Data Collection in the U.S. BLS

1
Data Collection in the U.S. BLS CES Survey

• New Frontiers for Data Collection
• October 31 November 2, 2012
• Geneva, Switzerland
• Ken Robertson, Assistant Commissioner
• Bureau of Labor Statistics

2
Outline

• Current Employment Statistics (CES) Survey
  Background
• A History of Data Collection in CES
• Current CES Data Collection

3
CES Background

• Bureau of Labor Statistics (BLS) CES survey is
  also known as the payroll or the establishment
  survey.
• Each month the CES program surveys approximately
  141,000 businesses representing 486,000
  individual worksites.
• Collection days before first release of the data
  varies each month, ranging from 10-16 days.
• About 25 of the non-certainty businesses are
  rotated out-of-sample each year

4
Major Uses of CES Data

• Economic indicator one of the earliest
  available each month
• These data are used to publish 4 news releases
  each month 2 national, and 2 sub-national
• Publish employment, hours, and earnings by
  industry and geography
• Input to other economic series

5
History of CES Data Collection

• 1915 1983
• Data were collected almost exclusively by Mail in
  a decentralized environment
• Collection rates were typically between 40-50
  for 1st release of the data, 90 for the final
• A mail shuttle form was utilized

6
Computer Assisted Telephone Interviewing

• 1984 CES CATI testing started
• 1987 Large-scale CATI test, involved 11 States,
  lasted 7 years, and ended with a test size of
  5,500 cases
• 1995 Collecting about 10,000 cases per month via
  CATI

7
Touchtone Data Entry

• 1987 CES began to experiment with TDE as a way
  to lower collection costs compared to CATI but
  to retain a higher response rate than achieved by
  mail

8
Voice Recognition

• 1989 CES briefly explored voice recognition as
  a data collection alternative

9
Electronic Data Interchange

• 1995 the BLS began to hear from larger
  employers, who were participating in multiple BLS
  surveys, that they wanted some way to reduce
  their cost of participation
• Developed the EDI Center to work with these large
  multi-site businesses

10
FAX

• 1995 BLS developed FAX collection for medium
  size firms.
• Used for collection where CATI is too burdensome,
  but business is not large enough for EDI

11
One Point TDE

• 1996 BLS developed the One-Point TDE system to
  take over the TDE responsibility for two states.
  Up to this point each state had independently
  maintained their own TDE system and helpdesk.
• By 2004 all states transferred their TDE
  operations to the One Point TDE

12
Web

• 1996 tested Web-based collection system
• 1998 decided to support Web collection.
• 2004 - CES started using the Internet Data
  Collection Facility (IDCF), a centralized service
  utilized for multiple BLS surveys
• WEB-LITE
• 2004 CES decided to try a streamlined version of
  Web-based collection.

13
E-Mail Web-FTP

• 2006 tested email collection utilized embedded
  HTML that allowed respondent to access BLS
  website
• Problems encountered because of different HTML
  rendering standards employed by different email
  clients
• Web-FTP
• 2007 BLS began utilizing a spreadsheet data
  collection form that the State of West Virginia
  had developed for medium sized firms
• Offered to respondents who have at least 5
  worksites but less than 100

14
A New Form

• 2011 Started a field test of a major redesign to
  the form, which had not changed significantly
  from the 1-page grid design since 1939
• New form is a 4-page form, printed on 11 X 17,
  folded to produce four 8½ X 11 pages

15
Collection Form 1915-2012
16
New Form - OutsideFront-introduction
Back-Thank You
17
New Form - InsideLeft -Instructions
Right-Data
18
History of major innovations in CES data
collection

• 1915 1983 Mail
• 1984 CATI
• 1987 Touchtone Data Entry (TDE)
• 1989 Voice-Recognition
• 1995 EDI, FAX
• 1996 One-Point TDE, WEB
• 2004 Web-Lite, IDCF
• 2006 Email
• 2007 WebFTP
• 2011 New Form

19
Current CES Data Collection

• Current Methods
• Costs
• Collection Rates

20
CES Methods Data Collection

• Current CES sample is collected through a variety
  of methods CATI, Fax, TDE, Web, EDI, WebFTP
• CATI yields highest response rates but is the
  most expensive
• Providing options helps sustain response rates in
  a voluntary survey
• Collection modes have evolved away from mail to
  automated methods

21
CES Collection Environment

• Challenging, especially for 1st preliminary CES
  has 10-16 days for collection
• Reference period is the pay period including the
  12th
• Collection begins as early as the 13th of the
  reference month, continues until 600 pm Monday,
  for the Employment Situation release, typically
  the first Friday of the month
• some firms do not have payroll available until
  after 1st preliminary cut-off

22
AE Collection Rates 1st and 3rd Closings,
January 2003 to Present
23
CES Collection Over Time
Table 1. Distribution of CES sample by
collection mode over time
Collection Mode 1915 1993 2004 2011
Mail 100 86 3 0
CATI 0 4 20 18
TDE 0 8 27 4
EDI 0 0 30 45
FAX 0 0 14 5
WEB 0 0 1 25
Other 0 2 5 3
24
CES Collection Over Time
Table 2. Collection rates and costs by mode,
2011 Average
Mode Collection rates at first release On-going collection cost, per unit
CATI 90.8 10.38
TDE 84.6 2.88
EDI 59.2 0.50
FAX 85.8 5.86
WEB 78.5 2.40
Other Varies Varies
25
Concluding Remarks

• CES data collection is a very large monthly
  operation with critical deadlines leading to
  multiple news releases each month
• Operations are managed centrally, and conducted
  in 4 Data Collection Centers, and an Electronic
  Data Interchange Center
• Data from about 141,000 businesses representing
  486,000 establishments are collected each month
  using various modes of collection
• Transition from decentralized operations in over
  50 states has resulted in substantial efficiencies

26
Concluding Remarks

• Managing multi-modal operations is very
  challenging
• must have highly qualified and motivated
  professional staff to pay attention to all the
  moving parts
• must fix problems quickly
• Must monitor distribution among modes closely to
  maximize response with available funding
• Must continue to innovate and take advantage of
  new technologies where it makes sense
• To maintain high collection rates
• To take advantage of new efficiencies

27
Ken RobertsonAssistant CommissionerIndustry
Employment Statistics202-691-5440Robertson.Ken_at_
bls.gov