Upload
christa-bowen
View
214
Download
0
Tags:
Embed Size (px)
Citation preview
Development of the 2007 Census of Agriculture Mail List
Stan Hoge and Bill Iwig
National Agricultural Statistics Service
Overview
• Background
• List Development Cycle
• 2002 Census Coverage Results
• Census Coverage Goals
• Steps to Meet Established Goals
• List Building Results
• Suggestions for the 2012 CML
Background
• NASS has responsibility for an annual agricultural survey program and the Census of Agriculture every five years.
• The NASS list frame supports both programs requiring:• Adequate production coverage and accurate control data
for the survey programs.
• Adequate farm coverage for the census.
• List development is an on-going process involving adding new records and updating current records.
List Development Cycle
NewList
Sources
NewList
Sources
Annual
Survey
Program
RecordLinkage
Updates andMaintenance
Updated List Frame (active, inactive and potential farm records)
Screen forPotential Farms
CensusMailList
2002 Census Coverage Results
Category TotalPercent from
Coverage Adjustment
All Farms 2,128,982 17.90
Land in Farms (acres) 938,279,056 2.40
Farms by Value of Sales:
Less than $1,000 570,919 32.77
$1,000 to $2,499 255,639 33.92
$2,500 to $4,999 213,326 15.46
$5,000 to $9,999 223,168 11.45
2002 Census Coverage Results
Minority Group No. FarmsPercent from
Coverage Adjustment
Female 237,819 24.18
Hispanic 50,592 43.14
Am. Indian or Alaska Native 15,494 20.19
Asian 6,285 18.60
Black/African Am. 29,090 43.87
Native Hawaiian 983 18.01
More than one race 7,661 22.88
2002 Census Coverage ResultsCoverage Adjustment of Total Number of Farms
Census Coverage Goals
• Increase minority farm coverage from 68% to 73%.
• Maintain coverage of farms with $50,000 or more in agricultural sales at 95%.
• Reach at least 75% coverage of all farms for all States.
• Set a Target CML Size = 3.1 million records.
CML Size vs. Farm Coverage
89.6
88.2
86.3
82.1
4.13.6
3.22.85
78
80
82
84
86
88
90
92
1987 1992 1997 2002
Far
m C
ove
rag
e (%
) .
0
2
4
6
CM
L S
ize
(Millio
ns)
.
Steps to Meet Established Goals
• Aggressive list-building guided by annual Field Office (FO) list-building plans.
• Outreach to minority and socially disadvantaged farm operators.
• Introduction of the Agricultural Identification Survey (AIS).
• Improvements to record-linkage processing.
• Set State level target CMLs.
Census List Development by Source, 2002 and 2007
Source
New Adds
(Million)Reactivated Criteria
(Million)
1998-2002 2003-2007 1998-2002 2003-2007
FTI/FSA/InfoUSA .526 .678 .268 .370
State Sources .452 1.098 .051 .047
Total .987 1.776 .319 .417
Minority Counts (U.S.)List Control vs. Census Adjusted
Type2002
CensusJanuary
2005June 2007 % Increase
Am Indian 15,494 10,993 44,683 306
Asian 8,375 5,148 8,160 59
Black/African American
29,090 14,770 43,871 197
Female 237,819 163,755 333,435 104
More than one race
7,661 7,394 21,092 185
Native Hawaiian
983 1,481 3,384 129
Hispanic 50,592 21,914 44,518 103
Agriculture Identification Survey Results(Screener for Potential Farms)
Survey Period Total
Mailed
In-Scope
1/
Out of
Scope
Not Returned
Fall 2004 171,717 32,833 18,542 110,990
Spring 2005 102,722 31,112 17,481 34,257
Fall 2005 133,179 47,540 19,044 49,902
Spring 2006 260,663 51,260 32,734 108,053
Fall 2006 1,065,311 472,728 157,990 361,558
Total 1,733,592 635,473 245,791 664,760
1/ Response with data and available for CML
Set State-level CML Targets
• Allocation of U.S. Target CML (3.1 million).• Based on State coverage history, population
distribution and current farm number estimates.• Provided guide to FOs in managing list building.• Expect approximately 3.4 million available
records so plan to trim marginal records for some States.• Potential farm records with probable bad addresses• Records from poor quality list sources• FO Input through CML Target Evaluation and Trim Plans• Records identified by Classification And Regression Trees
Model (CART), based on percent active on survey frame.
Suggestions for the 2012 CML
• Analyze the effectiveness of available list sources.
• Analyze the effectiveness of the AIS and the entire Census list-building process.
• Provide firmer and earlier guidance to the FOs on list-building expectations.
• Increase automation of record linkage as appropriate.
END