Tech Paper 2011 Sampling

Technical Paper 2011-1

SAMPLING METHODOLOGY
FOR
FEDERAL WAGE SYSTEM
WAGE SURVEYS

Civilian Personnel Management Service
Wage and Salary Division

1 March 2011

Supercedes Technical Paper 91-1

CONTENTS

I. INTRODUCTION
A. Scope
B. Wage Area
C. Types of Surveys
D. Bureau of Labor Statistics Function
E. Purpose

1
1
2
2
2

II. SAMPLING METHODOLOGY
A. Source of Universe
B. Universe Definition
1. Location
2. Industry
3. Number of Employees
C. Sampling Process
1. Definition of certainty units
2. Definition of additional industries
3. Determination of sample size
4. Stratification
5. Allocation of the sample over the strata
6. Measure of size
7. Sample selection
8. Computation of the weight
9. Establishment Survey List

3
3
3
4
6
6
7
7
7
7
8
8
9
9
10

III. WEIGHT ADJUSTMENTS FOR NON-RESPONSE
A. General
B. Noncertainty Units
C. Certainty Units
D. Unique Units
E. Combined Returns
1. Recognizable Units
2. Hidden Units

11
11
13
13
14
14
14

APPENDIX A
Appropriated Fund NAICS Groupings
Non-Appropriated Fund CT NAICS Groupings
Non-Appropriated Fund AS/PS NAICS Groupings
APPENDIX B
Allocation of Establishments to Strata

I. INTRODUCTION

A. Scope
The Federal Government uses the Federal Wage System (FWS) to fix and adjust the
rates of pay for prevailing rate employees. These employees are in trade, craft, and
unskilled labor type jobs, more commonly referred to as blue collar. In 1968, the Civil
Service Commission (now the Office of Personnel Management (OPM)) issued the basic
policies of the FWS as Chapter 532 of the Federal Personnel Manual (FPM). The FWS
provides common policies, systems, practices, and job grading standards for uniform
application by all agencies subject to section 5342 of Title 5, United States Code. The
system covers most trade, craft, and unskilled labor employees in the executive branch of
the federal government who are paid from appropriated funds (ApF). In 1972, the system
was extended by Public Law 92-392 to cover Nonappropriated Fund (NAF) craft and
trade employees. To assure equitable treatment for all Department of Defense (DoD)
NAF hourly paid employees, the principles of the law pertaining to the fixing and
adjusting of pay rates were extended by DoD to hourly paid employees engaged in
Administrative Support (AS) and Patron Services (PS) positions. Such positions as sales
clerks, secretaries, display assistants, and cashiers fall into this group. OPM issued
regulation title 5 Code of Federal Regulations (CFR) Part 532, dated May 11, 1981, to
fulfill its responsibility for the overall administration of the Prevailing Rate System under
sections 5342 and 5346 of Title 5, United States Code. OPM determines the basic
policies, practices, and procedures with the advice of the Federal Prevailing Rate
Advisory Committee, other executive agencies, and labor organizations.
B. Wage Area
Since rates of pay are based on prevailing rates in a locality, OPM has defined
geographical areas, referred to as wage areas, for which rates of pay are set. The wage
areas are defined in terms of state, county (or parish), and town boundaries. There are 132
ApF wage areas and 121 NAF wage areas. The NAF wage areas are defined separately
from the ApF areas. For each wage area, OPM designates an agency as the lead agency
responsible for fixing and administering rates of pay. This designation is generally given
to the agency which has the largest number of employees covered by the FWS in the wage
area. Prior to 1995, the Department of Veterans Affairs was the lead agency for 25 wage
areas and the National Aeronautics and Space Administration was the lead agency for one
wage area. In 1995 OPM designated the Department of Defense (DoD) as the lead agency
for all ApF wage areas and for all NAF wage areas. Within DoD, the Civilian Personnel
Management Service, Wage and Salary Division (WAGE) executes this authority with the
advice of the DoD Wage Committee.

C. Types of Surveys
The lead agency conducts wage surveys of private enterprise establishments in their
respective wage areas annually to determine the level of prevailing wages paid for survey
jobs that are representative by occupation or grade level of local Federal Government
employment. The scope of the survey is defined to include establishments in certain
industries with a minimum number of employees. Two types of wage surveys are
conducted on an alternating cycle. A full scale survey involves the listing of
establishments in the scope of the survey for the wage area and collecting pay data by
personally visiting the establishments. A wage change survey is conducted the second
year in a wage area including only those establishments that participated in the previous
full scale survey. Wage change data may be collected by telephone, mail, or personal
visit. Thus, in any one year, about half of the wage areas have a full scale survey. ApF
surveys are conducted in 132 areas. NAF surveys are conducted in 121 areas. In universe
survey wage areas, all in-scope establishments are included in the survey. In sample
survey wage areas, a sample of the in-scope establishments is statistically selected to be in
the survey since including all establishments in the survey would be too costly and require
too much time. Sample surveys are conducted in more than half of the wage areas.
D. Bureau of Labor Statistics Function
In the beginning of the FWS, the Bureau of Labor Statistics (BLS) of the Department
of Labor had the responsibility to furnish the lead agencies with the list of establishments
to be included in the wage survey for each survey area. The list contains the name,
address, point of contact, number of employees, and main industry of the establishments.
The list also has a numerical weight for each establishment on the list, which shows the
numerical value of the establishment in representing establishments not selected to be on
the list. In a universe survey, all initial weights are one since all in-scope establishments
are on the list. After the close of the survey data collection, BLS adjusted the
establishment weights for establishment non-response. BLS used statistical sampling
methodology to select establishments for wage areas having sample surveys and used a
post-survey weighting procedure to adjust weights for non-response in both universe and
sample surveys. In April 1989, OPM transferred these responsibilities from BLS to the
Department of Defense Wage Fixing Authority Technical Staff (DODWFATS). In 1994
DODWFATS became the Civilian Personnel Management Service, Wage and Salary
Division (WAGE).
E. Purpose
The purpose of this technical paper is to describe the following methodologies used by
WAGE:
•

Generating sample area establishment lists

•

Adjusting weights for non-response

2

II. SAMPLING METHODOLOGY
A. Source of Universe
When BLS started providing lists of establishments to the lead agencies for their
surveys, the information was obtained from state unemployment insurance agencies and
supplemented by data from trade associations, trade directories, and labor organizations.
In 1983, BLS had to change the source of the information from state unemployment
insurance agencies because of the Privacy Act and other confidentiality concerns. Dun
and Bradstreet was chosen as the source for the establishment lists, since it maintains
records on all business establishments in the United States. Each record contains the
name, address, number of employees, and the primary industry of the company. The
industry was identified by the Standard Industrial Classification (SIC) code. Beginning in
2005, the SIC code was replaced by the North American Industry Classification System
(NAICS) code. In April 1989, the DODWFATS officially took over this BLS function.
The computer programs used to produce the list of establishments for wage surveys were
obtained from BLS. The establishment records are updated annually with new
information from Dun and Bradstreet. WAGE maintains a database of in-scope
establishments for all wage areas. The database is updated by wage specialists with
information collected on survey as well as by updates from Dun and Bradstreet.
B. Universe Definition
Three factors determine if an establishment is in the definition of the universe for a
wage area: the physical location of the establishment; the primary industry of the
establishment; and the number of employees in the establishment.
1. Location: A wage area consists of two parts, the survey area and the area of
application, which are defined geographically in terms of counties, parishes, or
independent cities. In New England states, for ApF wage areas, definitions may consist of
entire townships or cities. Survey data are collected from in-scope establishments located
in the survey area. The area of application includes the survey area plus that part of the
wage area from which no data are collected, but to which the wage schedule applies. In
Washington, D.C., for example, the ApF survey area includes the city of Washington,
D.C.; the independent cities of Alexandria, Fairfax, Falls Church, Manassas, and
Manassas Park in Virginia; the counties of Charles, Frederick, Montgomery, and Prince
George’s in Maryland; and the counties of Arlington, Fairfax, Loudoun, and Prince
William in Virginia. The area of application includes all the survey area plus the counties
of Calvert and St. Mary’s in Maryland; the city of Fredericksburg in Virginia; the counties
of Clark, Facquier, King George, Spotsylvania, Stafford, and Warren in Virginia; and the
county of Jefferson in West Virginia. NAF wage areas are usually smaller than ApF wage
areas. For example, the NAF wage area for Washington, D.C. is defined as the survey
area being the city of Washington, D.C. and the area of application being the same as the
survey area.

3

2. Industry: Establishments engaged in the following industries are included in all
ApF wage surveys for regular wage schedules as defined in title 5 CFR Part 532.
The industries are identified by 2007 NAICS codes:
311 through 339
(except 323)
221
481
482
484
485 (except 4853)
487 (except 4872)
488 (except 4883
and 4884)
492
493
515
517
5621
5622
423
424

All manufacturing classes except printing and related support
activities (NAICS 323)
Utilities
Air transportation
Rail transportation
Truck transportation
Transit and ground passenger transportation except taxi and
limousine service (NAICS 4853)
Scenic and sightseeing transportation except scenic and
sightseeing transportation, water (NAICS 4872)
Support activities for transportation except support activities for
water transportation (NAICS 4883) and support activities for road
transportation (NAICS 4884)
Couriers and messengers
Warehousing and storage
Broadcasting (except Internet)
Telecommunications
Waste collection
Waste Treatment and Disposal
Merchant wholesalers, durable goods
Merchant wholesalers, nondurable goods

Other industries may be added to the standard industry coverage when those industries
account for significant proportions of local area employment or when it is expected that
sizeable numbers of job matches might be found. For example, NAICS 212210, iron ore
mining, is added to the scope of the wage survey in Northwestern Michigan and NAICS
622110, General Medical and Surgical Hospitals, is added to the scope in many wage
surveys.

4

Establishments engaged in the following industries are included in all NAF wage surveys
for regular wage schedules as defined in title 5 CFR Part 532. The industries are
identified by 2007 NAICS codes:
42312
4232
42362
42369
42371
42391
42399
4241
42421
4243
42445
4247
4249
44132
44311
44411
44611
4471
44814
4521
45299
45321
4542
71391
71395
72111
7221
7222
7224

Motor Vehicle Supplies and New Parts Merchant Wholesalers
Furniture and Home Furnishing Merchant Wholesalers
Electrical and Electronic Appliance, Television, and Radio Set Merchant
Wholesalers
Other Electronic Parts and Equipment Merchant Wholesalers
Hardware Merchant Wholesalers
Sporting and Recreational Goods and Supplies Merchant Wholesalers
Other Miscellaneous Durable Goods Merchant Wholesalers
Paper and Paper Product Merchant Wholesalers
Drugs and Druggists’ Sundries Merchant Wholesalers
Apparel, Piece Goods, and Notions Merchant Wholesalers
Confectionary Merchant Wholesalers
Petroleum and Petroleum Products Merchant Wholesalers
Miscellaneous Nondurable Goods Merchant Wholesalers
Tire Dealers
Appliance, Television, and Other Electronics Stores
Home Centers
Pharmacies and Drug Stores
Gasoline Stations
Family Clothing Stores
Department Stores
All Other General Merchandise Stores
Office Supplies and Stationery Stores
Vending Machine Operators
Golf Courses and Country Clubs
Bowling Centers
Hotels (except Casino Hotels) and Motels
Full-service Restaurants
Limited-service Eating Places
Drinking Places (Alcoholic Beverages)

In addition to the above industries defined in title 5 CFR Part 532, on-base food
contractors (NAICS 561320) are included in NAF wage surveys.

5

The following industries are included in all NAF surveys for the AS and PS schedules as
defined by WAGE. The industries are identified by 2007 NAICS codes:
Finance and Insurance
522110
522120
522130
522210
524113
524114
524126

Commercial Banking
Savings Institutions
Credit Unions
Credit Card Issuing
Direct Life Insurance Carriers
Direct Health and Medical Insurance Carriers
Direct Property and Casualty Insurance Carriers

3. Number of Employees: The minimum number of employees an establishment must
have to be in the universe for an ApF wage survey is 50. The Alaska wage survey is an
exception with a minimum number of 20. For NAF wage surveys, the minimum number
is 20 employees except for establishments in NAICS codes 441320, 447110, 447190,
454210, 713910, and 713950, for which the minimum number is eight. For the NAF
AS/PS schedules, the minimum number of employees is also 20.
C. Sampling Process
The sampling design used in the wage surveys is sampling proportionate to
establishment employment. The universe is divided into groups based on the industry in
which the establishment is engaged. These groups are in turn divided into subgroups
based on the establishment employment. These subgroups are called strata. Within each
stratum, a sample of establishments is selected independently from the other strata.
Within each stratum, a sample of establishments is selected independently from the other
strata. The number of selected establishments is allocated over the strata based on total
stratum employment. The basic steps for the selection of sample establishments are:
•

Definition of certainty units, if any

•

Definition of additional industries to be in the scope of the universe for the wage
survey, such as General Medical and Surgical Hospitals, NAICS 622110

•

Determination of sample size

•

Stratification of the universe by industry and employment size

•

Allocation of the sample over the strata based on the number of employees

•

Computation of the measure of size for each establishment to determine its
probability of selection

•

Selection of establishments within each stratum

•

Computation of the weight for each selected unit

•

Listing of the selected establishments

6

1. Definition of certainty units: title 5 CFR Part 532 permits the inclusion of
individual establishments in the survey with certainty where such inclusion is considered
highly important and the establishment is within scope for the survey area. The automated
system allows definition of establishments individually by an identification number and
also collectively by industry grouping using the NAICS code. For example, on-base food
service contractors are included with certainty for NAF wage surveys.
2. Definition of additional industries: title 5 CFR Part 532 allows a lead agency to
define the addition of other industry classes to a regular survey in a wage area where those
industries account for signification proportions of the local private employment of the
kinds and levels found in local Federal employment. These added industries are defined
in terms of the appropriate NAICS code(s). For example, in the Orange County
(Orlando), Florida wage area, NAICS code 713110, Amusement and Theme Parks, is
added to the universe.
3. Determination of sample size: The number of units to be in the sample is
determined by the user or may be calculated as a function of the sample size of the last
full scale survey. This number is the number of establishments to be selected in the
sampling process and does not include the number of establishments that are to be
included with certainty. Presently, the sample size is determined by the user and is based
on the size of the sample for the wage area in the last full scale survey. The automated
system will compute the sample size if the user does not enter one. The necessary data
file from the last full scale sample selection must exist and be input to the sampling
process. In this case the sample size, n, will be calculated as:
n=

1
( M − m) 1
+
Mm
N

M = number of establishments in universe for last full scale excluding
establishments defined as certainty
m = number of establishments in sample for last full scale excluding certainty
establishments
N = number of establishments in universe currently excluding certainty
establishments
4. Stratification: The universe of establishments is determined by three factors: (1)
the physical location of the establishment; (2) the primary industry; and (3) the number of
employees. Each establishment in the scope of the universe for the wage area is put into a
group based on its primary industry and employment size. Establishments are first
grouped by industry. Then each industry grouping is divided into size groupings based on
the number of employees. These groupings are referred to as strata and the universe is
considered stratified. Establishments identified to be selected with certainty are grouped
into separate strata called certainty strata. The certainty strata have the same size

7

groupings as the sampled strata. See Appendix A for the definitions of the industry and
size groupings.
5. Allocation of the sample over the strata: The number of establishments to be
selected from each noncertainty stratum is calculated by allocating the sample size over
these noncertainty strata. The allocation is based on the number of employees in the
stratum. The allocation procedure is described in Appendix B. The number of
establishments to be selected from a certainty stratum is by definition all the
establishments in the stratum.
6. Measure of size: A measure of size is calculated for each establishment. It
determines the probability of selection for the establishment. The measures are calculated
independently for each stratum. The magnitude of the measure is related to the status of
the establishment in the previous full scale survey. Establishments that were selected in
the previous full scale survey have higher measures of size, ones that were in the universe
previously but not selected have small measures of size, and establishments not in the
universe at all previously have measures of size somewhere in between. This is done to
make the survey more efficient since there is less cost and time involved in contacting a
previously interviewed establishment. The measures of size are computed using formulae
derived from probability theories of statistician Nathan Keyfitz as first formulated in an
article in the American Statistical Association Journal, March 1951. The formulae were
presented in the internal BLS memorandum written by Phil D. Gilliland to Thomas N.
Wakin, subject: Automated Sample Selection Procedure for the Federal Wage System.
The formulae are shown below.
To calculate the measure of size for establishments in stratum r, the following quantities
are needed:
r = stratum that establishment is currently in
p = stratum that establishment was in last survey, if any
Nr = total number of establishments in current stratum r
Nrp = number of establishments in current stratum r which were in stratum p in
last survey
np = number of establishments in sample last survey for stratum p
nrp = number of establishments in current stratum r which were in sample last
survey for stratum p
The measures of size for establishments in stratum r are:
a. For establishments that were in the sample last survey in stratum p:
Nrp + np − nrp
n pN r

8

b. For establishments that were in stratum p last survey but not in the sample:
np − nrp
n pN r

c. For all other establishments, i.e., establishments in the universe for the last survey:
1
Nr
It is possible that the measure of size may be so large that an establishment will be
selected twice. This can happen if the measure of size is greater than or equal to the
1
sampling interval. See formula a above. The sampling interval is
where nr is the
nr
number of allocated establishments for stratum r. If this is the case, the establishments in
the stratum are reselected with equal probability by assigning all establishments a measure
1
where Nr is the number of establishments in the stratum. The sampling
of size of
Nr
process is fully automated and uses data files with records of the establishments.
Therefore, if the data file from the last full scale survey for the wage area is not available,
all establishments within the stratum are assigned the same measure of size using formula
c above.
7. Sample selection: The selection of establishments within each stratum is done
independently. Each stratum has N units from which n units are to be selected for the
sample as determined by the allocation procedure above. The units in a stratum are
numbered in a certain order, i.e., by NAICS code and number of employees. The first of
the n units to be selected is selected at random and then every kth unit on the list after that
is selected where
k=

N
.
n

Thus n units are selected. This method is called systematic sampling. The probability of
a unit being selected is its measure of size as described above.
8. Computation of the weight: Each establishment selected has a weight assigned.
The weight is a whole number and indicates the number of establishments in the same
stratum that the selected establishment represents. For example, if there are three
companies in the universe in a particular stratum and one is selected for the survey, it will
be given a weight of 3 because it represents itself and two other companies. All
companies in a universe survey have a weight of one since each represents only itself.
The sum of the weights for selected establishments in a stratum must equal the number of
establishments in the stratum.

9

The computation of the weight is as follows:
N
r
N
is calculated:
= W where N is the number of
n
n
n
establishments in the stratum; n is the number of selected establishments in the stratum; W
is the integer, whole number part of the quotient; r is the remainder part of the quotient.

a. The quantity

b. Since the sum of the weights must equal the number of establishments in the
stratum, r establishments are given a weight of W + 1 and n – r establishments are given a
weight of W. The weights are assigned randomly among the selected establishments.
9. Establishment Survey List: The selected establishments are printed in a specific
format showing the name of the company, address, employment size, industry (NAICS),
weight assigned, and other information. The certainty establishments are listed first in
ascending order by industry (NAICS) and then in increasing employment within industry;
then the sampled establishments are listed next and in the same order. This list is called
the Establishment Survey List (ESL).

10

III. WEIGHT ADJUSTMENTS FOR NON-RESPONSE
A. General
After data collection is complete, the appropriate WAGE regional office enters the
data into the survey database. This data includes the total employment and the collection
status (DAC, REF, OOB, etc.) for each establishment on the ESL. This data is then used
to adjust the weights.
B. Noncertainty Units
The procedures for noncertainty units (establishments selected by random sample and
usually having a weight greater than one) are designed to ensure that weights for refusing
establishments are randomly and evenly distributed among survey participants.
Nonresponding certainty establishments (found in universe surveys or specifically
requested to appear on the sample survey list with a weight of one) are accounted for by
increasing the weight of a comparable responding establishment to account for the total
employment. Nonresponding unique establishments may not be reweighted for since
some aspect of the unique establishment, such as employment size or industry type, may
preclude an assumption of comparability to another unit. For noncertainty establishments,
an analyst will determine whether the number of responding units in a given stratum is
equal to or greater than the number of refusals in that stratum.
If the number of responding units is equal to or greater than the number of refusals,
proceed to step 1. If not, proceed to step 7.
1. Add the weights of the refusing establishments and the responding establishments
in the stratum. Divide by the number of respondents. The result is the Adjusted Sampling
Ratio Reciprocal (ASRR).
ASRR = Weights of Refusals and Respondents_
Number of Responding Establishments
Example: The weights of the refusal and responding establishments totals 34 and
there are 8 responding establishments:
34
2
= 4 (do not reduce fraction to lowest common denominator)
8
8

2. If the ASSR is a whole number, assign it as the new weight of each responding
establishment in the stratum. If the ASSR is not a whole number continue to step 3.
3. The ASRR is expressed as a compound number (whole number plus fraction). Use
the remainder (the numerator of the fraction) as the number of establishments that get the
larger weight (the whole number plus one).

11

2
Example: The ASRR is 4 . Since the remainder is 2, two units will get the larger
8
weight of 5 and the other six units will get the smaller weight of 4.

4. Assign the new weights randomly (using random numbers selected by the
computer) among the responding establishments in the stratum.
5. Verify that the new weights are correct by adding them up. The sum should equal
the total original weights of the refusing and responding establishments.
2 x 5 + 6 x 4 = 34 total weights
6. Verify also that the new weights in the stratum are within one integer of each other,
i.e. weights of 5 and 4 rather than weights of 6 and 3.
7. If the number of refusing establishments is significantly greater than the number of
responding establishments in a cell size grouping, the stratum is collapsed into either the
next highest employment stratum or the next lowest to account for all employment in both
strata. This procedure will allow the weights of the refusing establishments to be
distributed as evenly as possible between responding establishments in both strata.
Example:
Cell Size

Weight

50 – 99
100 – 249

3
2

Establishments
Responding
Refusing
2
4
5
1

To collapse the strata, first compute the weighted total employment. This is done
by multiplying each within-scope unit (responding or refusing establishments) in each
stratum by its weight and then by the average employment size:
a. (2 x 3 + 4 x 3) x 75 = 1,350

(5 x 2 + 1 x 2) x 175 = 2,100

1,350 + 2,100 = 3,450 (total weighted employment)
b. The total weighted employment is then divided by the number of responding
units, which in this case is 7.
3,450
= 492 (Truncate the answer.)
7
c. This figure is then divided by the average employment size in each cell to obtain
the new weights. Round to the appropriate fractional ASRR:
492
1
→ 6 : therefore in the stratum with size grouping 50 – 99, the new
75
2
weights would be one unit with a weight of 7 and one unit with a weight of 6.

12

492
4
→ 2 : therefore in the stratum with size grouping 100 – 249, the new
175
5
weights would be four units with a weight of 3 and one unit with a weight of 2.

C. Certainty Units
In the case of a certainty establishment, the following procedure is used so that the
comparable establishment receives a new weight that accounts for the total employment of
both the comparable and the refusing establishment.
1. Sum the weighted assigned (expected) employments of the certainty establishment
and the actual (observed) weighted employment of the comparable establishment.
2. Divide this sum by the unweighted actual employment of the comparable
establishment.
3. Round the result to the nearest whole number. This is the revised weight of the
comparable establishment.
Example:
Certainty Establishment: assigned employment 1,222
Comparable Establishment: actual employment 600, weight 2
Sum of weighted employment: 1,222 + (600 x 2) = 2,422
Divide by comparable establishment’s unweighted employment:

2,422
= 4.03
600

Round to the nearest integer: 4.03 → 4
Comparable establishment’s revised weight: 4.
4. In some cases the comparable establishment’s weight will not change. This may
occur if the comparable establishment’s unweighted employment is much larger than that
of the certainty establishment.
D. Unique Units
If the establishment is a unique establishment (i.e., there is something about the
company such as employment size or industry that sets it apart from other companies),
reweighting is generally not recommended.
E. Combined Returns
A combined return (CMB) refers to a situation where data are collected for multiple
units from one responding unit.

13

1. Recognizable Units: When a participating establishment represents a combined
return, it is necessary to reweight for shifts or additions to the represented populations. In
a situation where data are collected from a unit clearly identified on the ESL that includes
one or more other clearly recognizable unit(s) appearing on the ESL, reweighting is
accomplished by computing the total weighted employment for these units, then dividing
this number by the actual employment for which data were collected. The result is the
new weight for the establishment from which data were collected.
2. Hidden Units: If a combined return includes data from an establishment on the
ESL plus establishments that are not recognizable on the ESL (hidden units in weight):
a. Identify the strata that contain the hidden establishments and randomly reduce
the weights of establishments on the ESL in those strata since the hidden establishments
are accounted for in the reweighting of the combined participating establishment.
b. Compute the new weight for the combined participating establishment.
a) multiply its assigned weight by its assigned employment;
b) add the total assigned employment of the hidden establishments;
c) divide the sum by the total actual collected employment.
Note: A combined return for hidden establishments can be reweighted only if all
establishments which the combined return represents are clearly within survey scope and
their individual employment is known.

14

APPENDIX A
Appropriated Fund NAICS Groupings
Industry Groupings
NAICS Code

Grouping
Number

311 - 316, 322,
324 – 326

1

321, 327, 331 - 339

2

Air

481

6

Rail

482

3

Truck

484

5

Industry
Manufacturing
Nondurables
Durables
Transportation & Warehousing

All other transportation, except taxi &
water
Support activities except water & road
Air
Rail
Truck
Couriers & messengers
Warehousing & storage

485 (except 4853)
487 (except 4872)
488 (except 4883 &
4884)
4881
4882
4885
492

4

6
3
5
4

493

5

515 (except Internet),
517

7

221

8

Waste Collection

5621

10

Waste treatment & disposal

5622

10

Durables

423

9

Nondurables

424

9

Communications
Utilities
Sanitation services

Wholesale trade

Size class groupings: 50 – 99; 100 – 249; 250 – 499; 500 – 999; 1,000 – 2,499; 2,500 and
more.
1

Non-Appropriated Fund CT NAICS Groupings

Industry Groupings for CT Establishments
Industry

NAICS Code

All Wholesale Trade
Gas Stations and Tire Dealers
Electronic Stores
Home Centers
Drug and Variety Stores
Clothing and Department Stores
Vending Machine Operators
Golf Courses
Bowling Centers
Hotels

423 – 424
441320, 447110, 447190
443111, 443112, 453210
444110
446110, 452990
448140, 452111, 452112
454210
713910
713950
721110
722110, 722211, 722212, 722213,
722410

Eating and Drinking Places

Grouping
Number
1
2
3
4
5
6
7
8
9
10
11

Size class groupings:
For all above NAICS except 441320, 447110, 447190, 454210, 713910, and 713950: 20 – 49;
50 – 99; 100 – 249; 250 – 499; 500 – 999; 1,000 – 2,499; 2,500 and more.
For NAICS 441320, 447110, 447190, 454210, 713910, and 713950: Same size class
groupings as above plus 8 – 19 grouping.
Non-Appropriated Fund AS/PS NAICS Groupings

Industry Groupings for AS/PS Establishments
Industry
Banking and Credit
Insurance

NAICS Code
522110, 522120, 522130, 522210
524113, 524126, 524114

Grouping
Number
12
13

Size class groupings:
For all above NAICS: 20 49; 50 – 99; 100 – 249; 250 – 499; 500 – 999; 1,000 – 2,499;
2,500 and more.

2

APPENDIX B
Allocation of Establishments to Strata
The procedure used to allocate the number of desired sample units over the strata that are
to be sampled is described below. By definition any stratum defined as a certainty stratum
has all establishments in the stratum in the sample and is excluded from this procedure.
1. Calculate survey sample size, n, for noncertainty strata only. Do not include the count
for certainty strata, if any.
2. Calculate the number of employees, Er, in each noncertainty stratum r, and the total
noncertainty universe employment, E, over all noncertainty strata.
3. Calculate the number of establishments, Nr in each noncertainty stratum r.
4. Calculate Fo:
Fo =

n (total noncertainty sample size)___
E (total noncertainty universe employment)

5. Calculate nr, allocated sample size for each noncertainty stratum r:
Nr = FoEr (Er = employment total in stratum r)
6. If nr is greater than or equal to Nr – 2, then set nr = Nr, i.e., all units in stratum r are in
the sample. The stratum is called a certainty sampled stratum. If there are no such strata, skip
to step 11.
7. Calculate the total number of units, Nc, allocated to certainty sampled strata (step 6).
Calculate the total employment, Ec, in the certainty sampled strata.
8. Calculate F1:
F1 = N (from step 1) – Nc (from step 7)
E (from step 2) – Ec (from step 7)
9. For all strata that are not certainty sample, recalculate nr:
nr = F1 er (er = employment total in stratum r)
10. Repeat steps 6 though 9, if Nc is greater than zero (step 7).
11. For all strata where nr is greater than zero and nr is less than one, allocate one unit, i.e.,
set nr = 1.
12. For all strata where nr is less than two and Nr is greater than one, allocate two units,
i.e., set nr = 2.

1

13. Otherwise, round nr, computed above (in step 5, 6, or 9 as appropriate to nearest whole
number.
14. For each stratum compute Wr =

Nr
nr

For any stratum where Wr is greater than the maximum weight, set nr =

Nr________
maximum weight

dropping any decimal portion.

2