Exploring Quarterly Agricultural Survey Questionnaire Version Reduction Scenarios

United States
Department of
Agriculture

National
Agricultural
Statistics
Service
Research and
Development Division
Washington DC 20250

Exploring Quarterly
Agricultural Survey
Questionnaire Version
Reduction Scenarios
Morgan Earp
Scott Cox
Jody McDaniel
Chadd Crouse

RDD Research Report
RDD-09-09
May 2009

This paper was prepared for limited distribution to the research community outside the United
States Department of Agriculture. The views expressed herein are not necessarily those of the
National Agricultural Statistics Service or of the United States Department of Agriculture.

EXECUTIVE SUMMARY
The United States Department of Agriculture (USDA), National Agricultural Statistics
Service’s (NASS) Agricultural Survey Program (ASP) collects national agricultural data
from farmers and ranches quarterly and annually to estimate the size of local and national
crop production and stock inventories using the Quarterly Agricultural Survey
Questionnaire (QAS). The QAS provides a basis for estimating seasonal and annual crop
production, supplies, and grain storage. The farming industry uses QAS estimates for
both short-term and long-term crop planning.
Currently the QAS collects data on a variety of crops and stocks across all 50 states
anywhere from one to four times a year. Variations in the crop data collection from state
to state are believed to reduce respondent burden (e.g., Maine farmers are not surveyed
for soybeans and Georgia farms are not surveyed for alfalfa); however, such
customization in turn creates a number of QAS versions, which is both time-consuming
and more costly to administer. Therefore, this report explores multiple scenarios for
reducing the number of QAS versions.
A table consisting of all 31 crops and stocks surveyed by each state was analyzed using
hierarchical clustering to identify possibilities for regional versions. The number of
clusters was limited to 20 in order to reduce the number of potential QAS versions. The
resulting clusters’ (regions’) crop/stock survey frequencies were determined by
comparing the crop/stock survey frequency across states within the given cluster.

i

RECOMMENDATIONS
1. There are no current plans to pursue regionalizing the QAS. However, should
regionalization be pursued in the future, this report demonstrates a useful
methodology for doing so. This approach should also be explored when
proposing regionalization of questionnaires in other surveys.

ii

Exploring Quarterly Agricultural Survey Questionnaire Version Reduction
Scenarios
Morgan Earp, Scott Cox, Jody McDaniel, & Chadd Crouse1
Abstract
The United States Department of Agriculture’s National Agricultural
Statistics Service (NASS) conducts the Agricultural Survey Program
(ASP), which consists of crop/stocks and livestock surveys. The Quarterly
Agricultural Survey (QAS) questionnaire serves as a primary data
collection instrument for the Agency’s estimates of seasonal and annual
crop production, supplies, and grain storage, which are used by the
farming industry for both short and long term planning. The QAS is
administered in all 50 states and collects data on 31 different crops and
stocks, in varying combinations and frequencies throughout the year
depending on the state. Such variation allegedly reduces respondent
burden; however, it greatly increases the complexity of the survey
administration process. Hierarchical clustering was done to investigate the
potential of creating 20 regional QAS versions. Such an approach, if
implemented operationally, would reduce the number of QAS versions by
60 percent (50 to 20). The research explored further clustering the QAS
into only five regional versions, which, if implemented, would reduce the
number of QAS versions by 90 percent (50 to 5).
Key Words: Quarterly Agricultural Survey Questionnaire; Questionnaire Version
Reduction; Item Reduction

1.

BACKGROUND

In September of 2007, the United States Department of Agriculture’s (USDA) National
Agricultural Statistics Service (NASS) created a team to improve the efficiency of the
Quarterly Agricultural Survey Program through a general review of survey content and
by reducing the number of questionnaire versions. The Quarterly Agricultural Survey
(QAS) Questionnaire Reduction and Review Team was established to improve the
efficiency of the quarterly agricultural survey program through a reduction in the number
1

Scott Cox and Jody McDaniel initiated this research while serving as the Commodity Surveys Section
Head and the Quarterly Crops/Stocks Survey Administrator with the USDA/NASS – Census and Survey
Division (CSD). Chadd Crouse provided and defined the data for this research while a mathematical
statistician with the USDA/NASS – CSD. Morgan Earp is a survey and mathematical statistician with the
USDA/NASS in its Research and Development Division (RDD), located in Room 305, 3251 Old Lee
Highway, Fairfax, VA 22030.
Jaki McCarthy provided assistance with this research while the Chief Cognitive Research Methodologist
with the USDA/NASS/RDD.

of questionnaire versions and a general review of survey content (National Agricultural
Statistics Service, 2008). The desired result of this team was to facilitate standardization
and regionalization of NASS data collection efforts and produce the following benefits:
allow maximization of resources within Data Collection Centers, promote consistency of
data collection within regions, ensure consistent standards between the Agricultural
Survey and Census Program, guarantee the quality of estimates, potentially reduce survey
cost and respondent burden, reduce processing time for headquarters units and gain
efficiency in survey training.
The team hoped to produce the following deliverable after the work was completed:
1. Draft regional Quarterly Agricultural Survey instruments that meet NASS
questionnaire design standards and improve the overall efficiency of the data
collection process. (Goal was dropped).
2.

Complete review of Quarterly Agricultural Survey instruments to determine if all
items are needed for the NASS estimation program and to determine if
efficiencies can be gained by either changing the format or number survey items
collected. (Goal was achieved).

3. Review all modes of collection (paper, Blaise, EDR) for the Quarterly
Agricultural Survey program to ensure that questions are standardized across each
mode of data collection. (Goal was achieved).
4. Ensure that all Quarterly Agricultural Survey paper instruments are 12 pages or
less to ensure that NASS standard survey mailing procedures can be utilized.
(Goal was achieved).
Each above deliverable focuses on improving the overall efficiency of the QAS. The
relative success of each deliverable affected the efficiency of the QAS program. The
collective success of all four ultimately was expected to provide the greatest opportunity
for improving efficiency of the QAS program. Deliverables two through four were
completed and are contributing towards the efficiency of the QAS program.
The last three deliverables are being used to make the QAS program more efficient;
however, deliverable one was not completed. The goal of drafting regional QAS survey
instruments was dropped after much discussion with CSD management. This was due to
the realization that the continued use of the Questionnaire Repository System (QRS)
provides many of the efficiencies that would be gained by regionalizing QAS survey
instruments. Specifically, it was determined that the benefits achieved by using the QRS
for questionnaire development outweighed potential gains of using regional
questionnaires, when it came to reducing respondent burden, in the QAS program. The
original intent of having regional questionnaire versions was to improve the overall
efficiency of the QAS program. This includes but is not limited to the following items:
reducing respondent burden, improving the quality of estimates, potentially reducing
survey cost, improving survey training, and better utilizing DCCs. However, the change

in objectives for the team did not allow completion of all steps (such as testing, etc.) that
would have been completed if the regional questionnaires had been adapted.
Furthermore, this report should be a starting point for any future discussions regarding
the regionalization of QAS questionnaires (or any other survey instruments). The main
principals discussed could probably be applied to other survey programs without any
major changes in the methodology.
2.

INTRODUCTION

This report documents options for reducing the number of QAS versions administered by
the Agricultural Survey Program (ASP). The QAS is administered in all 50 states and
collects data on 31 different crops and stocks throughout the year, in varying
combinations and frequencies depending on the state.
The QAS provides clear indications of the potential production and supply of major
commodities in the United States. NASS surveys producers on their total acres operated,
acres planted and harvested of specific commodities, and quantities of grains and oilseeds
produced and stored on-farm, in order to set national and state estimates. NASS
publishes the results of the QAS in a series of reports, including the annual acreage and
quarterly grain stocks reports. The entire agricultural community including producers,
buyers, providers, processors, state and federal agencies, and policymakers depends on
the estimates set using the QAS. Users of these estimates include commodity markets,
educational institutions, state and federal agencies, and the farming and ranching
operations themselves.
In an effort to reduce respondent burden and maintain high response rates, state-specific
versions of the QAS are used in the ASP data collection process. The content of these
QAS versions varies by state and the time of year the questionnaire is administered.
1.1

Problem

Currently, each state utilizes its own version of the QAS. Although this process is
thought to reduce respondent burden, it requires considerable resources to develop the
survey instruments, administer the survey, and summarize the data.
1.2

Purpose

This report describes the potential for regionalizing the QAS. It is merely an exploratory
summarization of current QAS survey frequencies across states, and does not account for
state preferences. The purpose of the report is to improve the efficiency of the quarterly
agricultural survey program by reducing the number of QAS versions. It provides insight
into possibilities for reducing the number of QAS versions and, thus, the associated
survey administration burden.
3.

METHOD

3.1

Data

Crop and stock survey frequencies were assessed and compared across all states using
hierarchical clustering. Appendix A shows all crops and stocks included in one or more
QAS questionnaires across the column headings. The table cells are coded by sampling
frequency, not publication frequency. The QAS surveying frequencies of crops and
stocks are coded using the Field Crops Section classifications: annual (A), full season (F),
“included” (I), all other states (AOS), silage (S) or not surveyed (NS) (Table A-1). The
notation used to classify states does not necessarily reflect the crop/stock specifically
sampled for the QAS. For the purposes of this analysis, “annual” crops/stocks are
considered to be surveyed annually, “full season” crops/stocks are considered to be
surveyed quarterly, “included” crops/stocks are special crops considered to be surveyed
at least annually2, “all other states” crops/stocks are grouped and published together with
other crops by state (survey frequency varies), “silage” corn is considered to be surveyed
annually to estimate silage (not grain), and “not surveyed” crops/stocks are not surveyed.
All six survey frequencies were generated for purposes of analysis; however, they were
consolidated as follows for purposes of summary: crops and stocks identified as annual,
included, or silage were considered annual; crops and stocks identified as full season, all
other states, or not surveyed were left as is. The analytical data set consists of 50 states
and 31 variables: Each variable represents one of the 31 crops/stocks surveyed.
3.2

Procedure

States sharing similar crops and stocks surveyed were dynamically grouped together
using a dendogram to form clusters that were ultimately referred to as regions. A
dendogram is a tree with individual elements at one end, building (agglomeratively) into
a single cluster containing every element at the other end (JMP, 2008). Dendograms may
be cut at any point to provide a specific number of clusters. For the purposes of this
report, two target numbers of clusters were specified as follows: 1) the QAS 20 regions
(versions) were identified using 20 clusters; and 2) the QAS 5 regions (versions) were
identified by consolidating all 20 original clusters into 5 clusters.
4.

RESULTS

Clustering was used to reduce the number of QAS versions by 60 percent (50 to 20) and
then ultimately by 90 percent (50 to 5). Survey administration frequency data, comprised
of indicator variables as to whether specific items are surveyed as well as the frequency
at which they are surveyed across states, were combined with state sampling data to
create regional state groupings known as clusters using JMP’s hierarchical clustering
algorithm. Using hierarchical clustering, these data were compared across states to create
clusters of like states in terms of survey administration frequency. Clusters were first
limited to 20 and then to 5 to identify regional survey version scenarios. The initial
hierarchical clustering, using the 20 cluster cutoff, revealed 20 regions (Figure 1).

2

Special crops included in the full program are not necessarily sampled quarterly; therefore, they are
classified as annual for purposes of this study given that they are surveyed at least annually.

The frequency (annually, seasonally, all other states or not surveyed) at which
crops/stocks were surveyed within each cluster was examined and used to determine the
frequency at which they would be surveyed for a given region. When the survey
frequency was tied, annual, full season, or all other survey trumped not surveyed, and
seasonal trumped annual. When survey frequency was not tied between annual, full
season, all other survey, or not surveyed, the frequency at which the majority of states
within the cluster surveyed specific crops/stocks was used (even if it meant reclassifying
crops/stocks from surveyed to not surveyed)

Cluster

State

Dendogram

One

Two
Three
Four
Five
Six
Seven

Eight
Nine
Ten

Eleven
Twelve

Thirteen
Fourteen
Fifteen

Sixteen
Seventeen
Eighteen

Nineteen
Twenty

Figure 1. State Cluster Dendogram

4.1

Twenty Region Survey Design

4.1.1 Region 1: Alabama, Mississippi, Georgia, South Carolina, North Carolina, &
Virginia
The first cluster identified Region 1, which includes six states: Alabama, Mississippi,
Georgia, South Carolina, North Carolina, and Virginia (Figure 2). Analysis of individual
crops/stocks survey frequency within Cluster 1 indicates that Region 1 would survey
oats, potatoes, sorghum, sweet potatoes, tobacco, and watermelons annually; and corn,
cotton, other hay, peanuts, soybeans, and winter wheat seasonally (Table 1).

Figure 2. Region 1 States

Table 1: Region 1 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

Crops/Stocks
Oats
Potatoes
Sorghum
Sweet Potatoes
Tobacco
Watermelons
Corn
Cotton
Other Hay
Peanuts
Soybeans
Winter Wheat

4.1.2

Region 2: Arkansas & Louisiana

The second cluster identified Region 2, which includes two states: Arkansas and
Louisiana (Figure 3). Analysis of individual crops/stocks survey frequency within
Cluster 2 indicates that Region 2 would survey alfalfa, potatoes, sweet potatoes, and
watermelons annually and corn, cotton, other hay, rice, sugar cane, sorghum, soybeans,
and winter wheat seasonally (Table 2).

Figure 3. Region 2 States

Table 2: Region 2 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

Crops/Stocks
Alfalfa
Potatoes
Sweet Potatoes
Watermelons
Corn
Cotton
Other Hay
Rice
Sugar Cane
Sorghum
Soybeans
Winter Wheat

4.1.3

Region 3: Missouri & Oklahoma

The third cluster identified Region 3, which includes two states: Missouri and Oklahoma
(Figure 4). Analysis of individual crops/stocks survey frequency within Cluster 3
indicates that Region 3 would survey forage, oats, potatoes, rye, tobacco, and
watermelons annually; alfalfa, corn, cotton, other hay, peanuts, rice, sorghum, soybeans,
and winter wheat seasonally; and canola and sunflower in unison with other crops/stocks
as the region sees fit (Table 3).

Figure 4. Region 3 States

Table 3: Region 3 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Forage
Oats
Potatoes
Rye
Tobacco
Watermelons
Alfalfa
Corn
Cotton
Other Hay
Peanuts
Rice
Sorghum
Soybeans
Winter Wheat
Canola
Sunflower

4.1.4

Region 4: California

The fourth cluster identified Region 4, which includes one state: California (Figure 5).
Analysis of individual crops/stocks survey frequency within Cluster 4 indicates that
Region 4 would survey dry edible beans, forage, garbanzo beans, potatoes, safflower,
sorghum, sweet potatoes, and watermelons annually; alfalfa, corn, cotton, durum wheat,
oats, other hay, rice, and winter wheat; and sunflower in unison with other crops/stocks
as California sees fit (Table 4).

Figure 5. Region 4 States

Table 4: Region 4 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Forage
Garbanzo Beans
Potatoes
Safflower
Sorghum
Sweet Potatoes
Watermelons
Alfalfa
Corn
Cotton
Durum Wheat
Oats
Other Hay
Rice
Winter Wheat
Sunflower

4.1.5

Region 5: Texas

The fifth cluster identified Region 5, which includes one state: Texas (Figure 6).
Analysis of individual crops/stocks survey frequency within Cluster 5 indicates that
Region 5 would survey dry edible beans, forage, potatoes, sweet potatoes, and
watermelons annually; alfalfa, corn, cotton, oats, other hay, peanuts, rice, sugar cane,
sorghum, soybeans, sunflower, and winter wheat seasonally; and rye in unison with other
crops/stocks as Texas sees fit (Table 5).

Figure 6. Region 5 States

Table 5: Region 5 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Forage
Potatoes
Sweet Potatoes
Watermelons
Alfalfa
Corn
Cotton
Oats
Other Hay
Peanuts
Rice
Sugar Cane
Sorghum
Soybeans
Sunflower
Winter Wheat
Rye

4.1.6

Region 6: Alaska & Hawaii

The sixth cluster identified Region 6, which includes two states: Alaska and Hawaii
(Figure 7). Analysis of individual crops/stocks survey frequency within Cluster 6
indicates that Region 6 would survey oats annually and sugar cane seasonally (Table 6).

Figure 7. Region 6 States

Table 6: Region 6 Crops/Stocks Survey Frequency
Survey Frequency

Annual
Seasonal

Crops/Stocks
Oats
Sugar Cane

4.1.7

Region 7: Connecticut, Maine, Massachusetts, Nevada, New Hampshire, Rhode
Island, & Vermont

The seventh cluster identified Region 7, which includes seven states: Connecticut, Maine,
Massachusetts, Nevada, New Hampshire, Rhode Island, and Vermont (Figure 8).
Analysis of individual crops/stocks survey frequency within Cluster 7 indicates that
Region 7 would survey alfalfa, corn (silage), other hay, and potatoes annually (Table 7).

Figure 8. Region 7 States

Table 7: Region 7 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Crops/Stocks
Alfalfa
Corn (Silage)
Other Hay
Potatoes

4.1.8

Region 8: Arizona & Florida

The eighth cluster identified Region 8, which includes two states: Arizona and Florida
(Figure 9). Analysis of individual crops/stocks survey frequency within Cluster 8
indicates that Region 8 would survey corn, other hay, potatoes, sorghum, soybeans,
tobacco, watermelons, and winter wheat annually; alfalfa, cotton, durum wheat, peanuts,
and sugarcane seasonally; and safflower in unison with other crops/stocks as the region
sees fit (Table 8).

Figure 9. Region 8 States

Table 8: Region 8 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Corn
Other Hay
Potatoes
Sorghum
Soybeans
Tobacco
Watermelons
Winter Wheat

Seasonal

All Other States

Alfalfa
Cotton
Durum Wheat
Peanuts
Scane
Safflower

4.1.9

Region 9: New Mexico

The ninth cluster identified Region 9, which includes one state: New Mexico (Figure 10).
Analysis of individual crops/stocks survey frequency within Cluster 9 indicates that
Region 9 would survey dry edible beans, forage, other hay, potatoes, and winter wheat
annually; and alfalfa, corn, cotton, peanuts, and sorghum seasonally (Table 9).

Figure 10. Region 9 States

Table 9: Region 9 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

Crops/Stocks
Dry Edible Beans
Forage
Other Hay
Potatoes
Winter Wheat
Alfalfa
Corn
Cotton
Peanuts
Sorghum

4.1.10 Region 10: Delaware, Maryland, & New Jersey
The tenth cluster identified Region 10, which includes three states: Delaware, Maryland,
and New Jersey (Figure 11). Analysis of individual crops/stocks survey frequency
within Cluster 10 indicates that Region 10 would survey alfalfa, other hay, potatoes, and
watermelons annually; and corn, soybeans, and winter wheat seasonally (Table 10).

Figure 11. Region 10 States

Table 10: Region 10 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

Crops/Stocks
Alfalfa
Other Hay
Potatoes
Watermelons
Corn
Soybeans
Winter Wheat

4.1.11 Region 11: Indiana & Iowa
The eleventh cluster identified Region 11, which includes two states: Indiana and Iowa
(Figure 12). Analysis of individual crops/stocks survey frequency within Cluster 11
indicates that Region 11 would survey forage and watermelons annually; and alfalfa,
corn, other hay, soybeans, and winter wheat seasonally (Table 11).

Figure 12. Region 11 States

Table 11: Region 11 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Forage
Watermelons

Seasonal

Alfalfa
Corn
Other Hay
Soybeans
Winter Wheat

4.1.12 Region 12: Kentucky, Tennessee, & West Virginia
The twelfth cluster identified Region 12, which includes three states: Kentucky,
Tennessee, and West Virginia (Figure 13). Analysis of individual crops/stocks survey
frequency within Cluster 12 indicates that Region 12 would survey alfalfa, sorghum, and
tobacco annually; and corn, other hay, soybeans, and winter wheat seasonally (Table 12).

Figure 13. Region 12 States

Table 12: Region 12 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Alfalfa
Sorghum
Tobacco

Seasonal

Corn
Other Hay
Soybeans
Winter Wheat

4.1.13 Region 13: Colorado
The thirteenth cluster identified Region 13, which includes one state: Colorado (Figure
14). Analysis of individual crops/stocks survey frequency within Cluster 13 indicates
that Region 13 would survey dry edible beans, oats, potatoes, proso millet, and spring
wheat annually; alfalfa, corn, other hay, sorghum, sunflower, and winter wheat
seasonally; and canola and sunflower in unison with other crops/stocks /stocks as
Colorado sees fit (Table 13).

Figure 14. Region 13 States

Table 13: Region 13 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Oats
Potatoes
Proso Millet
Spring Wheat
Alfalfa
Corn
Other Hay
Sorghum
Sunflower
Winter Wheat
Canola
Safflower

4.1.14 Region 14: Utah & Wyoming
The fourteenth cluster identified Region 14, which includes two states: Utah and
Wyoming (Figure 15). Analysis of individual crops/stocks survey frequency within
Cluster 14 indicates that Region 14 would survey corn, dry edible beans, oats, spring
wheat, and winter wheat annually; alfalfa and other hay seasonally; and safflower and
sunflower in unison with other crops/stocks as the region sees fit (Table 14).

Figure 15. Region 14 States

Table 14: Region 14 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Crops/Stocks
Corn
Dry Edible Beans
Oats
Spring Wheat
Winter Wheat

Seasonal

Alfalfa
Other Hay

All Other States

Safflower
Sunflower

4.1.15 Region 15: Idaho, Montana, & Oregon
The fifteenth cluster identified Region 15, which includes three states: Idaho, Montana,
and Oregon (Figure 16). Analysis of individual crops/stocks survey frequency within
Cluster 15 indicates that Region 15 would survey Austrian winter peas, corn, dry edible
beans, dry edible peas, garbanzo beans, lentils, and potatoes annually; alfalfa, durum
wheat, oats, other hay, spring wheat, and winter wheat seasonally; and canola, rapeseed,
and mustard seed in unison with other crops/stocks as the region sees fit (Table 15).

Figure 16. Region 15 States

Table 15: Region 15 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Austrian Winter Peas
Corn
Dry Edible Beans
Dry Edible Peas
Garbanzo Beans
Lentils
Potatoes
Alfalfa
Durum Wheat
Oats
Other Hay
Spring Wheat
Winter Wheat
Canola
Rapeseed
Mustard Seed

4.1.16 Region 16: Washington
The sixteenth cluster identified Region 16, which includes one state: Washington (Figure
17). Analysis of individual crops/stocks survey frequency within Cluster 16 indicates
that Region 16 would survey dry edible beans, dry edible peas, forage, garbanzo beans,
lentils, oats, and potatoes annually; alfalfa, corn, other hay, spring wheat, and winter
wheat seasonally; and canola and mustard seed in unison with other crops/stocks as
Washington sees fit (Table 16).

Figure 17. Region 16 States

Table 16: Region 16 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Dry Edible Beans
Dry Edible Peas
Forage
Garbanzo Beans
Lentils
Oats
Potatoes

Seasonal

All Other States

Alfalfa
Corn
Other Hay
Spring Wheat
Winter Wheat
Canola
Mustard Seed

4.1.17 Region 17: Minnesota & North Dakota
The seventeenth cluster identified Region 17, which includes two states: Minnesota and
North Dakota (Figure 18). Analysis of individual crops/stocks survey frequency within
Cluster 17 indicates that Region 17 would survey dry edible beans, dry edible peas,
flaxseed, forage, garbanzo beans, lentils, potatoes, and winter wheat annually; alfalfa,
canola, corn, durum wheat, oats, other hay, soybeans, spring wheat, and sunflower
seasonally; and rapeseed, mustard seed, rye, safflower in unison with other crops/stocks
as the region sees fit (Table 17).

Figure 18. Region 17 States

Table 17: Region 17 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Dry Edible Peas
Flaxseed
Forage
Garbanzo Beans
Lentils
Potatoes
Winter Wheat
Alfalfa
Canola
Corn
Durum Wheat
Oats
Other Hay
Soybeans
Spring Wheat
Sunflower
Rapeseed
Mustard Seed
Rye
Safflower

4.1.18 Region 18: Illinois, Michigan, New York, Ohio, Pennsylvania, & Wisconsin
The eighteenth cluster identified Region 18, which includes six states: Illinois, Michigan,
New York, Ohio, Pennsylvania, and Wisconsin (Figure 19). Analysis of individual
crops/stocks survey frequency within Cluster 18 indicates that Region 18 would survey
forage and potatoes annually; alfalfa, corn, oats, other hay, rye, soybeans, sunflower, and
winter wheat seasonally; and rye and sunflower in unison with other crops/stocks as the
region sees fit (Table 18).

Figure 19. Region 18 States

Table 18: Region 18 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Forage
Potatoes

Seasonal

All Other States

Alfalfa
Corn
Oats
Other Hay
Soybeans
Winter Wheat
Rye
Sunflower

4.1.19 Region 19: Kansas & Nebraska
The nineteenth cluster identified Region 19, which includes two states: Kansas and
Nebraska (Figure 20). Analysis of individual crops/stocks survey frequency within
Cluster 19 indicates that Region 19 would survey dry edible beans, forage, garbanzo
beans, lentils, potatoes, and proso millet annually; alfalfa, corn, cotton, oats, other hay,
sorghum, soybeans, sunflower, and winter wheat seasonally; and canola and rye in unison
with other crops/stocks as the region sees fit (Table 19).

Figure 20. Region 19 States

Table 19: Region 19 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Forage
Garbanzo Beans
Lentils
Potatoes
Proso Millet
Alfalfa
Corn
Cotton
Oats
Other Hay
Sorghum
Soybeans
Sunflower
Winter Wheat
Canola
Rye

4.1.20 Region 20: South Dakota
The twentieth cluster identified Region 20, which includes one state: South Dakota
(Figure 21). Analysis of individual crops/stocks survey frequency within Cluster 20
indicates that Region 20 would survey dry edible beans, durum wheat, flaxseed, forage,
garbanzo beans, and proso millet annually; alfalfa, corn, oats, other hay, sorghum,
soybeans, spring wheat, sunflower, and winter wheat seasonally; and rye and safflower in
unison with other crops/stocks as South Dakota sees fit (Table 20).

Figure 21. Region 20 States

Table 20: Region 20 Crops/Stocks Survey Frequency
Survey Frequency

Annual

Seasonal

All Other States

Crops/Stocks
Dry Edible Beans
Durum Wheat
Flaxseed
Forage
Garbanzo Beans
Proso Millet
Alfalfa
Corn
Oats
Other Hay
Sorghum
Soybeans
Spring Wheat
Sunflower
Winter Wheat
Rye
Safflower

4.2

Five Region Survey Design

Hierarchical clustering was also done using five clusters, which created five QAS
regions, ultimately reducing the number of versions by 90 percent (from 50 to 5).
3.2.1 Region One: Alabama, Arkansas, California, Georgia, Louisiana, Mississippi,
Missouri, North Carolina, Oklahoma, South Carolina, Texas, & Virginia
Clusters 1 through 5 were merged to identify the first consolidated region, Region 1,
which includes twelve states: Alabama, Arkansas, California, Georgia, Louisiana,
Mississippi, Missouri, North Carolina, Oklahoma, South Carolina, Texas, and Virginia
(Figure 22 & 23). Analysis of individual crops/stocks survey frequency across clusters 1
through 5 indicates that Region 1 would survey oats, sorghum, potatoes, sweet potatoes,
and watermelons annually; and alfalfa, corn, cotton, other hay, peanuts, rice, soybeans,
and winter wheat seasonally (Table 21).

Figure 22. Region 1 States

Table 21: Region 1 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Oats
Sorghum
Potatoes
Sweet Potatoes
Watermelons

Seasonal

Alfalfa
Corn
Cotton
Other Hay
Peanuts
Rice
Soybeans
Winter Wheat

Cluster

State

Dendogram

One

Two
Three
Four
Five

Figure 23. Five Region Dendogram: Region 1 States

4.2.2 Region Two: Alaska, Connecticut, Hawaii, Maine, Massachusetts, Nevada, New
Hampshire, Rhode Island, & Vermont
Clusters 6 and 7 were merged to identify the second consolidated region, Region 2, which
includes nine states: Alaska, Connecticut, Hawaii, Maine, Massachusetts, Nevada, New
Hampshire, Rhode Island, and Vermont (Figure 24 & 25). Analysis of individual
crops/stocks survey frequency across clusters 1 through 5 indicates that Region 2 would
survey alfalfa, corn (silage), and other hay annually (Table 23).

Figure 24. Region 2 States

Table 22: Region 2 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Alfalfa
Other Hay
Corn (Silage)

Cluster

State

Dendogram

Six
Seven

Figure 25. Five Region Dendogram: Region 2 States

4.2.3 Region Three: Arizona, Florida, New Mexico, Delaware, Maryland, New Jersey,
Indiana, Iowa, Kentucky, Tennessee, West Virginia, Colorado, Utah, & Wyoming
Clusters 8 through 14 were merged to identify the third consolidated region, Region 3,
which includes fourteen states: Arizona, Florida, New Mexico, Delaware, Maryland,
New Jersey, Indiana, Iowa, Kentucky, Tennessee, West Virginia, Colorado, Utah, and
Wyoming (Figure 26 & 27). Analysis of individual crops/stocks survey frequency across
Clusters 8 through 14 indicates that Region 3 would survey potatoes and winter wheat
annually; and alfalfa, corn, other hay, and soybeans seasonally (Table 23).

Figure 26. Region 3 States

Table 23: Region 3 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Potatoes
Winter Wheat
Alfalfa
Corn
Other Hay
Soybeans

Seasonal

Cluster

State

Dendogram

Eight
Nine
Ten

Eleven
Twelve

Thirteen
Fourteen

Figure 27. Five Region Dendogram: Region 3 States

4.2.4 Region Four: Idaho, Oregon, Montana, Washington, Minnesota, & North Dakota
Clusters 15 through 17 were merged to identify the fourth consolidated region, Region 4,
which includes six states: Idaho, Oregon, Montana, Washington, Minnesota, and North
Dakota (Figure 28 & 29). Analysis of individual crops/stocks survey frequency across
Clusters 15 through 17 indicates that Region 4 would survey Australian winter peas, dry
edible beans, dry edible peas, flaxseed, forage, garbanzo beans, lentils, and potatoes
annually; alfalfa, canola, corn, oats, other hay, spring wheat, and winter wheat seasonally;
and canola, rapeseed, and mustard seed in unison with other crops/stocks as the region
sees fit (Table 24).

Figure 28. Region 4 States

Table 24: Region 4 Crops/Stocks Survey Frequency
Survey Frequency

Australian Winter Peas
Dry Edible Beans
Dry Edible Peas
Flaxseed
Forage
Garbanzo Beans
Lentils
Potatoes

Annual

Alfalfa
Canola
Corn
Oats
Other Hay
Spring Wheat
Winter Wheat

Seasonal

All Other Surveys

Cluster

Crops/Stocks

State

Canola
Rapeseed
Mustard Seed

Dendogram

Fifteen

Sixteen
Seventeen

Figure 29. Five Region Dendogram: Region 4 States

4.2.5 Region Five: Illinois, Wisconsin, Michigan, New York, Ohio, Pennsylvania,
Kansas, Nebraska, & South Dakota
Clusters 18 through 20 were merged to identify the fifth consolidated region, Region 5,
which includes nine states: Illinois, Wisconsin, Michigan, New York, Ohio,
Pennsylvania, Kansas, Nebraska, and South Dakota (Figure 30 & 31). Analysis of
individual crops/stocks survey frequency across Clusters 18 through 20 indicates that
Region 5 would survey dry edible beans, forage, and potatoes annually; alfalfa, corn,
oats, other hay, soybeans, and winter wheat seasonally; and rye in unison with other
crops/stocks as the region sees fit (Table 24).

Figure 30. Region 5 States

Table 25: Region 5 Crops/Stocks Survey Frequency
Survey Frequency

Crops/Stocks

Annual

Dry Edible Beans
Forage
Potatoes

Seasonal

Alfalfa
Corn
Oats
Other Hay
Soybeans
Winter Wheat

All Other Surveys

Rye

Cluster

State

Dendogram

Eighteen

Nineteen
Twenty

Figure 31. Five Region Dendogram: Region 5 States

5.

DISCUSSION

The QAS is used to collect data on up to 31 crops and stocks anywhere from one to four
times a year, depending on the state. Since no two states are identical in their
questionnaire content, national data collection and summary is highly complex. It
appears that the survey administration process could be greatly simplified by
regionalizing the QAS. Creating 20 QAS regions reduces the amount of questionnaire
versions by 60 percent, helping standardize questionnaire administration and summaries,
and ultimately allowing states to share in printing costs and data editing tools. Creating
five QAS regions further reduces the number of questionnaire versions by 90 percent,
further standardizing questionnaire administration and summary code, and allowing
numerous states to share in printing costs and data editing tools. Regionalizing the QAS
also reduces survey administration burden and helps simplify the process for setting
national estimates.
6.

LIMITATIONS

Obviously, by reducing the number of QAS versions, states lose some autonomy and
respondent burden may be slightly increased. It is expected that the respondent burden
will be minimal using the 20-region design and only slightly increased by using the 5region design. However, respondent burden should be assessed by each state
individually. States need to compare the number of items as well as the frequency at
which they are asked before and after regionalization.
7.

RECOMMENDATIONS
There are no current plans to pursue regionalizing the QAS. However, should
regionalization be pursued in the future, this report demonstrates a useful
methodology for doing so. This approach should also be explored when
proposing regionalization of questionnaires for other surveys.

8.

REFRENCES

JMP. (2007). Clustering. Statistics and Graphics Guide . SAS Institute Inc.
National Agricultural Statistics Service. (2008, January 15). Charter for the Quarterly
Agricultural Survey Questionnaire Reduction and Review Team. Retrieved from nassnet:
http://nassnet/library/info/teams/qas/index_qas_response.html

Table A-1: Crops/Stocks Questionnaire Reduction Team Data

7.

APPENDIX A

3

3

Note: A = Annual, F = Full Season, I = “Included”, AOS = All Other States, S = Silage, and NS = Not
Surveyed (classifications are based on sampling frequency, not on publication frequency).

Table A-1: Crops/Stocks Questionnaire Reduction Team Data (Continued)