ORGANIZATION AND MANAGEMENT OF A DATA COLLECTION CENTER
Paul R. Calhoun and Laura W. Jackson
Bureau of Labor Statistics, 61 Forsyth Street SW, Atlanta Georgia 30303
John C. Wohlford
Bureau of Labor Statistics, 2 Massachusetts Avenue NE, Washington DC 20212
KEY WORDS: data collection, interviewers,
management, teams, specialization
DATA COLLECTION CENTER
GROWTH-MANAGEMENT MODEL
Optimum Management Structure
All opinions expressed in this paper are those of the
authors and do not constitute policy of the Bureau of
Labor Statistics
I. BACKGROUND
The Atlanta Data Collection Center (DCC) of the
Bureau of Labor Statistics opened in the fall of 1990
with three interviewers collecting employment data via
Computer-Assisted Telephone Interviews (CATI) for
the Current Employment Statistics (CES) survey. In
1992 a similar DCC was opened in Kansas City
followed by Dallas in 1997. By 2001 Atlanta CES staff
and workload had grown to 40 interviewers handling
over 45,000 monthly CES sample units with multiple
projects and software. The Job Openings and Labor
Turnover Survey (JOLTS) DCC, also in Atlanta,
opened in the fall of 1999, now with 21 interviewers
handling over 11,600 JOLTS sample units. From the
first decade into the second, the DCC presented
management challenges to maintain trained staff who
worked many projects while adjusting to organizational
and workload shifts.
This paper will provide a case-study analysis of the
interrelationship between organizational growth and
management of a Data Collection Center (DCC), and
its impact on the successful implementation of CATI
collection for an establishment survey.
II. ANALYTICAL FRAMEWORK
Team
Specialization
Team
Supervision
Direct
Supervision
Startup
Increasing Workload & Time
Growth
Maturity
The DGM model groups organizational change into
three phases; Startup, Growth, and Maturity. Startup
represents the period of time from the opening of the
functional unit, or beginning of actual operation when
new staff are hired and trained. Growth occurs as the
organization begins to expand activities and staff.
Finally, at Maturity efficiencies become realized with
expansion and specialization of staff assignments.
The DGM model matches three unique management
structures to each of the three change phases. Each
matching set effectively provides the most optimum
structure for the relevant change phase. The first
structure is Direct Supervision. This structure is
characterized by direct production control by
management. The second structure is Team
Supervision. Direct management control is exchanged
with supervisory teams, allowing for economies of
scale. Finally, Team Specialization allows an
organization to achieve efficiencies in the Maturity
phase while maintaining an optimum level of
performance.
Our case-study will analyze the Atlanta DCC
organization and management utilizing what can be
described as the DCC Growth-Management (DGM)
Model. This model describes the performance
optimizing relationship between management structure
in a DCC and organizational change. In the model,
performance is measured by survey collection rates and
workload (Chart 1).
A N A L Y T IC A L F R A M E W O R K
D C C G R O W T H -M A N A G E M E N T M O D E L
n D e scrib e s th e p e rfo rm a nc e o ptim iz in g re la tio n sh ip
b e tw e e n m a na g e m e n t s tru cture in a D C C an d
o rg a n iz a tio n a l c ha n g e .
O rg C ha n g e P h a s e
S ta rtu p
G ro w th
M a tu rity
M an a g e m e n t S tru cture
D ire c t S u p e rv is io n
T e a m S u p e rvisio n
T e a m S p e c ia liza tion
n P e rfo rm a n ce m e a su re d b y S u rv e y C o llectio n
R a te s
0 8 /05 /0 1
JSM -A SA A u g us t 2 00 1 ,B L S-C alh o u n/Jackso n /W o h lfo rd
3
III. CASE STUDY – Atlanta DCC
2.
1.
The Start-up phase had concerted support of upper
management, both locally and nationally. One BLS
manager was assigned to the Center with an assistant
added later. The Center opened with state-of-the-art
equipment for the period in a new and attractive setting.
All software was DOS-based, and printers were not
initially networked. Interviewers and first-line
supervisors were contractual staff obtained through the
DOL competitive-bid process. Since 1992 the
contractor has been Computer Based Systems,
Incorporated, (CBSI) now a subsidiary of the Titan
Corporation. CBSI recruited applicants and screened
them with BLS management review and approval.
Candidates were chosen for interviewer positions based
on computer experience, data entry skills, strong
customer service background, good telephone skills,
and poise. Interviewer staff worked part-time hours
during the first two-weeks of the month preparing mail
packages for respondents. Staff worked full-time
during the last two weeks of the month. This schedule
was a result of the CES collection period for current
month’s data which follows the pay period including
the 12th of the month. These last two weeks were
dedicated to intensive CATI collection.
Center management prepared training materials,
incorporating mock interviews and hands-on
workshops. Training emphasized persuasion skills and
data movements for employment, payroll and hours edit
parameters reflective of CES data experience. After six
months of interviewer contact, respondents were asked
to report to a touch-tone data entry (TDE) system or
return to mail response with State Employment Security
Agencies (SESA’s). At this point, time-consuming
decisions related to address refinement and reporting
site coverage were rare.
CASE STUDY - Atlanta DCC
ATL DCC
Director
CES DCC
Manager
JOLTS DCC
Manager
Asst. Manager
08/05/01
Growth (1992 – 1998)
Start-up (1990 – 1992)
Supervisor 1
Supervisor 2
Interviewers 1-10
Interviewers 1-10
Asst. Manager
Supervisor 3
Supervisor 4
Interviewers 1-10 Interviewers 1-10
Supervisor 1
Supervisor 2
Interviewers 1-10 Interviewers 1-10
JSM-ASA August 2001,BLS-Calhoun/Jackson/Wohlford
6
In the 1990’s, CES workload, projects, critical software
and equipment changes required management and staff
adaptation. Interviewers were trained to handle one
project where CES respondents reporting via mail were
contacted by telephone interviewers to provide data that
would be entered, edited and corrected on-line in BLS
developed software. From 1990 through 1996, we
enjoyed early success with collection rates averaging 95
percent throughout this project. Staff excelled at single
focus CATI collection. Each supervisor was assigned
ten interviewers to monitor and evaluate with standards
similar for duties surrounding survey procedures and
collection. There were some intermittent projects for
experienced interviewers who had an in-depth
understanding of basic work and could easily adjust to
new or additional related activities. During 1994 we
conducted the Response Analysis Survey of Payroll
Processing Firms that developed into a BLS report. By
1996 we were engaged in additional activities that were
permanently maintained and added to our monthly
workload and training requirements. CES work
projects included:
CATI collection of CES sample for six months,
primarily larger establishments
Mail-to-TDE conversion of SESA mail reporting
units to TDE reporting, primarily smaller
establishments
NRP (non-response prompt) calls to delinquent TDE
respondents
Operating and maintaining a TDE system
In 1995 the Atlanta and Kansas City DCC’s worked
with the University of Michigan’s Institute for Social
Research to determine best sample solicitation
techniques. In these three locations, through randomly
drawn establishment samples from BLS-Washington,
we assigned stratified panels among senior interviewers
for control testing. During this growth phase, the
Atlanta DCC expanded to four supervisors with 40
interviewers in order to continue the ongoing CES work
and complete solicitation testing. The goal was to
establish “best practices” for the CES Redesign when
the program changed from a quota sample to a
probability sample. From controlled testing and
interviewer debriefings, the project that began as
solicitation of CES establishments was changed to
“enrollment,” replacing a marketing term with survey
lexicon. In addition, interviewers perceived a negative
connotation from the public to the term “survey.” We
substituted the term “report” in our literature and script
where appropriate.
During this time, the DCC received upgraded software
to handle the many new functions associated with
enrollment of sample members. As in any office, and
particularly one with production standards, new
software can have a critical impact on staff who must
adapt to new processes. Handling new sample units
meant that not only were interviewers required to learn
new persuasion skills, but they also had to engage in
address refinement, including locating the potential
respondent who had the payroll information to
complete the CES form. (In JOLTS, through separate
pre-survey tests, BLS learned that target contacts could
be located in human resource departments where job
openings and vacancy data were more likely retained.)
Analysis by BLS Washington reported that address
refinement average time exceeded enrollment and
collection times. In Atlanta, average time spent on each
function was 44 percent on address refinement, 35
percent on enrollment and 21 percent on collection. As
the CES solicitation project grew, workload and activity
shifted toward consuming all four weeks of the month
with address refinement and enrollment work filling out
the first two weeks of the month. Most staff became
full-time interviewers, and were expected to handle all
projects. In addition to other work, four interviewers
were assigned to another project, the All Employee
Payroll test, which resulted in another BLS report.
office morale and performance slipped in some
categories.
In July of 1998, Atlanta Collection Center organization
changed to team management where each supervisor’s
group of ten was separated into two teams of
approximately five members. DCC management
distributed instructions for team conduct and
management changes. Management, supervisors and
interviewers were excited to engage in a new structure
that added more staff involvement. Teams met and
elected senior and experienced interviewers as team
leaders. The Center was fortunate that over half the
staff was experienced, many of whom had been with us
in excess of three years. Team members chose their
own names like High Percenters, SolSurvivors
(referencing our solicitation software, Solcati), Touch
of Class and other bonding nicknames. The DCC
structure was changed to support teams.
Atlanta DCC Organizational Chart
Team Specialization
CES DCC
Manager
Asst. Manager
AR & EN
Supervisor
CO & RC
Supervisor
FAX
Supervisor
ER/NRP
Supervisor
Team A
Team B
Team C
Team D
Interviewers 1-10
Interviewers 1-10
Interviewers 1-3
Interviewers 1-10
Growth in Enrollment Workload
60000
50000
08/05/01
JSM-ASA August 2001,BLS-Calhoun/Jackson/Wohlford
23
Units/Call s
40000
Total Units
30000
Total Calls
Enroll Units
20000
10000
Ap
r-9
8
Ju
n-9
8
Au
g98
O
ct98
De
c98
Fe
b99
Ap
r -9
9
Ju
n99
Au
g99
O
ct
-9
9
De
c99
Fe
b00
Ap
r -0
0
Ju
n00
Au
g00
Oc
t- 0
0
De
c00
Fe
b01
Ap
r-0
1
0
Month-Year
08/05/01
JSM-ASA August 2001,BLS-Calhoun/Jackson/Wohlford
19
3. Maturity (1998 - 2001)
During the late 1990’s, DCC staff turnover increased
from almost nonexistent to severe. Particularly, the
Atlanta job market became very tight with an unusually
high number of telephone centers vying for similar
skills. An article in the Atlanta Journal-Constitution,
July 26, 1999, titled “Call Centers Ring Georgia’s
Number,” cited that Atlanta was home to 184 telephone
centers which employed over 82,000 people. With new
software, growing turnover and new work processes,
Since CES is a monthly survey, all performance
measures are calculated each month at the time of the
first closing deadline. The monthly progress reports
were modified to highlight team performances. Office
atmosphere noticeably improved with team
competitions stimulating improved performance. In
January of 1999, enrollment was over 50 percent of our
call workload. Except for new staff, we had all teams
working on enrollment cases, including refining
addresses, processing fax reporters and handling TDE
support work, covering NRP and edit reconciliation
calls to TDE respondents.
As fax workload increased, we decided to assign two
interviewers to control all contacts to fax respondents.
For the rest of the staff, every month interviewers had
to manage their time to meet various quotas. In
October of 2000 management decided to differentiate
projects among supervisors and teams for specialization
of work. Teams were initially assigned to the following
functions:
Address refinement
Enrollment
CATI collection and refusal conversion
TDE support with limited CATI collection
Fax and refusal conversion
From October 2000 through February 2001, we
evaluated the effectiveness of team specialization.
During the first two months, we eliminated a backlog of
edit reconciliation calls to respondents. Address
refinement and enrollment rates immediately improved;
however, supervisors reported growing tensions
between the address refinement and enrollment teams.
In CES and JOLTS, the sample unit is drawn from
address team, having relinquished responsibility for the
case, felt their work was completed, and was not aware
of extra research performed by the enrollment
interviewers. We then assigned each address team
member a partner on the enrollment team, and much
progress was made.
After five months of separate address refinement
assignment, we regrouped again. At the current time
with many projects and quotas, we organized into our
most productive structure for CES with specialized
teams assigned to:
Enrollment
Enrollment and address refinement
Collection
CATI collection and refusal conversion
Fax and refusal conversion
Support activities
TDE support and limited CATI collection
Average time per case by mode in Atlanta DCC Oct 2000-Feb 2001
Ave ra ge tim e in m inute s
20.00
15.00
10.00
5.00
0.00
Oct00
Nov00
Dec00
Jan01
AR
11.11
13.37
14.57
15.03
17.40
EN
6.87
8.80
8.59
7.78
10.21
CO
7.79
8.57
7.26
7.93
8.71
08/05/01
Feb01
JSM-ASA August 2001,BLS-Calhoun/Jackson/Wohlford
22
SESA’s file of firms registered with the state
unemployment insurance (UI) system. In CES the
entire UI number is drawn with its site children while in
JOLTS the UI account is disaggregated into sites prior
to sampling. With each, there are complications in
address refinement. One of the observations, not
addressed in this paper, is the growing centralization of
payroll and administrative records by firms in this
country. With mergers and the increased use of payroll
processing software or firms, more often we are
directed to a central office to collect employment data.
As a result, the address refinement process has become
increasingly complicated. It was not a simple address
that the team had to research. Often during the process
of enrollment, interviewers might be directed to
business operations in different states, or to completely
different parent firms with the potential for sample unit
overlaps or conflicts. Consequently the enrollment
team who inherited work from the address team
complained of unresolved contact identification. The
Comparing performance for all staff to performance by
specialty team, we see increased uniformity within the
specialty team. In other words, there is a high
dispersion of performance measures when all
interviewers are included. When performance
measures are charted by team, we see higher
correlations in performance measures. It is easy to see
the effectiveness of the new structure. The team charts
depict structures with similar targets in meeting the
DCC goals. For example, enrollment interviewers
encounter higher refusal rates than interviewers
handling permanent CATI cases. We then evaluate
interviewers by team, by performance measure.
Workload and performance evaluations had to be
adapted to mirror the specialized activity within the
overall expectations for the DCC.
Interviewer Performance for Enrollment Team, March 2001
3.5
Inte rvie w e r m e a n com pa re d to DCC m e a n
25.00
3
2.5
2
1.5
1
0.5
0
A01
B01
B02
C01
D01
D02
F01
J01
J02
K01
K02
L01
L02
N01
P01
P02
R01
1.04
0.99
1.04
1.04
0.93
1.04
0.91
1
0.96
0.91
1.01
1.01
1.04
1.01
1.04
0.97
1.01
1.12
0.93
1.11
1.02
0.89
0.87
1.14
1.09
0.81
1.11
0.73
1.17
0.92
1.13
CORATE
1.33
1.21
1.17
1.34
1.37
1.29
1.19
1.18
1.08
1.29
1.41
1.03
0.72
1.07
1.08
1.29
1.05
1.14
RRATE
1.13
1.04
1.03
0.88
1.01
1.12
1.12
0.75
0.92
1
1.13
4PANCO
1.34
1.29
0.95
0.91
0.9
0.95
1.01
0.55
0.96
0.92
1.01
0.15
REFRATE
0.4
1.77
2.71
1.76
1.41
1.38
3.13
1.72
1.03
1.91
0.54
1.19
PCTLESS2
1
0.58
0.9
0.76
1.48
1.87
1.17
0.53
0.75
0.43
1.56
0.82
0.47
1.2
1.68
0.75
1.43
0.62
IR
1.03
1.31
1.13
1.26
1.1
1.22
0.99
1.01
0.92
0.97
0.61
0.75
1.24
0.99
1.2
0.89
1.05
0.65
CC
0.68
1.17
0.91
0.84
0.9
1.07
0.79
0.78
0.88
0.58
0.4
0.86
0.69
1.09
0.93
0.87
0.71
1.29
ARRATE
ENRATE
Interviewer
V01
1.04
1.05
1.12
IV. DCC Growth-Management Model and JOLTS
V. CONCLUSIONS
The successful development of the Atlanta DCC
prompted BLS to follow the same model with the new
JOLTS program. While work in JOLTS has been
growing rapidly, the first year of enrollment and
collection consumed all available interviewer staff, all
performing fairly similar activities.
The Atlanta DCC developed into a mature organization,
extending to cover two BLS surveys (CES and JOLTS)
while adapting to changing workload and functions
within each survey. Structural changes were initiated
that fit workload with staff skills and preferences. In
line with easing of Atlanta’s tight job market, turnover
has again slowed, decreasing time spent on
interviewing and training new staff. Even so,
organizational change reduced interviewer burnout by
offering different and rotating duties. Team structures
developed employee-bonding groups, and supervisors
became more expert in their specialty areas.
Specialization increased our total output with more
units handled per interviewer.
1.
JOLTS Start-up (1999)
The JOLTS DCC began with four full-time
interviewers and one supervisor in 1999 and quickly
increased to eight within the first four months. The unit
immediately began to utilize training materials from the
Atlanta CES DCC. With these tools at hand the JOLTS
DCC was able to immediately implement full-cycle
CATI Collection including refinement, enrollment, and
collection activities. With such a small staff, Direct
Management Supervision was the key to success at this
phase.
2.
Based on our Case-Study of the Atlanta DCC, we offer
the following conclusions concerning an effective way
to manage this type of data collection center:
1) A successful survey collection center goes through
an organizational cycle from startup through
growth to maturity.
JOLTS Growth (2000 – 2001, current)
During the period 2000-2001, the unit had grown to 21
full-time interviewers and two supervisors. Additional
activities including Touch-tone Data Entry (TDE) and
Refusal Conversion activities were implemented. With
the increased workload and more than doubling of staff,
management quickly moved its operational structure to
Team Supervision, as did the Atlanta CES DCC at this
junction in its life-cycle.
2) Effective management must be willing to change
organization structures to handle changing
workload through periods of growth.
3) Team management supports staff involvement and
improves morale.
4) Team specialization improves production
efficiencies and increases output.
GROWTH-MANAGEMENT MODEL
JOLTS
Workload Growth
14000
12000
# Cases
10000
STARTUP
GROWTH
8000
6000
4000
2000
0
10/99 11/99 12/99 1/00
2/00
3/00
4/00
5/00
6/00
7/00
8/00
9/00 10/00 11/00 12/00 1/01
Month-Year
2/01
3/01
4/01
5/01
6/01