.00
10Pe
ll ca
tego
ries
for
all
Pell
Pell
amou
nt <
= $
1,18
32,
460
910
29.5
28.9
0.17
0029
.629
.40.
2060
reci
pien
ts$1
,183
< P
ell a
mou
nt <
= $
1,95
32,
390
1,01
023
.224
.4-0
.320
023
.323
.6-0
.315
0$1
,953
< P
ell a
mou
nt4,
840
1,88
047
.246
.70.
1500
47.2
47.1
0.11
00W
hat w
as th
e am
ount
of
the
Pell
gran
t rec
eive
d?9,
690
3,80
019
10.4
1910
.5-0
.008
319
09.9
1910
.4-0
.504
8
Inst
itutio
n se
ctor
Publ
ic le
ss-t
han-
2-ye
ar74
032
00.
70.
70.
0000
0.7
0.7
0.00
00Pu
blic
2-y
ear
5,90
02,
980
43.8
49.5
-1.7
000*
45.4
45.5
-0.0
830
Publ
ic 4
-yea
r no
n-do
ctor
ate-
gran
ting
5,78
01,
950
12.8
10.3
0.75
00*
12.1
12.1
0.00
40Pu
blic
4-y
ear
doct
orat
e-gr
antin
g10
,520
3,78
021
.719
.50.
6500
*21
.121
.10.
0540
Priv
ate
not-
for-
prof
it 2-
year
or
less
970
530
0.8
0.9
-0.0
400
0.8
0.8
-0.0
010
Priv
ate
not-
for-
prof
it 4-
year
non
-doc
tora
te-g
rant
ing
4,71
01,
560
9.4
8.0
0.44
00*
9.0
9.0
-0.0
090
Priv
ate
not-
for-
prof
it 4-
year
doc
tora
te-g
rant
ing
3,26
01,
280
5.9
5.6
0.09
005.
85.
80.
0280
Priv
ate
for-
prof
it le
ss-t
han-
2-ye
ar2,
340
1,29
01.
92.
2-0
.100
02.
02.
00.
0000
Priv
ate
for-
prof
it 2-
year
780
390
1.9
2.0
-0.0
100
1.9
1.9
0.00
00
Priv
ate
for-
prof
it 4-
year
530
320
1.1
1.3
-0.0
700
1.2
1.2
0.00
80St
uden
t's m
arita
l sta
tus
Sing
le:
I#
:$
76.1
76.9
-0.7
700*
IM
arri
ed#
#:
#$
22.5
21.6
0.84
60*
Sepa
rate
d$
:I
#:
1.4
1.5
-0.0
770
Staf
ford
cat
egor
ies
for
Staf
ford
am
ount
<=
$2,
625
3,71
01,
340
32.7
33.1
-0.1
000
32.9
32.8
0.16
10St
affo
rd r
ecip
ient
s$2
,625
< S
taff
ord
amou
nt <
= $
4,42
53,
000
1,02
022
.423
.2-0
.210
022
.222
.6-0
.355
0$4
,425
< S
taff
ord
amou
nt <
= $
5,50
03,
860
1,08
022
.220
.00.
5500
*22
.021
.70.
3010
$5,5
00 <
Sta
ffor
d am
ount
3,08
01,
060
22.8
23.7
-0.2
400
22.9
23.0
-0.1
070
See
foot
note
s at
end
of
tabl
e.
Tab
le 4
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
unde
rgra
duat
e st
uden
ts -
Con
tinue
d
Bef
ore
CA
TI n
onre
spon
se a
djus
tmen
t-un
impu
ted
data
Afte
r w
eigh
adj
ustm
ents
-impu
ted
data
CA
TI
resp
onde
ntC
AT
Ino
nres
pond
ent
CA
TI u
nwei
ghte
dC
AT
I unw
eigh
ted
mea
n, s
tudy
mea
n, s
tudy
Est
imat
edM
ean,
CA
TI
Mea
n, s
tudy
Est
imat
edD
escr
iptio
nR
espo
nse
resp
onde
nts
nonr
espo
nden
tsw
eigh
tsw
eigh
tsbi
asw
eigh
tsw
eigh
tsbi
asA
mou
nt o
f Sta
fford
loan
rec
eive
dD
id th
e st
uden
t rec
eive
aY
es13
,650
4,50
046
06.3
4547
.114
.824
345
99.6
4591
.58.
1385
Staf
ford
loan
?N
o13
,650
4,50
029
.523
.21.
8700
*27
.627
.6-0
.031
0D
id th
e st
uden
t rec
eive
any
sta
teY
es21
,890
9,90
070
.576
.8-1
.870
0*72
.472
.40.
0310
fina
ncia
l aid
?N
o6,
310
1,96
016
.913
.41.
0200
*15
.915
.90.
0380
Stud
ent t
ype
- sa
mpl
edB
acca
laur
eate
rec
ipie
nt29
,220
12,4
4083
.186
.6-1
.020
0*84
.184
.2-0
.038
0O
ther
und
ergr
adua
te s
tude
nt10
,900
3,52
07.
86.
10.
4900
*7.
17.
3-0
.144
02G
radu
ate
stud
ent
24,2
8010
,830
91.3
93.7
-0.6
900*
92.0
92.0
-0.0
340
Firs
t-pr
ofes
sion
al s
tude
nt33
040
0.8
0.2
0.19
00*
0.8
0.7
0.18
00*
3010
0.1
0.1
0.00
000.
10.
1-0
.002
0
* B
ias
is s
igni
fica
nt a
t the
0.0
5/(c
1)
leve
l, w
here
c is
the
num
ber
of c
ateg
orie
s w
ithin
the
prim
ary
vari
able
.
:Suf
fici
ent d
ata
from
oth
er n
on-C
AT
I so
urce
s w
ere
not a
vaila
ble
prio
r to
impu
tatio
n.
'The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l, an
d th
ere
wer
e not
suf
fici
ent d
ata
avai
labl
e fr
omot
her
non-
CA
TI
sour
ces
to in
clud
e th
e va
riab
le in
the
nonr
espo
nse
mod
els.
2The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l. Sa
mpl
ed s
tude
nt ty
pew
as n
ot in
clud
ed in
the
nonr
espo
nse
mod
els
beca
use
it is
not
an
actu
al s
tude
nt c
hara
cter
istic
and
may
not
ref
lect
true
stu
dent
type
.
t\c7,
3Enr
ollm
ent c
ateg
orie
s w
ere
defi
ned
by q
uart
iles
base
d on
tota
l enr
ollm
ent f
or th
e 19
97-1
998
scho
ol y
ear.
CY
)N
OT
E: E
stim
ated
bia
s be
fore
CA
TI
nonr
espo
nse
adju
stm
ent i
s th
e di
ffer
ence
bet
wee
n th
e m
ean
for
CA
TI
resp
onde
nts
and
nonr
espo
nden
ts m
ultip
lied
by th
e w
eigh
ted
nonr
espo
nse
rate
.A
fter
wei
ght a
djus
tmen
ts, e
stim
ated
bia
s is
the
diff
eren
ce b
etw
een
mea
ns b
ased
on
the
CA
TI
wei
ghts
and
the
stud
y w
eigh
ts. T
o pr
otec
t con
fide
ntia
lity
of th
e da
ta, s
ome
num
bers
hav
e be
enro
unde
d.
SOU
RC
E: U
.S. D
epar
tmen
t of
Edu
catio
n, N
atio
nal C
ente
r fo
r E
duca
tion
Stat
istic
s, N
atio
nal P
osts
econ
dary
Stu
dent
Aid
Stu
dy, 1
999-
2000
(N
PSA
S:20
00).
Tab
le 5
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
grad
uate
/fir
st-
prof
essi
onal
stu
dent
s
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
'C
AT
Iun
wei
ghte
dre
spon
dent
s
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
ts
Mea
n,st
udy
wei
ghts
Est
imat
edbi
as
Stud
ent's
age
Stud
ent a
ge g
roup
s19
or
youn
ger
8,94
02,
780
32.9
31.9
0.23
30*
32.7
32.6
0.01
5320
to 2
340
100.
80.
20.
1000
*0.
80.
70.
1470
*24
to 2
982
028
010
.010
.9-0
.200
010
.110
.2-0
.059
030
to 3
93,
380
1,15
038
.143
.5-1
.300
0*39
.639
.30.
3000
40 o
r ol
der
2,55
079
027
.725
.60.
5000
26.8
27.3
-0.5
020
Has
stu
dent
rec
eive
d an
y ty
pe o
f ai
d?Y
es2,
150
550
23.4
19.8
0.80
00*
22.7
22.6
0.11
30N
o4,
330
1,30
045
.442
.60.
7000
*44
.044
.7-0
.736
00D
id s
tude
nt a
ttend
inst
itutio
n in
the
fall?
Yes
, ful
l tim
e4,
630
1,58
054
.657
.5-0
.700
0*56
.055
.30.
7360
*Y
es, h
alf
time
4,42
01,
020
44.1
32.9
2.70
00*
41.9
41.5
0.40
30Y
es, l
ess
than
hal
f tim
e1,
500
700
18.0
24.9
-1.6
000*
19.1
19.7
-0.6
020
No
1,59
058
019
.221
.4-0
.500
020
.019
.80.
2020
Atte
ndan
ceFu
ll tim
e1,
410
500
18.6
20.8
-0.5
000
19.0
19.0
-0.0
020
Hal
f tim
e:
II
##
26.8
27.8
-0.9
650*
zL
ess
than
hal
f tim
e:
::
::
19.2
18.8
0.36
40M
ixed
::
:#
:28
.327
.80.
4720
Citi
zens
hip
stat
usU
.S. c
itize
n#
##
:#
25.7
25.6
0.12
90R
esid
ent
7,26
02,
050
89.0
80.8
2.00
00*
87.5
86.8
0.73
10*1
Vis
a24
013
02.
94.
4-0
.300
0*2.
93.
2-0
.351
0*C
PS m
atch
Yes
900
500
8.0
14.8
-1.6
000*
9.7
10.0
-0.3
800
No
2,90
082
033
.530
.50.
7000
*32
.232
.8-0
.560
0D
epen
denc
y st
atus
- tw
o-le
vel
Dep
ende
nt6,
060
2,06
066
.569
.5-0
.700
0*67
.867
.20.
5600
Inde
pend
ent
##
#:
#4.
43.
21.
2470
*2
Dep
ende
ncy
stat
us -
thre
e-le
vel
Dep
ende
nt:
##
::
95.6
96.9
-1.2
470*
Inde
pend
ent w
/out
dep
ende
nts
##
##
#4.
43.
21.
2470
*2
Inde
pend
ent w
/dep
ende
nts
#:
##
#59
.963
.0-3
.072
0*E
nrol
lmen
t tot
al a
t the
stu
dent
's in
stitu
tion
:#
::
#35
.733
.91.
8240
*
Enr
ollm
ent c
ateg
orie
s3E
nrol
lmen
t<=
3,26
78,
960
2,88
017
666.
018
587.
8-2
21.2
910*
1776
0.1
1788
7.3
-127
.142
1
3,26
7<er
trol
lmen
t<=
11,0
961,
410
390
14.1
12.5
0.40
0013
.513
.8-0
.215
0
11,0
96<
enro
llmen
t<24
,120
2,16
064
025
.423
.20.
5000
25.6
24.9
0.71
90*
24,1
20<
enr
ollm
ent
2,50
085
030
.331
.0-0
.100
030
.630
.50.
0980
Was
the
stud
ent e
nrol
led
in in
stitu
tion
inY
es, a
t a N
PSA
S in
stitu
tion
2,89
01,
000
30.1
33.3
-0.8
000*
30.3
30.9
-0.6
020
the
fall?
:Yes
, not
at a
NPS
AS
inst
itutio
n7,
450
2,37
080
.679
.30.
3000
80.3
80.3
-0.0
210
No
100
100.
90.
30.
1000
*0.
70.
70.
0230
Did
the
stud
ent r
ecei
ve a
ny f
eder
al f
inan
cial
aid
?Y
es1,
410
500
18.5
20.4
-0.4
000
19.0
19.0
-0.0
020
No
2,61
075
030
.427
.70.
6000
*29
.329
.7-0
.401
06,
340
2,13
069
.672
.3-0
.600
0*70
.770
.30.
4010
..
See
foot
note
s at
end
of
tabl
e.
Tab
le 5
.-N
onre
spon
se b
ias
befo
re C
AT
Ino
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
grad
uate
/fir
st-
rofe
ssio
nal s
tude
nts
-Con
tinue
d
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI
nonr
espo
nse
adju
stm
ent-
unim
pute
d da
taA
fter
wei
ght a
djus
tmen
ts-i
mpu
ted
data
CA
T!
unw
eigh
ted
resp
onde
nts
CA
TI
unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n, C
AT
Iw
eigh
ts
Mea
n,st
udy
wei
ghts
Est
imat
edbi
as
Stud
ent's
sex
Mal
e3,
780
1,31
042
.243
.5-0
.300
043
.042
.60.
4110
Fem
ale
4,91
01,
530
57.8
56.5
0.30
0057
.057
.4-0
.411
0
Did
the
stud
ent r
ecei
ve a
nyY
es2,
430
760
22.2
21.4
0.20
0021
.422
.0-0
.537
0'1
inst
itutio
n fi
nanc
ial a
id?
No
6,53
02,
120
77.9
78.6
-0.2
000
78.6
78.0
0.53
70*
Inst
itutio
n re
gion
New
Eng
land
620
240
6.3
7.1
-0.2
000
6.4
6.5
-0.0
530
Mid
Eas
t1,
670
580
19.1
20.3
-0.3
000
19.4
19.4
0.02
60G
reat
Lak
es1,
520
490
17.5
17.9
-0.1
000
17.7
17.6
0.05
50Pl
ains
740
210
8.4
7.1
0.30
008.
38.
10.
2250
Sout
heas
t1,
810
510
20.5
18.1
0.60
0019
.719
.9-0
.232
0So
uthw
est
910
320
9.9
11.7
-0.4
000
10.5
10.3
0.13
00R
ocky
Mou
ntai
n29
050
3.2
2.1
0.30
00*
3.1
2.9
0.18
60Fa
r W
est
1,33
044
014
.414
.9-0
.100
014
.214
.5-0
.369
0O
utly
ing
area
7030
0.7
0.9
0.00
000.
80.
80.
0320
Inst
itutio
n se
ctor
Publ
ic le
ss-t
han-
2-ye
ar0
00.
00.
00.
0000
0.0
0.0
0.00
00'
Publ
ic 2
-yea
r60
02.
20.
10.
5100
*2.
31.
70.
5160
*
Publ
ic 4
-yea
r no
n-do
ctor
ate-
gran
ting
940
270
12.0
10.5
0.36
0011
.611
.7-0
.023
0
Publ
ic 4
-yea
r do
ctor
ate-
gran
ting
4,12
01,
310
44.0
44.9
-0.2
000
43.9
44.2
-0.3
360
Priv
ate
not-
for-
prof
it 2-
year
or
less
100
0.0
0.0
0.01
000.
00.
00.
0070
Priv
ate
not-
for-
prof
it 4-
year
non
-doc
tora
te-g
rant
ing
700
220
9.2
9.7
-0.1
200
9.4
9.4
0.05
40
Priv
ate
not-
for-
prof
it 4-
year
doc
tora
te-g
rant
ing
2,89
098
030
.532
.6-0
.510
030
.831
.0-0
.172
0Pr
ivat
e fo
r-pr
ofit
less
-tha
n-2-
year
00
0.0
0.0
0.00
000.
00.
00.
0030
Priv
ate
for-
prof
it 2-
year
00
0.0
0.0
0.00
000.
00.
00.
0000
Priv
ate
for-
prof
it 4-
year
240
902.
02.
2-0
.050
02.
02.
0-0
.048
0
Stud
ent's
mar
ital s
tatu
sSi
ngle
##
##
#53
.656
.0-2
.439
0*2
Mar
ried
##
##
#45
.643
.22.
3830
*
Sepa
rate
dI
::
::
0.9
0.8
0.05
70St
affo
rd c
ateg
orie
s fo
rSt
affo
rd a
mou
nt <
= $
8,00
064
019
023
.423
.40.
0000
22.8
23.4
-0.6
300
Staf
ford
rec
ipie
nts
$8,0
00<
Sta
ffor
d am
ount
<=
$12
,521
620
180
23.3
23.7
-0.0
900
22.9
23.4
-0.4
840
$12,
521
< S
taff
ord
amou
nt <
= $
18,5
0095
026
039
.937
.50.
5500
40.1
39.4
0.72
90$1
8,50
0 <
Sta
ffor
d am
ount
320
110
13.4
15.5
-0.4
600
14.3
13.9
0.38
50
See
foot
note
s at
end
of
tabl
e.
tV2
CO
* B
ias
is s
igni
fica
nt a
t the
0.0
5/(c
1)
leve
l, w
here
c is
the
num
ber
of c
ateg
orie
s w
ithin
the
prim
ary
vari
able
.
:Suf
fici
ent d
ata
from
oth
er n
on-C
AT
I so
urce
s w
ere
not a
vaila
ble
prio
r to
impu
tatio
n.
Tab
le 5
.-N
onre
spon
se b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t and
aft
er w
eigh
t adj
ustm
ents
for
sel
ecte
d va
riab
les
for
grad
uate
/fir
st-
prof
essi
onal
stu
dent
s -C
ontin
ued
Des
crip
tion
Res
pons
e
Bef
ore
CA
TI n
onre
spon
se a
djus
tmen
t-un
impu
ted
data
Afte
r w
eigh
adj
ustm
ents
-impu
ted
data
CA
TI u
nwei
ghte
dre
spon
dent
sC
AT
I unw
eigh
ted
nonr
espo
nden
ts
CA
TI
resp
onde
ntm
ean,
stu
dyw
eigh
ts
CA
TI
nonr
espo
nden
tm
ean,
stu
dyw
eigh
tsE
stim
ated
bias
Mea
n C
AT
Iw
eigh
ts
Mea
n, s
tudy
wei
ghts
Est
imat
edbi
as
2,54
073
014
078.
914
316.
2-5
3.59
0614
339.
214
132.
520
6.71
80*
Am
ount
of
Staf
ford
loan
rec
eive
d2,
540
730
29.6
27.3
0.54
00*
28.6
29.0
-0.4
540
Did
the
stud
ent r
ecei
ve a
Yes
Staf
ford
loan
?N
o6,
420
2,15
070
.472
.7-0
.540
0*71
.471
.00.
4540
260
903.
13.
00.
0200
3.0
3.1
-0.1
110
Did
the
stud
ent r
ecei
ve a
ny s
tate
Yes
fina
ncia
l aid
?N
o8,
690
2,79
096
.997
.0-0
.020
097
.096
.90.
1110
440
180
1.8
2.5
0.18
00*
1.8
2.0
0.19
50*
Stud
ent t
ype
- sa
mpl
edB
acca
laur
eate
rec
ipie
ntO
ther
und
ergr
adua
te s
tude
nt34
060
7.0
3.3
0.87
00*
6.9
6.1
0.81
00*
Gra
duat
e st
uden
t7,
280
2,36
078
.881
.6M
.670
0*79
.179
.5-0
.313
0Fi
rst-
prof
essi
onal
stu
dent
890
280
12.5
12.6
-0.0
200
12.2
12.5
-0.3
020
Stud
ent t
ype
- C
AD
EU
nder
grad
uate
stu
dent
00
0.0
0.0
0.00
000.
00.
00.
0000
Gra
duat
e st
uden
t8,
040
2,60
087
.688
.2-0
.140
087
.887
.80.
0000
Firs
t-pr
ofes
sion
al s
tude
nt92
028
012
.411
.80.
1400
12.2
12.2
0.00
00
'The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l. T
he in
tera
ctio
n te
rmof
this
var
iabl
e cr
osse
d w
ith s
tude
nt ty
pew
as n
ot in
clud
ed in
the
nonr
espo
nse
mod
els
beca
use
the
wei
ghtin
g w
as d
one
at th
e al
l-st
uden
t lev
el a
nd n
ot s
epar
atel
y by
stu
dent
type
.
2The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
at t
he 0
.05
leve
l, an
d th
ere
wer
eno
t suf
fici
ent d
ata
avai
labl
e fr
om o
ther
non
-C
AT
I so
urce
s to
incl
ude
the
vari
able
in th
e no
nres
pons
e m
odel
s.
3Enr
ollm
ent c
ateg
orie
s w
ere
defi
ned
by q
uart
iles
base
d on
tota
l enr
ollm
ent f
or th
e 19
97-1
998
scho
olye
ar.
NO
TE
: Est
imat
ed b
ias
befo
re C
AT
I no
nres
pons
e ad
just
men
t is
the
diff
eren
ce b
etw
een
the
mea
n fo
r C
AT
I re
spon
dent
s an
d no
nres
pond
ents
mul
tiplie
d by
the
wei
ghte
d no
nres
pons
e ra
te. A
fter
wei
ght a
djus
tmen
ts, e
stim
ated
bia
s is
the
diff
eren
ce b
etw
een
mea
ns b
ased
on
the
CA
TI
wei
ghts
and
the
stud
y w
eigh
ts. T
o pr
otec
t con
fide
ntia
lity
of th
e da
ta, s
ome n
umbe
rs h
ave
been
rou
nded
.
SOU
RC
E: U
.S. D
epar
tmen
t of
Edu
catio
n, N
atio
nal C
ente
r fo
r E
duca
tion
Stat
istic
s, N
atio
nal P
osts
econ
dary
Stu
dent
Aid
Stu
dy, 1
999-
2000
(N
PSA
S:20
00).
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Table 6.Summary of significant nonresponse bias before CATI nonresponse adjustment bystudent tune
DescriptionAll students
Baccalaureaterecipients
Undergraduatestudents
Graduate/first-professional
students
Student's age T T TStudent age groups T T T THas student received any type of aid? T T T TDid student attend institution in the fall? T T T TCitizenship status T T T TCPS match T T T TEnrollment total at the student's
institutionT T T T
Enrollment categories2 T T T TWas the student enrolled in institution in
the fall?T T T T
Did the student receive any federalfinancial aid?
T T T T
Student's sex T T TDid the student receive any institution
financial aid?T T T
Institution region T T T TDid the student receive any Pell grants? T T tPell categories for all Pell recipientsWhat was the amount of the Pell grant
received?
T tt
Institution sector T T T TStafford categories for Stafford recipients3 TAmount of Stafford loan receivedDid the student receive a Stafford loan? T T T TDid the student receive any state financial
aid?T T T
Student type sampled T t T TStudent type CADE T T tT denotes significance at the 0.05/(c-1) level for at least one category of the primary variable, where c is the number of categorieswithin the primary variable.
t Not applicable
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
20
30
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
3. Weight Adjustments
Weight adjustments are typically used to reduce bias due to unit nonresponse, and the resultsin tables 1, 2, 3, 4, and 5 show that these adjustments are important for reducing the potential fornonresponse bias due to the differences between CATI respondents and nonrespondents. Aftercomputing study weights for study respondents by making various adjustments to the design-basedweights, adjustments were made for CATI nonresponse. In the initial nonresponse models allvariables were incorporated that were thought to be predictive of CATI nonresponse and weremissing for five percent or less of all study respondents including:
age (categorical),any aid receipt indicator,fall attendance status,citizenship,CPS record indicator,institution enrollment from IPEDS IC file (categorical),fall enrollment status,federal aid receipt indicator,sex,Hispanic indicator,institutional aid receipt indicator,OBE region,student date of birth preloaded into CATI,parent data preloaded into CATI,total number of phone numbers obtained for student,Social Security number indicator,Pell grant status,Pell grant amount (categorical),Stafford loan status,Stafford loan amount (categorical),institution type,state aid receipt indicator,number of institutions attended in 1999-2000, andstudent type.
Other variables that were considered but excluded from the "not located" model because they weremissing for more than five percent of all study respondents were:
dependents indicator, dependency status, number of dependents,full-year attendance status,high school degree indicator and type,high school graduation year,local residence,parents' income, parents' family size, parent's marital status,student's marital status
21
3I
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
student's income, andrace.
Table 7 lists the predictor variables used for each of the three final nonresponse adjustment models.Dependency status and student's marital status were included in the final other nonresponse models(see discussion below of the three models). Marital status was also included in the final refusalmodel.
Also, a Chi-squared automatic interaction detector (CHAID) analysis was performed on thecandidate predictor variables to determine important interactions. The CHAID analysis divided thedata into segments that differed with respect to the response variable: not located, refusal, or othernonresponse. The segmentation process first divided the sample into groups based on categories ofthe most significant predictor of response. It then split each of these groups into smaller subgroupsbased on other predictor variables. It also merged categories of a variable that were foundinsignificant. This splitting and merging process continued until no more statistically significantpredictors were found (or until some other stopping rule was met). The interactions from the finalCHAID segments were then defined.
The resulting segment interactions and all the main effect variables were then subjected tovariable screening in the logistic procedure. Variables significant at the 15 percent significancelevel were retained, with the exception of institution type, student type, Pell grant status, andStafford loan status, which were retained whether or not they were significant. It was determinedthat Pell grant status and Stafford loan status are important predictors of federal aid receipt, so thesevariables were retained in all nonresponse models to preserve the population totals of thesepredictor variables. Additionally, institution type and student type were retained in all nonresponsemodels because of their importance as stratification variables.
The adjustment for CATI nonresponse was performed in three stages because the predictorsof response propensity were potentially different at each stage:
(1) inability to locate the student(2) refusal to be interviewed(3) other non-interview
Using these three stages of nonresponse adjustment achieved greater reduction in nonresponse biasto the extent that different variables were significant predictors of response propensity at each stage.Six of the variables are only in one model as main effects, seven variables are in two models asmain effects, and eight variables, including the four variables forced into all models, are in all threemodels as main effects. Additionally, some variables were included as a main effect in one modeland as part of an interaction in another model. For example, ethnicity is a main effect in the refusalmodel but part of interactions in the other two models, as shown in table 7.
22
32
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Table 7.Variables used in final NPSAS:2000 CATI nonresponse models
Variable sector Not located model Refusal model Other nonresponse model
Institutional sector X X XRegion X X XStudent type X X XAge group X XSex X X XInstitutional aid recipient X XFederal aid recipient XPell grant recipient X X XStafford loan recipient X X XCitizenship X XEthnicity XFall enrollment XFall attendance XEnrollment X XNumber of phone numbers X XNumber of schools attended X X XDate of birth preloaded in CATI X X XCPS match XParent information preloaded in CATI X XMarital status X XDependency X2 CHAID segments based on ethnicity,institutional aid receipt, and number ofschools attended
X
10 CHAID segments based on aid receipt,number of schools attended, fallattendance, region, enrollment, and agegroup
X
11 CHAID segments based on citizenship,number of schools attended, ethnicity,federal aid receipt, institutional sector, fallattendance, marital status, and fallenrollment
X
NOTE: The variables institution sector, student type, receipt of Pel grant, and receipt of Stafford loan were forced intoall three models.
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
Poststratification to control totals was used to adjust for the potential for bias resulting fromframe errors. The CATI weights were adjusted to control totals using a generalized rakingprocedure. The control totals established during the poststratification of the study weights also wereused for the CATI weights. These control totals were for annual student enrollment, by institutiontype; total number of Pell grants awarded; amount of Pell grants awarded, by institution type; andamount of Stafford loans awarded, by institution type. To help reduce nonresponse bias further,additional control totals were formed for annual enrollment by student type as well as control totalsby:
23
33
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
sex,age group (less-than-24, 24-29, and 30+),federal aid applicant,federal aid receipt,state aid receipt,institution aid receipt, andfall attendance status.
The annual enrollment control totals by student type were obtained from the study weights so thatestimates of the annual enrollment using the study or CATI weights would be the same. The otherseven control totals listed above were also computed using the study weights because thesevariables were known for most CATI respondents and nonrespondents.
All nonresponse adjustment and poststratification models were fit using RTI's proprietarygeneralized exponential models (GEMs)2, which are logistic models incorporating bounds on theadjustment factors. Section 6.1 of the NPSAS methodology report describes the weightingprocedure in more detail.
4. Bias for CATI Variables
The before-CATI nonresponse adjustment bias was also estimated for several CATIvariables that were missing for CATI nonrespondents but known for more than 90 percent of CATIrespondents. For the CATI respondents, it was assumed that the respondents who initially refusedto be interviewed had characteristics similar to CATI refusals, and that the respondents who weredifficult to contact, based on the number of phone call attempts, had characteristics similar tostudents who were never located. Table 8 shows the estimated bias before adjustment under theseassumptions.
The bias due to refusals was estimated as the difference between the mean for CATIrespondents who were initial refusals and the mean for all other respondents, using the CATIweight. T-tests were used to test each level of the variables for significance of the bias at the0.05/(c-1) significance level, where c is the number of categories within the primary variable. Chi-squared tests were used to test if the distribution based on the CATI weights was significantlydifferent at the 0.05 level from the distribution based on the study weights. To conduct thesestatistical tests, the study and CATI respondents were combined and the study respondents based onstudy weights were contrasted with the CATI respondents based on CATI weights. Then,SUDAAN was used to compute the variance and to test for significant differences. SUDAANcomputed the variance using institution strata and PSUs and took account of the correlation in theestimates caused by having students on both sides of the contrast.
2 Folsom, R.E. and A.C. Singh (2000). "The Generalized Exponential Model for Sampling Weight Calibration forExtreme Values, Nonresponse, and Poststratification." Proceedings of the Section on Survey Research Methods of theAmerican Statistical Association, pp. 598-603.
24
34
Tab
le 8
.-N
onre
spon
se b
ias
for
CA
TI
vari
able
s fo
r al
l stu
dent
s
Des
crip
tion
Res
pons
e
Mea
n, C
AT
I w
eigh
tsM
ean,
CA
TI
wei
ghts
Initi
al r
efus
alre
spon
dent
sO
ther
resp
onde
nts
Est
imat
ed b
ias
Dif
ficu
lt to
cont
act
resp
onde
nts
Oth
erre
spon
dent
sE
stim
ated
bia
s11
.59.
61.
948*
I8.
010
.4-2
.415
*IR
ecei
ved
any
empl
oyer
aid
Yes
No
88.5
90.4
-1.9
48*
92.0
89.6
2.41
5'W
orke
d w
hile
in s
choo
lY
es74
.678
.7-4
.087
*I79
.077
.81.
244'
No
20.7
19.4
1.23
617
.120
.4-
3.33
5'
Mis
sing
4.8
1.9
2.85
2*3.
91.
82.
091*
Wor
ked
20 o
r m
ore
hour
s pe
r w
eek
whi
le in
sch
ool
Yes
62.9
65.5
-2.5
97"
66.6
64.6
2.01
8*1
No
32.4
32.6
-0.2
5429
.533
.6-
4.10
9'
Mis
sing
4.8
1.9
2.85
2*3.
91.
8-4
.109
*
Wor
ked
mul
tiple
jobs
in 1
999-
2000
Yes
17.0
21.2
-4.1
62*1
21.4
20.3
1.12
5'
No
79.3
78.1
1.20
975
.779
.0-3
.362
*
Mis
sing
3.8
0.8
2.95
3*2.
90.
72.
238*
Bor
n ou
tsid
e th
e U
.S.
Yes
7.5
12.2
-4.6
49*1
9.8
12.0
-2.1
19*1
No
92.5
87.9
4.64
9*90
.288
.02.
119*
Reg
iste
red
to v
ote
Yes
81.5
82.4
-0.9
4080
.682
.8-2
.189
*1
No
18.6
17.6
0.94
019
.417
.22.
189*
Vot
ed in
the
2000
ele
ctio
nsY
es71
.578
.1-6
.551
*I64
.881
.0-1
6.20
2*1
No
28.5
21.9
6.55
1*35
.219
.016
.202
*
Has
a d
isab
ility
Yes
9.7
10.2
-0.5
57'
8.8
10.6
-1.7
77*1
NO
84.3
88.7
-4.4
29*
86.7
88.5
- 1.
775'
Mis
sing
6.1
1.1
4.98
6*4.
51.
03.
552*
Atte
nded
mor
e th
an o
ne in
stitu
tion
in 1
999-
2000
Yes
5.0
6.0
-0.9
49"
5.9
5.8
0.05
7
No
95.0
94.0
0.94
9*94
.194
.2-0
.057
Has
dep
ende
nts
othe
r th
an a
spo
use
Yes
27.8
28.7
-0.8
9323
.430
.2-6
.808
*1
No
72.2
71.3
0.89
376
.769
.96.
808'
Has
chi
ldre
n un
der
5 ye
ars
old
Yes
13.8
14.5
-0.6
6712
.115
.2-3
.090
*1
No
86.2
85.5
0.66
787
.984
.83.
090*
Has
chi
ldre
n ag
ed 5
to 1
2 ye
ars
old
Yes
14.7
15.3
-0.5
8011
.416
.4-5
.062
"N
o85
.384
.70.
580
88.6
83.6
5.06
2*
U.S
. Arm
ed F
orce
s ve
tera
nY
es4.
64.
40.
156
3.1
4.9
-1.7
68*1
No
89.0
88.1
0.88
589
.987
.72.
172*
_Mis
sing
6.4
7.5
-1.0
417.
07.
4-0
.404
* B
ias
is s
igni
fica
nt a
t the
0.0
5/(c
-1)
leve
l, w
here
c is
the
num
ber
of c
ateg
orie
s w
i hin
the
prim
ary
vari
able
.
'The
dis
trib
utio
n ba
sed
on th
e C
AT
I w
eigh
ts is
sig
nifi
cant
ly d
iffe
rent
at t
he 0
.05
leve
l fro
m th
e di
stri
butio
n ba
sed
on th
e st
udy
wei
ghts
.
SOU
RC
E: U
.S. D
epar
tmen
t of
Edu
catio
n, N
atio
nal C
ente
r fo
r E
duca
tion
Stat
istic
s, N
atio
nal P
osts
econ
dary
Stu
dent
Aid
Stu
dy, 1
999-
2000
(N
PSA
S:20
00).
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
The bias due to inability to contact the student was estimated as the difference between the meanfor CATI respondents who were difficult to contact and the mean for all other respondents, usingthe CATI weight. Again, t-tests were performed to test the significance of the bias for each levelof the variables, and Chi-Squared tests were performed to test the significance of thedistributions of each variable.
The bias was generally higher when comparing difficult-to-locate students to the otherrespondents than when comparing the initial refusals to the other respondents. These biasestimates indicate that using the three nonresponse models was the proper approach becauseinitial refusals differ from other respondents and difficult-to-locate students also differ fromother respondents.
5. Bias After Weight Adjustments
Although tables 2 through 5 show that some bias remains after all weight adjustments forseveral variables, the magnitude of the residual bias shown in these tables is usually very small.The second set of columns in tables 2 through 5 shows the estimated bias after weightadjustments for the variables available for most responding and nonresponding students. Thebias after weight adjustments is the difference between the means based on the CATI weightsand the study weights. For all students combined, Pell grant receipt, Pell grant amount,institution sector, and student type CADE have zero bias after weight adjustments because allstudents combined were controlled to known totals.
For baccalaureate recipients and graduate/first-professional students, some sectors had nostudents and therefore no bias. For undergraduate students, some sectors that were all or mostlycomprised of undergraduate students had zero bias because all students combined werecontrolled to totals for sectors. For graduate/first-professional students, student type - CADEhad zero bias because all students combined were controlled to graduate and first-professionalstudent totals.
Figures 1 through 4 compare the estimated relative bias before CATI nonresponseadjustments with the estimated relative bias after weight adjustments. All four figures indicatethat when the relative bias was large before CATI nonresponse adjustment, it was almost alwaysreduced dramatically after weight adjustments. When the relative bias was small before CATInonresponse adjustment, it stayed small after weight adjustments with occasional smallincreases. These figures clearly show that the CATI weight adjustments significantly reducedbias for all students combined, baccalaureate recipients, undergraduate students, andgraduate/first-professional students.
The exceptions when the bias was large before CATI nonresponse adjustment andremained large after weight adjustments were due to small sample sizes. For example, infigure 3, the outlier is for undergraduate students sampled as graduate students, and in figure 4,the outliers are for graduate students in less-than-4-year institution sectors.
26
36
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
The absolute bias decreased after weight adjustments for many variables. For variousstudent groups, the percentage of variable categories that did not increase after weightadjustments were:
all students combined 94.7 percentbaccalaureate recipients 79.4 percentundergraduate students 89.9 percentgraduate/first-professional students 65.2 percent.
For all students combined, some of the Pell grant and Stafford loan amount categories hadincreased bias after weight adjustments. The estimated bias is not significant for thesecategories, and this increase occurred because Pell grant and Stafford loan amounts werepoststratified to known program totals by sector (different categories than shown in the table).For baccalaureate recipients, undergraduate students, and graduate/first-professional students, thereasons for this increase were poststratification to totals for some of these variables, some samplesizes are small for some student types, and the weighting was done at the all-student level andnot separately by student type.
Similarly to the CATI variable bias, t-tests were performed to test the significance of thebias for each level of the variables, and Chi-Squared tests were performed to test the significanceof the distributions of each variable. Below and in table 9 are summaries of the after-weightingbias across the four tables:
for all students combined, six variables had significant t-tests and five variableshad significant Chi-Squared testsfor baccalaureate recipients, nine variables had significant t-tests and fivevariables had significant Chi-Squared testsfor undergraduate students, five variables had significant t-tests and five variableshad significant Chi-Squared testsfor graduate/first-professional students, 12 variables had significant t-tests and 8variables had significant Chi-Squared teststhe variables attendance status and dependency status (two-levels and three-levels) had significant t-tests and Chi-Squared tests for all four student typesstudent's marital status had significant t-tests for all four student types andsignificant Chi-squared tests for three of the student typessignificant biases are usually small and sometimes are due to small sample sizes.
There is not sufficient reported data available for the variables that are significantlybiased for all students combined to eliminate the bias altogether. That is, there is too muchmissing data for these variables to be included as poststratification control totals. Other variablesshow significant bias when analyzed separately for baccalaureate recipients, undergraduatestudents, and graduate/first-professional students, but not for all students combined.
Bias remaining after weight adjustments for variables based exclusively (or primarily)upon CATI data cannot be estimated because there is no data on these variables for CATInonrespondents. This analysis focused on the bias due to CATI nonresponse.
27
37
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Figure 1.Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for all students
Estimated relativebias after weightadjustments
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
4iiitiotootast
0.05 0.1 0.15 0.2 0.25 0.3 0.35
Estimated relative bias before CATI nonresponse adjustment
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
28
33
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Figure 2.Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for students sampled as baccalaureaterecipients
Estimatedrelative biasafter weightadjustments
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
et.****
0.05 0.1 0.15 0.2 0.25 0.3 0.35
Estimated relative bias before CATI nonresponse adjustment
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
29
39
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Figure 3.Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for undergraduate students
Estimated relative bias before CATI nonresponse adjustment
0.35
Outlier due to small sample size.
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
30
40
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Figure 4.Nonresponse bias before CATI nonresponse adjustment and after weightadjustments for selected variables for graduate/first professional students
Estimatedrelative biasafter weightadjustments
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0 roltt.0 0.05 0.1 0.15 0.2 0.25 0.3
Estimated relative bias before CATI nonresponse adjustment
0.35
Outliers due to small sample size.
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
31
41
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
Table 9.Summary of significant nonresponse bias after weight adjustments by student
Description All studentsBaccalaureate
recipientsUndergraduate
students
Graduate/first-professional
students
Student's age TStudent age groups TC THas student received any type of
aid?TC
Did student attend institution in thefall?
Attendance TC TC TC TCCitizenship status TC TCCPS matchDependency status two-level TC TC TC TCDependency status three-level TC TC TC TCEnrollment total at the student's
institutionEnrollment categories2 TWas the student enrolled in
institution in the fall?Did the student receive any federal
financial aid?Student's sexDid the student receive any
institution financial aid?TC
Institution regionDid the student receive any Pell
grants?Pell categories for all Pell
recipientsWhat was the amount of the Pell
grant received?
Tt
t
t
Institution sector T TCStudent's marital status TC T TC TCStafford categories for all Stafford
recipients3Amount of Stafford Loan received TDid the student receive a Stafford
loan?Did the student receive any state
financial aid?Student type sampled TC t TC TStudent type CADE T tT denotes significance at the 0.05/(c-1) level for at least one category of the primary variable, where c is the number of categorieswithin the primary variable.
C denotes significant difference at the 0.05 level between the distribution based on the CATI weights and the distribution basedon the study weights.
t Not applicable
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
42
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
6. ROC Curve
As described above, three nonresponse adjustment models were used. In order to assessthe overall predictive ability of the combined models, a Receiver Operating Characteristics(ROC) curve was used. As shown in figure 1, the area under the ROC curve developed for theoverall predicted response propensity was about 0.66 which corresponds to a highly significantWilcoxon test statistic.3 The curve indicates that in about two of every three randomly chosenpairs of sample students, one responding and the other nonresponding, the predicted overallresponse propensity of the respondent will be greater than that of the nonrespondent. This levelof discrimination implies that the variables used in the three models are highly informative butnot definitive predictors of a sample student's overall response propensity.
Figure 5.ROC curve for overall response propensity
1
0.9
0.8 -
0.7
0.6
0.5
0.4 -
0.3
0.2
0.1
0t0
SOURCE: U.S. Department of Education, National Center for Education Statistics, National Postsecondary Student Aid Study,1999-2000 (NPSAS:2000).
7. Conclusions
Information from multiple sources was used in weighting the data to reduce CATInonresponse bias. Examination of variables known for most respondents and nonrespondentsbefore CATI nonresponse adjustment revealed that some bias existed. In the initial nonresponsemodels all variables were incorporated that were thought to be predictive of CATI nonresponse
3 Hanley, J.A. and B.J. McNeil (1982). "The meaning and use of the area under a receiver-operating characteristic(ROC) curve. Diagnostic Radiology, 143:29-36.
33
43
National Postsecondary Student Aid Study 1999-2000 (NPSAS:2000)CATI Nonresponse Bias Analysis Report
and were missing for five percent or less of all study respondents. Important interactions amongthese variables were also included in the initial models. Three nonresponse models were used toreduce bias. Comparing CATI' respondents who were initial refusals with other respondents andcomparing CATI respondents who were difficult to contact with other respondents also indicatesthat three models would help reduce bias. Using these three stages of nonresponse adjustmentachieved greater reduction in nonresponse bias to the extent that different variables weresignificant predictors of response propensity at each stage. For poststratifying the CATI weights,control totals were used that were also used for poststratifying the study weights, and sevenadditional control totals were computed using the study weights for seven variables known formost respondents and nonrespondents.
The relative bias decreased considerably after weight adjustments--especially when it waslarge before CATI nonresponse adjustment. And the relative bias remained small after weightadjustments when it was small before CATI nonresponse adjustment. As shown in figures 1through 4, CATI nonresponse bias was reduced using weighting techniques, and the remainingrelative bias ranged from 0 to 0.35 percent.
44
Listing of NCES Working Papers to Date
Working papers can be downloaded as pdf files from the NCES Electronic Catalog (http://nces.ed.gov/pubsearch/).You can also contact Sheilah Jupiter at (202) 502-7444 (sheilah [email protected]) if you are interested in any of thefollowing papers.
Listing of NCES Working Papers by Program AreaNo. Title NCES contact
Baccalaureate and Beyond (B&B)98-15 Development of a Prototype System for Accessing Linked NCES Data
2001-15 Baccalaureate and Beyond Longitudinal Study: 2000/01 Follow-Up Field TestMethodology Report
Beginning Postsecondary Students (BPS) Longitudinal Study98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report98-15 Development of a Prototype System for Accessing Linked NCES Data
1999-15 Projected Postsecondary Outcomes of 1992 High School Graduates2001-04 Beginning Postsecondary Students Longitudinal Study: 1996-2001 (BPS:1996/2001)
Field Test Methodology Report
Common Core of Data (CCD)95-12 Rural Education Data User's Guide96-19 Assessment and Analysis of School-Level Expenditures97-15 Customer Service Survey: Common Core of Data Coordinators97-43 Measuring Inflation in Public School Costs98-15 Development of a Prototype System for Accessing Linked NCES Data
1999-03 Evaluation of the 1996-97 Nonfiscal Common Core of Data Surveys Data Collection,Processing, and Editing Cycle
2000-12 Coverage Evaluation of the 1994-95 Common Core of Data: PublicElementary/Secondary School Universe Survey
2000-13 Non-professional Staff in the Schools and Staffing Survey (SASS) and Common Core ofData (CCD)
2001-09 An Assessment of the Accuracy of CCD Data: A Comparison of 1988,1989, and 1990CCD Data with 1990-91 SASS Data
2001-14 Evaluation of the Common Core of Data (CCD) Finance Data Imputations2002-02 School Locale Codes 1987 - 2000
Data Development2000-16a Lifelong Learning NCES Task Force: Final Report Volume I2000-16b Lifelong Learning NCES Task Force: Final Report Volume II
Decennial Census School District Project95-12 Rural Education Data User's Guide96-04 Census Mapping Project/School District Data Book98-07 Decennial Census School District Project Planning Report
2001-12 Customer Feedback on the 1990 Census Mapping Project
Early Childhood Longitudinal Study (ECLS)96-08 How Accurate are Teacher Judgments of Students' Academic Performance?96-18 Assessment of Social Competence, Adaptive Behaviors, and Approaches to Learning with
Young Children97-24 Formulating a Design for the ECLS: A Review of Longitudinal Studies97-36 Measuring the Quality of Program Environments in Head Start and Other Early Childhood
Programs: A Review and Recommendations for Future Research1999-01 A Birth Cohort Study: Conceptual and Design Considerations and Rationale2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
1999 AAPOR Meetings
35
45
Steven KaufmanAndrew G. Malizio
Aurora D'Amico
Steven KaufmanAurora D'AmicoPaula Knepper
Samuel PengWilliam J. Fowler, Jr.Lee HoffmanWilliam J. Fowler, Jr.Steven KaufmanBeth Young
Beth Young
Kerry Gruber
John Sietsema
Frank JohnsonFrank Johnson
Lisa HudsonLisa Hudson
Samuel PengTai PhanTai PhanDan Kasprzyk
Jerry WestJerry West
Jerry WestJerry West
Jerry WestDan Kasprzyk
No. Title2001-02 Measuring Father Involvement in Young Children's Lives: Recommendations for a
Fatherhood Module for the ECLS-B2001-03 Measures of Socio-Emotional Development in Middle Childhood2001-06 Papers from the Early Childhood Longitudinal Studies Program: Presented at the 2001
AERA and SRCD Meetings
Education Finance Statistics Center (EDFIN)94-05 Cost-of-Education Differentials Across the States96-19 Assessment and Analysis of School-Level Expenditures97-43 Measuring Inflation in Public School Costs98-04 Geographic Variations in Public Schools' Costs
1999-16 Measuring Resources in Education: From Accounting to the Resource Cost ModelApproach
High School and Beyond (HS&B)95-12 Rural Education Data User's Guide
1999-05 Procedures Guide for Transcript Studies1999-06 1998 Revision of the Secondary School Taxonomy
HS Transcript Studies1999-05 Procedures Guide for Transcript Studies1999-06 1998 Revision of the Secondary School Taxonomy
International Adult Literacy Survey (IALS)97-33 Adult Literacy: An International Perspective
Integrated Postsecondary Education Data System (IPEDS)97-27 Pilot Test of IPEDS Finance Survey98-15 Development of a Prototype System for Accessing Linked NCES Data
2000-14 IPEDS Finance Data Comparisons Under the 1997 Financial Accounting Standards forPrivate, Not-for-Profit Institutes: A Concept Paper
National Assessment of Adult Literacy (NAAL)98-17 Developing the National Assessment of Adult Literacy: Recommendations from
Stakeholders1999-09a 1992 National Adult Literacy Survey: An Overview1999-09b 1992 National Adult Literacy Survey: Sample Design1999-09c 1992 National Adult Literacy Survey: Weighting and Population Estimates1999-09d 1992 National Adult Literacy Survey: Development of the Survey Instruments1999-09e 1992 National Adult Literacy Survey: Scaling and Proficiency Estimates1999-09f 1992 National Adult Literacy Survey: Interpreting the Adult Literacy Scales and Literacy
Levels1999-09g 1992 National Adult Literacy Survey: Literacy Levels and the Response Probability
Convention2000-05 Secondary Statistical Modeling With the National Assessment of Adult Literacy:
Implications for the Design of the Background Questionnaire2000-06 Using Telephone and Mail Surveys as a Supplement or Alternative to Door-to-Door
Surveys in the Assessment of Adult Literacy2000-07 "How Much Literacy is Enough?" Issues in Defining and Reporting Performance
Standards for the National Assessment of Adult Literacy2000-08 Evaluation of the 1992 NALS Background Survey Questionnaire: An Analysis of Uses
with Recommendations for Revisions2000-09 Demographic Changes and Literacy Development in a Decade2001-08 Assessing the Lexile Framework: Results of a Panel Meeting
National Assessment of Educational Progress (NAEP)95-12 Rural Education Data User's Guide97-29 Can State Assessment Data be Used to Reduce State NAEP Sample Sizes?97-30 ACT's NAEP Redesign Project: Assessment Design is the Key to Useful and Stable
Assessment Results
NCES contactJerry West
Elvira HauskenJerry West
William J. Fowler, Jr.William J. Fowler, Jr.William J. Fowler, Jr.William J. Fowler, Jr.William J. Fowler, Jr.
Samuel PengDawn NelsonDawn Nelson
Dawn NelsonDawn Nelson
Marilyn Binkley
Peter StoweSteven KaufmanPeter Stowe
Sheida White
Alex SedlacekAlex SedlacekAlex SedlacekAlex SedlacekAlex SedlacekAlex Sedlacek
Alex Sedlacek
Sheida White
Sheida White
Sheida White
Sheida White
Sheida WhiteSheida White
Samuel PengSteven GormanSteven Gorman
No. Title97-31 NAEP Reconfigured: An Integrated Redesign of the National Assessment of Educational
Progress97-32 Innovative Solutions to Intractable Large Scale Assessment (Problem 2: Background
Questionnaires)97-37 Optimal Rating Procedures and Methodology for NAEP Open-ended Items97-44 Development of a SASS 1993-94 School-Level Student Achievement Subfile: Using
State Assessments and State NAEP, Feasibility Study98-15 Development of a Prototype System for Accessing Linked NCES Data
1999-05 Procedures Guide for Transcript Studies1999-06 1998 Revision of the Secondary School Taxonomy2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
2001-08 Assessing the Lexile Framework: Results of a Panel Meeting2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP2001-19 The Measurement of Home Background Indicators: Cognitive Laboratory Investigations
of the Responses of Fourth and Eighth Graders to Questionnaire Items and ParentalAssessment of the Invasiveness of These Items
National Education Longitudinal Study of 1988 (NELS:88)95-04 National Education Longitudinal Study of 1988: Second Follow-up Questionnaire Content
Areas and Research Issues95-05 National Education Longitudinal Study of 1988: Conducting Trend Analyses of NLS-72,
HS&B, and NELS:88 Seniors95-06 National Education Longitudinal Study of 1988: Conducting Cross-Cohort Comparisons
Using HS&B, NAEP, and NELS:88 Academic Transcript Data95-07 National Education Longitudinal Study of 1988: Conducting Trend Analyses HS&B and
NELS:88 Sophomore Cohort Dropouts95-12 Rural Education Data User's Guide95-14 Empirical Evaluation of Social, Psychological, & Educational Construct Variables Used
in NCES Surveys96-03 National Education Longitudinal Study of 1988 (NELS:88) Research Framework and
Issues98-06 National Education Longitudinal Study of 1988 (NELS:88) Base Year through Second
Follow-Up: Final Methodology Report98-09 High School Curriculum Structure: Effects on Coursetaking and Achievement in
Mathematics for High School GraduatesAn Examination of Data from the NationalEducation Longitudinal Study of 1988
98-15 Development of a Prototype System for Accessing Linked NCES Data1999-05 Procedures Guide for Transcript Studies1999-06 1998 Revision of the Secondary School Taxonomy1999-15 Projected Postsecondary Outcomes of 1992 High School Graduates2001-16 Imputation of Test Scores in the National Education Longitudinal Study of 1988
National Household Education Survey (NHES)95-12 Rural Education Data User's Guide96-13 Estimation of Response Bias in the NHES:95 Adult Education Survey96-14 The 1995 National Household Education Survey: Reinterview Results for the Adult
Education Component96-20 1991 National Household Education Survey (NHES:91) Questionnaires: Screener, Early
Childhood Education, and Adult Education96-21 1993 National Household Education Survey (NHES:93) Questionnaires: Screener, School
Readiness, and School Safety and Discipline96-22 1995 National Household Education Survey (NHES:95) Questionnaires: Screener, Early
Childhood Program Participation, and Adult Education96-29 Undercoverage Bias in Estimates of Characteristics of Adults and 0- to 2-Year-Olds in the
1995 National Household Education Survey (NHES:95)96-30 Comparison of Estimates from the 1995 National Household Education Survey
(NHES:95)
37
47
NCES contactSteven Gorman
Steven Gorman
Steven GormanMichael Ross
Steven KaufmanDawn NelsonDawn NelsonArnold Goldstein
She ida WhiteArnold GoldsteinArnold GoldsteinArnold Goldstein
Jeffrey Owings
Jeffrey Owings
Jeffrey Owings
Jeffrey Owings
Samuel PengSamuel Peng
Jeffrey Owings
Ralph Lee
Jeffrey Owings
Steven KaufmanDawn NelsonDawn NelsonAurora D'AmicoRalph Lee
Samuel PengSteven KaufmanSteven Kaufman
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
No. Title97-02 Telephone Coverage Bias and Recorded Interviews in the 1993 National Household
Education Survey (NHES:93)97-03 1991 and 1995 National Household Education Survey Questionnaires: NHES:91 Screener,
NHES:91 Adult Education, NHES:95 Basic Screener, and NHES:95 Adult Education97-04 Design, Data Collection, Monitoring, Interview Administration Time, and Data Editing in
the 1993 National Household Education Survey (NHES:93)97-05 Unit and Item Response, Weighting, and Imputation Procedures in the 1993 National
Household Education Survey (NHES:93)97-06 Unit and Item Response, Weighting, and Imputation Procedures in the 1995 National
Household Education Survey (NHES:95)97-08 Design, Data Collection, Interview Timing, and Data Editing in the 1995 National
Household Education Survey97-19 National Household Education Survey of 1995: Adult Education Course Coding Manual97-20 National Household Education Survey of 1995: Adult Education Course Code Merge
Files User's Guide97-25 1996 National Household Education Survey (NHES:96) Questionnaires:
Screener/Household and Library, Parent and Family Involvement in Education andCivic Involvement, Youth Civic Involvement, and Adult Civic Involvement
97-28 Comparison of Estimates in the 1996 National Household Education Survey97-34 Comparison of Estimates from the 1993 National Household Education Survey97-35 Design, Data Collection, Interview Administration Time, and Data Editing in the 1996
National Household Education Survey97-38 Reinterview Results for the Parent and Youth Components of the 1996 National
Household Education Survey97-39 Undercoverage Bias in Estimates of Characteristics of Households and Adults in the 1996
National Household Education Survey97-40 Unit and Item Response Rates, Weighting, and Imputation Procedures in the 1996
National Household Education Survey98-03 Adult Education in the 1990s: A Report on the 1991 National Household Education
Survey98-10 Adult Education Participation Decisions and Barriers: Review of Conceptual Frameworks
and Empirical Studies
National Longitudinal Study of the High School Class of 1972 (NLS-72)95-12 Rural Education Data User's Guide
National Postsecondary Student Aid Study (NPSAS)96-17 National Postsecondary Student Aid Study: 1996 Field Test Methodology Report
2000-17 National Postsecondary Student Aid Study:2000 Field Test Methodology Report2002-03 National Postsecondary Student Aid Study, 1999-2000 (NPSAS:2000), CATI
Nonresponse Bias Analysis Report.
National Study of Postsecondary Faculty (NSOPF)97-26 Strategies for Improving Accuracy of Postsecondary Faculty Lists98-15 Development of a Prototype System for Accessing Linked NCES Data
2000-01 1999 National Study of Postsecondary Faculty (NSOPF:99) Field Test Report
Postsecondary Education Descriptive Analysis Reports (PEDAR)2000-11 Financial Aid Profile of Graduate Students in Science and Engineering
Private School Universe Survey (PSS)95-16 Intersurvey Consistency in NCES Private School Surveys95-17 Estimates of Expenditures for Private K-12 Schools96-16 Strategies for Collecting Finance Data from Private Schools96-26 Improving the Coverage of Private Elementary-Secondary Schools96-27 Intersurvey Consistency in NCES Private School Surveys for 1993-9497-07 The Determinants of Per-Pupil Expenditures in Private Elementary and Secondary
Schools: An Exploratory Analysis97-22 Collection of Private School Finance Data: Development of a Questionnaire98-15 Development of a Prototype System for Accessing Linked NCES Data
48
NCES contactKathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Peter StowePeter Stowe
Kathryn Chandler
Kathryn ChandlerKathryn ChandlerKathryn Chandler
Kathryn Chandler
Kathryn Chandler
Kathryn Chandler
Peter Stowe
Peter Stowe
Samuel Peng
Andrew G. MalizioAndrew G. MalizioAndrew Malizio
Linda ZimblerSteven KaufmanLinda Zimbler
Aurora D'Amico
Steven KaufmanStephen BroughmanStephen BroughmanSteven KaufmanSteven KaufmanStephen Broughman
Stephen BroughmanSteven Kaufman
No. Title2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
Recent College Graduates (RCG)98-15 Development of a Prototype System for Accessing Linked NCES Data
Schools and Staffing Survey (SASS)94-01 Schools and Staffing Survey (SASS) Papers Presented at Meetings of the American
Statistical Association94-02 Generalized Variance Estimate for Schools and Staffing Survey (SASS)94-03 1991 Schools and Staffing Survey (SASS) Reinterview Response Variance Report94-04 The Accuracy of Teachers' Self-reports on their Postsecondary Education: Teacher
Transcript Study, Schools and Staffing Survey94-06 Six Papers on Teachers from the 1990-91 Schools and Staffing Survey and Other Related
Surveys95-01 Schools and Staffing Survey: 1994 Papers Presented at the 1994 Meeting of the American
Statistical Association95-02 QED Estimates of the 1990-91 Schools and Staffing Survey: Deriving and Comparing
QED School Estimates with CCD Estimates95-03 Schools and Staffing Survey: 1990-91 SASS Cross-Questionnaire Analysis95-08 CCD Adjustment to the 1990-91 SASS: A Comparison of Estimates95-09 The Results of the 1993 Teacher List Validation Study (TLVS)95-10 The Results of the 1991-92 Teacher Follow-up Survey (TFS) Reinterview and Extensive
Reconciliation95-11 Measuring Instruction, Curriculum Content, and Instructional Resources: The Status of
Recent Work95-12 Rural Education Data User's Guide95-14 Empirical Evaluation of Social, Psychological, & Educational Construct Variables Used
in NCES Surveys95-15 Classroom Instructional Processes: A Review of Existing Measurement Approaches and
Their Applicability for the Teacher Follow-up Survey95-16 Intersurvey Consistency in NCES Private School Surveys95-18 An Agenda for Research on Teachers and Schools: Revisiting NCES' Schools and
Staffing Survey96-01 Methodological Issues in the Study of Teachers' Careers: Critical Features of a Truly
Longitudinal Study96-02 Schools and Staffing Survey (SASS): 1995 Selected papers presented at the 1995 Meeting
of the American Statistical Association96-05 Cognitive Research on the Teacher Listing Form for the Schools and Staffing Survey96-06 The Schools and Staffing Survey (SASS) for 1998-99: Design Recommendations to
Inform Broad Education Policy96-07 Should SASS Measure Instructional Processes and Teacher Effectiveness?96-09 Making Data Relevant for Policy Discussions: Redesigning the School Administrator
Questionnaire for the 1998-99 SASS96-10 1998-99 Schools and Staffing Survey: Issues Related to Survey Depth96-11 Towards an Organizational Database on America's Schools: A Proposal for the Future of
SASS, with comments on School Reform, Governance, and Finance96-12 Predictors of Retention, Transfer, and Attrition of Special and General Education
Teachers: Data from the 1989 Teacher Followup Survey96-15 Nested Structures: District-Level Data in the Schools and Staffing Survey96-23 Linking Student Data to SASS: Why, When, How96-24 National Assessments of Teacher Quality96-25 Measures of Inservice Professional Development: Suggested Items for the 1998-1999
Schools and Staffing Survey96-28 Student Learning, Teaching Quality, and Professional Development: Theoretical
Linkages, Current Measurement, and Recommendations for Future Data Collection97-01 Selected Papers on Education Surveys: Papers Presented at the 1996 Meeting of the
American Statistical Association97-07 The Determinants of Per-Pupil Expenditures in Private Elementary and Secondary
Schools: An Exploratory Analysis
39
49
NCES contactDan Kasprzyk
Stephen Broughman
Steven Kaufman
Dan Kasprzyk
Dan KasprzykDan KasprzykDan Kasprzyk
Dan Kasprzyk
Dan Kasprzyk
Dan Kasprzyk
Dan KasprzykDan KasprzykDan KasprzykDan Kasprzyk
Sharon Bobbitt &John RalphSamuel PengSamuel Peng
Sharon Bobbitt
Steven KaufmanDan Kasprzyk
Dan Kasprzyk
Dan Kasprzyk
Dan KasprzykDan Kasprzyk
Dan KasprzykDan Kasprzyk
Dan KasprzykDan Kasprzyk
Dan Kasprzyk
Dan KasprzykDan KasprzykDan KasprzykDan Kasprzyk
Mary Rollefson
Dan Kasprzyk
Stephen Broughman
No. Title97-09 Status of Data on Crime and Violence in Schools: Final Report97-10 Report of Cognitive Research on the Public and Private School Teacher Questionnaires
for the Schools and Staffing Survey 1993-94 School Year97-11 International Comparisons of Inservice Professional Development97-12 Measuring School Reform: Recommendations for Future SASS Data Collection97-14 Optimal Choice of Periodicities for the Schools and Staffing Survey: Modeling and
Analysis97-18 Improving the Mail Return Rates of SASS Surveys: A Review of the Literature97-22 Collection of Private School Finance Data: Development of a Questionnaire97-23 Further Cognitive Research on the Schools and Staffing Survey (SASS) Teacher Listing
Form97-41 Selected Papers on the Schools and Staffing Survey: Papers Presented at the 1997 Meeting
of the American Statistical Association97-42 Improving the Measurement of Staffing Resources at the School Level: The Development
of Recommendations for NCES for the Schools and Staffing Survey (SASS)97-44 Development of a SASS 1993-94 School-Level Student Achievement Subfile: Using
State Assessments and State NAEP, Feasibility Study98-01 Collection of Public School Expenditure Data: Development of a Questionnaire98-02 Response Variance in the 1993-94 Schools and Staffing Survey: A Reinterview Report98-04 Geographic Variations in Public Schools' Costs98-05 SASS Documentation: 1993-94 SASS Student Sampling Problems; Solutions for
Determining the Numerators for the SASS Private School (3B) Second-Stage Factors98-08 The Redesign of the Schools and Staffing Survey for 1999-2000: A Position Paper98-12 A Bootstrap Variance Estimator for Systematic PPS Sampling98-13 Response Variance in the 1994-95 Teacher Follow-up Survey98-14 Variance Estimation of Imputed Survey Data98-15 Development of a Prototype System for Accessing Linked NCES Data98-16 A Feasibility Study of Longitudinal Design for Schools and Staffing Survey
1999-02 Tracking Secondary Use of the Schools and Staffing Survey Data: Preliminary Results1999-04 Measuring Teacher Qualifications1999-07 Collection of Resource and Expenditure Data on the Schools and Staffing Survey1999-08 Measuring Classroom Instructional Processes: Using Survey and Case Study Fieldtest
Results to Improve Item Construction1999-10 What Users Say About Schools and Staffing Survey Publications1999-12 1993-94 Schools and Staffing Survey: Data File User's Manual, Volume III: Public-Use
Codebook1999-13 1993-94 Schools and Staffing Survey: Data File User's Manual, Volume IV: Bureau of
Indian Affairs (BIA) Restricted-Use Codebook1999-14 1994-95 Teacher Followup Survey: Data File User's Manual, Restricted-Use Codebook1999-17 Secondary Use of the Schools and Staffing Survey Data2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
1999 AAPOR Meetings2000-10 A Research Agenda for the 1999-2000 Schools and Staffing Survey2000-13 Non-professional Staff in the Schools and Staffing Survey (SASS) and Common Core of
Data (CCD)2000-18 Feasibility Report: School-Level Finance Pretest, Public School District Questionnaire
Third International Mathematics and Science Study (TIMSS)2001-01 Cross-National Variation in Educational Preparation for Adulthood: From Early
Adolescence to Young Adulthood2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
2002-01 Legal and Ethical Issues in the Use of Video in Education Research
50
NCES contactLee HoffmanDan Kasprzyk
Dan KasprzykMary RollefsonSteven Kaufman
Steven KaufmanStephen BroughmanDan Kasprzyk
Steve Kaufman
Mary Rollefson
Michael Ross
Stephen BroughmanSteven KaufmanWilliam J. Fowler, Jr.Steven Kaufman
Achievement (student) - mathematics2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics
Adult education96-14 The 1995 National Household Education Survey: Reinterview Results for the Adult
Education Component96-20 1991 National Household Education Survey (NHES:91) Questionnaires: Screener, Early
Childhood Education, and Adult Education96-22 1995 National Household Education Survey (NHES:95) Questionnaires: Screener, Early
Childhood Program Participation, and Adult Education98-03 Adult Education in the 1990s: A Report on the 1991 National Household Education
Survey98-10 Adult Education Participation Decisions and Barriers: Review of Conceptual Frameworks
and Empirical Studies1999-11 Data Sources on Lifelong Learning Available from the National Center for Education
Statistics2000-16a Lifelong Learning NCES Task Force: Final Report Volume I2000-16b Lifelong Learning NCES Task Force: Final Report Volume II
Adult literacysee Literacy of adults
American Indian education1999-13 1993-94 Schools and Staffing Survey: Data File User's Manual, Volume IV: Bureau of
Indian Affairs (BIA) Restricted-Use Codebook
Assessment/achievement95-12 Rural Education Data User's Guide .
95-13 Assessing Students with Disabilities and Limited English Proficiency97-29 Can State Assessment Data be Used to Reduce State NAEP Sample Sizes?97-30 ACT's NAEP Redesign Project: Assessment Design is the Key to Useful and Stable
Assessment Results97-31 NAEP Reconfigured: An Integrated Redesign of the National Assessment of Educational
Progress97-32 Innovative Solutions to Intractable Large Scale Assessment (Problem 2: Background
Questions)97-37 Optimal Rating Procedures and Methodology for NAEP Open-ended Items97-44 Development of a SASS 1993-94 School-Level Student Achievement Subfile: Using
State Assessments and State NAEP, Feasibility Study98-09 High School Curriculum Structure: Effects on Coursetaking and Achievement in
Mathematics for High School GraduatesAn Examination of Data from the NationalEducation Longitudinal Study of 1988
2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the ThirdInternational Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP2001-19 The Measurement of Home Background Indicators: Cognitive Laboratory Investigations
of the Responses of Fourth and Eighth Graders to Questionnaire Items and ParentalAssessment of the Invasiveness of These Items
Beginning students in postsecondary education98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report2001-04 Beginning Postsecondary Students Longitudinal Study: 1996-2001 (BPS:1996/2001)
Field Test Methodology Report
NCES contact
Patrick Gonzales
Steven Kaufman
Kathryn Chandler
Kathryn Chandler
Peter Stowe
Peter Stowe
Lisa Hudson
Lisa HudsonLisa Hudson
Kerry Gruber
Samuel PengJames HouserLarry OgleLarry Ogle
Larry Ogle
Larry Ogle
Larry OgleMichael Ross
Jeffrey Owings
Arnold Goldstein
Arnold GoldsteinArnold GoldsteinArnold Goldstein
Aurora D'Amico
Paula Knepper
No. Title NCES contact
Civic participation97-25 1996 National Household Education Survey (NHES:96) Questionnaires:
Screener/Household and Library, Parent and Family Involvement in Education andCivic Involvement, Youth Civic Involvement, and Adult Civic Involvement
Climate of schools95-14 Empirical Evaluation of Social, Psychological, & Educational Construct Variables Used
in NCES Surveys
Cost of education indices94-05 Cost-of-Education Differentials Across the States
Course-taking95-12 Rural Education Data User's Guide98-09 High School Curriculum Structure: Effects on Coursetaking and Achievement in
Mathematics for High School GraduatesAn Examination of Data from the National
1999-051999-06
Crime97-09
Curriculum95-11
98-09
Education Longitudinal Study of 1988Procedures Guide for Transcript Studies1998 Revision of the Secondary School Taxonomy
Status of Data on Crime and Violence in Schools: Final Report
Measuring Instruction, Curriculum Content, and Instructional Resources: The Status ofRecent Work
High School Curriculum Structure: Effects on Coursetaking and Achievement inMathematics for High School GraduatesAn Examination of Data from the NationalEducation Longitudinal Study of 1988
Customer service1999-10 What Users Say About Schools and Staffing Survey Publications2000-02 Coordinating NCES Surveys: Options, Issues, Challenges, and Next Steps2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
1999 AAPOR Meetings2001-12 Customer Feedback on the 1990 Census Mapping Project
Data quality97-13 Improving Data Quality in NCES: Database-to-Report Process
2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP2001-19 The Measurement of Home Background Indicators: Cognitive Laboratory Investigations
of the Responses of Fourth and Eighth Graders to Questionnaire Items and ParentalAssessment of the Invasiveness of These Items
Data warehouse2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
1999 AAPOR Meetings
Design effects2000-03 Strengths and Limitations of Using SUDAAN, Stata, and WesVarPC for Computing
Variances from NCES Data Sets
Dropout rates, high school95-07 National Education Longitudinal Study of 1988: Conducting Trend Analyses HS&B and
NELS:88 Sophomore Cohort Dropouts
Early childhood education96-20 1991 National Household Education Survey (NHES:91) Questionnaires: Screener, Early
Childhood Education, and Adult Education
Kathryn Chandler
Samuel Peng
William J. Fowler, Jr.
Samuel PengJeffrey Owings
Dawn NelsonDawn Nelson
Lee Hoffman
Sharon Bobbitt &John Ralph
Jeffrey Owings
Dan KasprzykValena PliskoDan Kasprzyk
Dan Kasprzyk
Susan AhmedArnold GoldsteinArnold GoldsteinArnold Goldstein
Dan Kasprzyk
Ralph Lee
Jeffrey Owings
Kathryn Chandler
No. Title96-22 1995 National Household Education Survey (NHES:95) Questionnaires: Screener, Early
Childhood Program Participation, and Adult Education97-24 Formulating a Design for the ECLS: A Review of Longitudinal Studies97-36 Measuring the Quality of Program Environments in Head Start and Other Early Childhood
Programs: A Review and Recommendations for Future Research1999-01 A Birth Cohort Study: Conceptual and Design Considerations and Rationale2001-02 Measuring Father Involvement in Young Children's Lives: Recommendations for a
Fatherhood Module for the ECLS-B2001-03 Measures of Socio-Emotional Development in Middle School2001-06 Papers from the Early Childhood Longitudinal Studies Program: Presented at the 2001
AERA and SRCD Meetings
Educational attainment98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report2001-15 Baccalaureate and Beyond Longitudinal Study: 2000/01 Follow-Up Field Test
Methodology Report
Educational research2000-02 Coordinating NCES Surveys: Options, Issues, Challenges, and Next Steps2002-01 Legal and Ethical Issues in the Use of Video in Education Research
Eighth-graders2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics
Employment96-03 National Education Longitudinal Study of 1988 (NELS:88) Research Framework and
Issues98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report2000-16a Lifelong Learning NCES Task Force: Final Report Volume I2000-16b Lifelong Learning NCES Task Force: Final Report Volume II2001-01 Cross-National Variation in Educational Preparation for Adulthood: From Early
Adolescence to Young Adulthood
Employment after college2001-15 Baccalaureate and Beyond Longitudinal Study: 2000/01 Follow-Up Field Test
Methodology Report
Engineering2000-11 Financial Aid Profile of Graduate Students in Science and Engineering
Enrollment after college2001-15 Baccalaureate and Beyond Longitudinal Study: 2000/01 Follow-Up Field Test
Methodology Report
Faculty higher education97-26 Strategies for Improving Accuracy of Postsecondary Faculty Lists
2000-01 1999 National Study of Postsecondary Faculty (NSOPF:99) Field Test Report
Fathers role in education2001-02 Measuring Father Involvement in Young Children's Lives: Recommendations for a
Fatherhood Module for the ECLS-B
Finance elementary and secondary schools94-05 Cost-of-Education Differentials Across the States96-19 Assessment and Analysis of School-Level Expenditures98-01 Collection of Public School Expenditure Data: Development of a Questionnaire
1999-07 Collection of Resource and Expenditure Data on the Schools and Staffing Survey1999-16 Measuring Resources in Education: From Accounting to the Resource Cost Model
Approach
43
53
NCES contactKathryn Chandler
Jerry WestJerry West
Jerry WestJerry West
Elvira HauskenJerry West
Aurora D'Amico
Andrew G. Malizio
Valena PliskoPatrick Gonzales
Patrick Gonzales
Jeffrey Owings
Aurora D'Amico
Lisa HudsonLisa HudsonElvira Hausken
Andrew G. Malizio
Aurora D'Amico
Andrew G. Malizio
Linda ZimblerLinda Zimbler
Jerry West
William J. Fowler, Jr.William J. Fowler, Jr.Stephen BroughmanStephen BroughmanWilliam J. Fowler, Jr.
No. Title2000-18 Feasibility Report: School-Level Finance Pretest, Public School District Questionnaire2001-14 Evaluation of the Common Core of Data (CCD) Finance Data Imputations
Finance postsecondary97-27 Pilot Test of IPEDS Finance Survey
2000-14 IPEDS Finance Data Comparisons Under the 1997 Financial Accounting Standards forPrivate, Not-for-Profit Institutes: A Concept Paper
Finance private schools95-17 Estimates of Expenditures for Private K-12 Schools96-16 Strategies for Collecting Finance Data from Private Schools97-07 The Determinants of Per-Pupil Expenditures in Private Elementary and Secondary
Schools: An Exploratory Analysis97-22 Collection of Private School Finance Data: Development of a Questionnaire
1999-07 Collection of Resource and Expenditure Data on the Schools and Staffing Survey2000-15 Feasibility Report: School-Level Finance Pretest, Private School Questionnaire
Geography98-04 Geographic Variations in Public Schools' Costs
Graduate students2000-11 Financial Aid Profile of Graduate Students in Science and Engineering
Graduates of postsecondary education2001-15 Baccalaureate and Beyond Longitudinal Study: 2000/01 Follow-Up Field Test
Methodology Report
Imputation2000-04
2001-102001-142001-162001-172001-18
Inflation97-43
Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and1999 AAPOR Meeting
Comparison of Proc Impute and Schafer's Multiple Imputation SoftwareEvaluation of the Common Core of Data (CCD) Finance Data ImputationsImputation of Test Scores in the National Education Longitudinal Study of 1988A Study of Imputation AlgorithmsA Study of Variance Estimation Methods
Measuring Inflation in Public School Costs
Institution data2000-01 1999 National Study of Postsecondary Faculty (NSOPF:99) Field Test Report
Instructional resources and practices95-11 Measuring Instruction, Curriculum Content, and Instructional Resources: The Status of
Recent Work1999-08 Measuring Classroom Instructional Processes: Using Survey and Case Study Field Test
Results to Improve Item Construction
International comparisons97-11 International Comparisons of Inservice Professional Development97-16 International Education Expenditure Comparability Study: Final Report, Volume I97-17 International Education Expenditure Comparability Study: Final Report, Volume II,
Quantitative Analysis of Expenditure Comparability2001-01 Cross-National Variation in Educational Preparation for Adulthood: From Early
Adolescence to Young Adulthood2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
International comparisons math and science achievement2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics
54
NCES contactStephen BroughmanFrank Johnson
Peter StowePeter Stowe
Stephen BroughmanStephen BroughmanStephen Broughman
Stephen BroughmanStephen BroughmanStephen Broughman
William J. Fowler, Jr.
Aurora D'Amico
Andrew G. Malizio
Dan Kasprzyk
Sam PengFrank JohnsonRalph LeeRalph LeeRalph Lee
William J. Fowler, Jr.
Linda Zimbler
Sharon Bobbin &John RalphDan Kasprzyk
Dan KasprzykShelley BurnsShelley Burns
Elvira Hausken
Arnold Goldstein
Patrick Gonzales
No. Title NCES contact
Libraries94-07 Data Comparability and Public Policy: New Interest in Public Library Data Papers
Presented at Meetings of the American Statistical Association97-25 1996 National Household Education Survey (NHES:96) Questionnaires:
Screener/Household and Library, Parent and Family Involvement in Education andCivic Involvement, Youth Civic Involvement, and Adult Civic Involvement
Limited English Proficiency95-13 Assessing Students with Disabilities and Limited English Proficiency
2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP
Literacy of adults98-17 Developing the National Assessment of Adult Literacy: Recommendations from
Stakeholders1999-09a 1992 National Adult Literacy Survey: An Overview1999-09b 1992 National Adult Literacy Survey: Sample Design1999-09c 1992 National Adult Literacy Survey: Weighting and Population Estimates1999-09d 1992 National Adult Literacy Survey: Development of the Survey Instruments1999-09e 1992 National Adult Literacy Survey: Scaling and Proficiency Estimates1999-09f 1992 National Adult Literacy Survey: Interpreting the Adult Literacy Scales and Literacy
Levels1999-09g 1992 National Adult Literacy Survey: Literacy Levels and the Response Probability
Convention1999-11 Data Sources on Lifelong Learning Available from the National Center for Education
Statistics2000-05 Secondary Statistical Modeling With the National Assessment of Adult Literacy:
Implications for the Design of the Background Questionnaire2000-06 Using Telephone and Mail Surveys as a Supplement or Alternative to Door-to-Door
Surveys in the Assessment of Adult Literacy2000-07 "How Much Literacy is Enough?" Issues in Defining and Reporting Performance
Standards for the National Assessment of Adult Literacy2000-08 Evaluation of the 1992 NALS Background Survey Questionnaire: An Analysis of Uses
with Recommendations for Revisions2000-09 Demographic Changes and Literacy Development in a Decade2001-08 Assessing the Lexile Framework: Results of a Panel Meeting
Literacy of adults international97-33 Adult Literacy: An International Perspective
Mathematics98-09 High School Curriculum Structure: Effects on Coursetaking and Achievement in
Mathematics for High School GraduatesAn Examination of Data from the NationalEducation Longitudinal Study of 1988
1999-08 Measuring Classroom Instructional Processes: Using Survey and Case Study Field TestResults to Improve Item Construction
2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance
Parental involvement in education96-03 National Education Longitudinal Study of 1988 (NELS:88) Research Framework and
Issues97-25 1996 National Household Education Survey (NHES:96) Questionnaires:
Screener/Household and Library, Parent and Family Involvement in Education andCivic Involvement, Youth Civic Involvement, and Adult Civic Involvement
1999-01 A Birth Cohort Study: Conceptual and Design Considerations and Rationale
Carrol Kindel
Kathryn Chandler
James HouserArnold GoldsteinArnold Goldstein
Sheida White
Alex SedlacekAlex SedlacekAlex SedlacekAlex SedlacekAlex SedlacekAlex Sedlacek
Alex Sedlacek
Lisa Hudson
Sheida White
Sheida White
She ida White
Sheida White
Sheida WhiteShe ida White
Marilyn Binkley
Jeffrey Owings
Dan Kasprzyk
Patrick GonzalesArnold Goldstein
Arnold Goldstein
Jeffrey Owings
Kathryn Chandler
Jerry West
No. Title2001-06 Papers from the Early Childhood Longitudinal Studies Program: Presented at the 2001
AERA and SRCD Meetings2001-19 The Measurement of Home Background Indicators: Cognitive Laboratory Investigations
of the Responses of Fourth and Eighth Graders to Questionnaire Items and ParentalAssessment of the Invasiveness of These Items
Participation rates98-10 Adult Education Participation Decisions and Barriers: Review of Conceptual Frameworks
and Empirical Studies
Postsecondary education1999-11 Data Sources on Lifelong Learning Available from the National Center for Education
Statistics2000-16a Lifelong Learning NCES Task Force: Final Report Volume I2000-16b Lifelong Learning NCES Task Force: Final Report Volume II
Postsecondary education persistence and attainment98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report1999-15 Projected Postsecondary Outcomes of 1992 High School Graduates
Postsecondary education staff97-26 Strategies for Improving Accuracy of Postsecondary Faculty Lists
2000-01 1999 National Study of Postsecondary Faculty (NSOPF:99) Field Test Report
Principals2000-10 A Research Agenda for the 1999-2000 Schools and Staffing Survey
Private schools96-16 Strategies for Collecting Finance Data from Private Schools97-07 The Determinants of Per-Pupil Expenditures in Private Elementary and Secondary
Schools: An Exploratory Analysis97-22 Collection of Private School Finance Data: Development of a Questionnaire
2000-13 Non-professional Staff in the Schools and Staffing Survey (SASS) and Common Core ofData (CCD)
2000-15 Feasibility Report: School-Level Finance Pretest, Private School Questionnaire
Projections of education statistics1999-15 Projected Postsecondary Outcomes of 1992 High School Graduates
Public school finance1999-16 Measuring Resources in Education: From Accounting to the Resource Cost Model
Approach2000-18 Feasibility Report: School-Level Finance Pretest, Public School District Questionnaire
Public schools97-43 Measuring Inflation in Public School Costs98-01 Collection of Public School Expenditure Data: Development of a Questionnaire98-04 Geographic Variations in Public Schools' Costs
1999-02 Tracking Secondary Use of the Schools and Staffing Survey Data: Preliminary Results2000-12 Coverage Evaluation of the 1994-95 Public Elementary/Secondary School Universe
Survey2000-13 Non-professional Staff in the Schools and Staffing Survey (SASS) and Common Core of
Data (CCD)2002-02 Locale Codes 1987 - 2000
56
NCES contactJerry West
Arnold Goldstein
Peter Stowe
Lisa Hudson
Lisa HudsonLisa Hudson
Aurora D'Amico
Aurora D'Amico
Linda ZimblerLinda Zimbler
Dan Kasprzyk
Stephen BroughmanStephen Broughman
Stephen BroughmanKerry Gruber
Stephen Broughman
Aurora D'Amico
William J. Fowler, Jr.
Stephen Broughman
William J. Fowler, Jr.Stephen BroughmanWilliam J. Fowler, Jr.Dan KasprzykBeth Young
Kerry Gruber
Frank Johnson
No. Title NCES contact
Public schools secondary98-09 High School Curriculum Structure: Effects on Coursetaking and Achievement in
Mathematics for High School GraduatesAn Examination of Data from the NationalEducation Longitudinal Study of 1988
Reform, educational96-03 National Education Longitudinal Study of 1988 (NELS:88) Research Framework and
Issues
Response rates98-02 Response Variance in the 1993-94 Schools and Staffing Survey: A Reinterview Report
School districts2000-10 A Research Agenda for the 1999-2000 Schools and Staffing Survey
School districts, public98-07 Decennial Census School District Project Planning Report
1999-03 Evaluation of the 1996-97 Nonfiscal Common Core of Data Surveys Data Collection,Processing, and Editing Cycle
School districts, public demographics of96-04 Census Mapping Project/School District Data Book
Schools97-42 Improving the Measurement of Staffing Resources at the School Level: The Development
of Recommendations for NCES for the Schools and Staffing Survey (SASS)98-08 The Redesign of the Schools and Staffing Survey for 1999-2000: A Position Paper
1999-03 Evaluation of the 1996-97 Nonfiscal Common Core of Data Surveys Data Collection,Processing, and Editing Cycle
2000-10 A Research Agenda for the 1999-2000 Schools and Staffing Survey2002-02 Locale Codes 1987 - 2000
Schools safety and discipline97-09 Status of Data on Crime and Violence in Schools: Final Report
Science2000-11 Financial Aid Profile of Graduate Students in Science and Engineering2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
Software evaluation2000-03 Strengths and Limitations of Using SUDAAN, Stata, and WesVarPC for Computing
Variances from NCES Data Sets
Staff97-42 Improving the Measurement of Staffing Resources at the School Level: The Development
of Recommendations for NCES for the Schools and Staffing Survey (SASS)98-08 The Redesign of the Schools and Staffing Survey for 1999-2000: A Position Paper
Staff higher education institutions97-26 Strategies for Improving Accuracy of Postsecondary Faculty Lists
Staff nonprofessional2000-13 Non-professional Staff in the Schools and Staffing Survey (SASS) and Common Core of
Data (CCD)
Jeffrey Owings
Jeffrey Owings
Steven Kaufman
Dan Kasprzyk
Tai PhanBeth Young
Tai Phan
Mary Rollefson
Dan KasprzykBeth Young
Dan KasprzykFrank Johnson
Lee Hoffman
Aurora D'AmicoArnold Goldstein
Ralph Lee
Mary Rollefson
Dan Kasprzyk
Linda Zimbler
Kerry Gruber
State1999-03 Evaluation of the 1996-97 Nonfiscal Common Core of Data Surveys Data Collection, Beth Young
Processing, and Editing Cycle
S7
No. Title NCES contact
Statistical methodology97-21 Statistics for Policymakers or Everything You Wanted to Know About Statistics But
Thought You Could Never Understand
Statistical standards and methodology2001-05 Using TIMSS to Analyze Correlates of Performance Variation in Mathematics
Students with disabilities95-13 Assessing Students with Disabilities and Limited English Proficiency
2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP
Survey methodology96-17 National Postsecondary Student Aid Study: 1996 Field Test Methodology Report97-15 Customer Service Survey: Common Core of Data Coordinators97-35 Design, Data Collection, Interview Administration Time, and Data Editing in the 1996
National Household Education Survey98-06 National Education Longitudinal Study of 1988 (NELS:88) Base Year through Second
Follow-Up: Final Methodology Report98-11 Beginning Postsecondary Students Longitudinal Study First Follow-up (BPS:96-98) Field
Test Report98-16 A Feasibility Study of Longitudinal Design for Schools and Staffing Survey
1999-07 Collection of Resource and Expenditure Data on the Schools and Staffing Survey1999-17 Secondary Use of the Schools and Staffing Survey Data2000-01 1999 National Study of Postsecondary Faculty (NSOPF:99) Field Test Report2000-02 Coordinating NCES Surveys: Options, Issues, Challenges, and Next Steps2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and
1999 AAPOR Meetings2000-12 Coverage Evaluation of the 1994-95 Public Elementary/Secondary School Universe
Survey2000-17 National Postsecondary Student Aid Study:2000 Field Test Methodology Report2001-04 Beginning Postsecondary Students Longitudinal Study: 1996-2001 (BPS:1996/2001)
Field Test Methodology Report2001-07 A Comparison of the National Assessment of Educational Progress (NAEP), the Third
International Mathematics and Science Study Repeat (TIMSS-R), and the Programmefor International Student Assessment (PISA)
2001-09 An Assessment of the Accuracy of CCD Data: A Comparison of 1988,1989, and 1990CCD Data with 1990-91 SASS Data
2001-11 Impact of Selected Background Variables on Students' NAEP Math Performance2001-13 The Effects of Accommodations on the Assessment of LEP Students in NAEP2001-19 The Measurement of Home Background Indicators: Cognitive Laboratory Investigations
of the Responses of Fourth and Eighth Graders to Questionnaire Items and ParentalAssessment of the Invasiveness of These Items
2002-01 Legal and Ethical Issues in the Use of Video in Education Research2002-02 Locale Codes 1987 - 20002002-03 National Postsecondary Student Aid Study, 1999-2000 (NPSAS:2000), CATI
Nonresponse Bias Analysis Report.
Teachers98-13
1999-142000-10
Response Variance in the 1994-95 Teacher Follow-up Survey1994-95 Teacher Followup Survey: Data File User's Manual, Restricted-Use CodebookA Research Agenda for the 1999-2000 Schools and Staffing Survey
Teachers instructional practices of98-08 The Redesign of the Schools and Staffing Survey for 1999-2000: A Position Paper
Teachers opinions regarding safety98-08 The Redesign of the Schools and Staffing Survey for 1999-2000: A Position Paper
Stephen BroughmanStephen BroughmanSusan WileyLinda ZimblerValena PliskoDan Kasprzyk
Beth Young
Andrew G. MalizioPaula Knepper
Arnold Goldstein
John Sietsema
Arnold GoldsteinArnold GoldsteinArnold Goldstein
Patrick GonzalesFrank JohnsonAndrew Malizio
Steven KaufmanKerry GruberDan Kasprzyk
Dan Kasprzyk
Dan Kasprzyk
Dan Kasprzyk
No. Title NCES contact
Teachers qualifications of1999-04 Measuring Teacher Qualifications Dan Kasprzyk
Teachers salaries of94-05 Cost-of-Education Differentials Across the States William J. Fowler, Jr.
Training2000-16a Lifelong Learning NCES Task Force: Final Report Volume I2000-16b Lifelong Learning NCES Task Force: Final Report Volume II
Lisa HudsonLisa Hudson
Variance estimation2000-03 Strengths and Limitations of Using SUDAAN, Stata, and WesVarPC for Computing Ralph Lee
Variances from NCES Data Sets2000-04 Selected Papers on Education Surveys: Papers Presented at the 1998 and 1999 ASA and Dan Kasprzyk
1999 AAPOR Meetings2001-18 A Study of Variance Estimation Methods Ralph Lee
Violence97-09 Status of Data on Crime and Violence in Schools: Final Report Lee Hoffman
Vocational education95-12 Rural Education Data User's Guide Samuel Peng
1999-05 Procedures Guide for Transcript Studies Dawn Nelson1999-06 1998 Revision of the Secondary School Taxonomy Dawn Nelson
5949
U.S. Department of EducationOffice of Educational Research and Improvement (OERI)
National Library of Education (NLE)Educational Resources Information Center (ERIC)
NOTICE
Reproduction Basis
EducallonalPmettes lalormation Culla
This document is covered by a signed "Reproduction Release (Blanket)"form (on file within the ERIC system), encompassing all or classes ofdocuments from its source organization and, therefore, does not require a"Specific Document" Release form.
This document is Federally-funded, or carries its own permission toreproduce, or is otherwise in the public domain and, therefore, may bereproduced by ERIC without a signed Reproduction Release form (either"Specific Document" or "Blanket").
