NISS Inaugurates Cross-Sector
Research Program
T
he National Institute of Statistical Sciences (NISS) has semicontinuous nature of agricultural data. There are also questions
established a cross-sector research in residence program in about the validity of the method when the prediction models under-
partnership with the National Agricultural Statistics Service lying its imputation fail.
(NASS), the survey and estimation arm of the U.S. Department of New Design and Estimation Methodologies for Biased Self-
Agriculture. The program will comprise five-person teams to Exclusion (Under-Coverage): Estimation of Small Farms from
address research problems of importance to NASS. Each team will Census Mail List NASS accounts for the incompleteness of its
consist of a faculty researcher in statistics, a NASS researcher, a Census Mail List (CML) by adjusting the weights of respondents
NISS mentor, a postdoctoral fellow, and a graduate student who to capture the estimated number of farms identified that are not on
will work together at NISS during the two consecutive summers. the list. When the 2007 Census was processed, NASS also identi-
The cross-sector research in residence program typifies the fied several valid farms not found in the area-frame, even
collaborations NISS has created in recent years, assem- though they were located in sampled area segments.
bling researchers from academia, government, and This poses the question of how many farms are
industry to work on projects in statistical research. missed by both CML and area frames. The chal-
One recent NISS project, Digital Government lenge is to develop statistical procedures to mea-
II, was a cross-disciplinary project addressing data sure the number of farms missing from both
confidentiality, data quality, and data integra- frames and to incorporate these into census
tion. Another project, Collaborative Research: weights. Cognitive issues also may need to be
Acquiring Accurate Dynamic Field Data Using addressed, as many qualifying small farms do
Lightweight Instrumentation, was conducted in not necessarily consider themselves farms and
collaboration with software engineers from Georgia hence fail to return the survey forms.
Tech, the University of Maryland, Vanderbilt New Statistical Editing and Imputation
University, and the University of Washington. Current Methods That Preserve Data Quality: Quarterly
NISS collaborators include the National Center for Agricultural Survey NASS uses data-cleaning pro-
Education Statistics, National Center for Health cedures in many of its surveys that are based on an
Statistics, Eli Lilly and Company, Merck, and the expert opinion/analysis review process and manual
Hamner Institutes for Health Sciences. intervention to correct identified data values out-
In 2006, NISS initiated the New Researcher side normally expected ranges. This manual process
Fellows Program. This program gives early career is time-consuming and can be inconsistent. It can
researchers an opportunity to join ongoing NISS projects in such lead to effects that are not reflected in the measurement error pro-
disciplines as bioinformatics, data confidentiality, chemoinformat- cess. The objective is to create automated statistical/selective editing
ics, education statistics, and software engineering. and imputation strategies that will reduce the nonsampling errors
and lower survey costs by reducing the extensive staff resources cur-
NISS Opens Search for Program Participants
rently used in data cleaning.
NISS invites faculty members, postdoctoral candidates, and graduate Statistical Multi-Source Predictive Models and Error
students to apply for positions in the cross-sector research in residence Estimates: Major USDA Crop Protection Forecasts and
program. Faculty and graduate student participants will receive finan- Estimates The USDA produces multiple forecasts of crop protec-
cial support under the program. Postdoctoral fellows will be appoint- tion throughout the growing season and estimates production at the
ed by NISS, but will spend significant time at NASS. end-of-season or after harvest. Official forecasts and estimates are
The first projects will be initiated in the summer of 2009 and derived from multiple surveys and administrative/auxiliary informa-
focus on advances in statistical methodology for USDA surveys. tion, including weather and remotely sensed data. Information is
Four candidates for the initial projects have been identified: collected from multiple sources and then synthesized by a panel of
Multivariate Imputation Mechanisms and Valid Mean- experts in USDA’s Agricultural Statistics Board (ASB), resulting in
Squared Error Estimation: Agricultural Resource Management the published official forecasts/estimates. These forecasts are com-
Survey – Phase III One objective of the Agricultural Resource pared to the use of the crops and assessed for accuracy. Later, when
Management Survey – Phase III is to allow statisticians and econo- the actual yields are known, can improvements be made to this pro-
mists to conduct multivariate statistical analyses of the farm econ- cess via increased use of modeling or through other approaches?
omy that produce valid estimates for the potential error in model How can these models or other techniques be validated during the
estimates and forecasts. NASS has been using a univariate approach short time period analysts have to review the inputs and publish the
to both imputation and mean-squared error estimates. Multivariate time-sensitive official estimates?
approaches stimulate multiple estimates and forecasts for multiple For more information about the NASS/NISS cross-sector
crops. Development of a multiple-imputation scheme will have to research in residence program or to apply for one of the positions,
handle the complexities associated with heterogeneous data and the go to
www.niss.org. n
DECEMBER 2008 AMSTAT NEWS 9
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82