TEXAS
I. Introduction to the State
A. Highlights
• Much innovative linking for research and program management
• Confidentiality and ownership issues cause some agencies to have limited participation. If
this is overcome, very few technical issues exist and even more data linking can take place.
B. Demographic and Caseload Description
In 1997, Texas had an estimated resident population of around 19.4 million people, ranked it
third among the states in absolute population. Twenty-nine percent of Texas's population in
1990 was under age 18 and 10% was over age 65. This places Texas ninth among the states in
the proportion of children in its population and forty-sixth in its proportion of elderly. The
median household income in Texas in 1995 (in constant 1994 dollars) was $32,039, ranking it
thirty-second among the states.
Compared with the rest of the country, Texas has a fairly high poverty rate. According to 1995
figures, 17% of the Texas population was living in poverty, ranking it sixth among the states.
Despite the high poverty rates, 1994 data indicate that Texas had a smaller proportion of its
population receiving public assistance (Aid to Families with Dependent Children (AFDC) and
Supplemental Security Income (SSI)) than most other states. With 6% of its resident population
receiving public aid, Texas ranked thirtieth among the states. In 1994, Texas had an estimated
794,000 AFDC recipients and 390,000 recipients of federal SSI payments. In 1996, there were
approximately 885,000 households with 2.4 million individuals receiving Food Stamps in Texas.
C. Interview Sources
Information about Texas's data systems and linked databases was collected through interviews
with over 15 people from the Texas Department of Human Services; Texas Department of
Health; Texas Health and Human Services Commission; Texas Workforce Commission; Texas
State Occupational Information Coordinating Committee; Texas Department of Protective and
Regulatory Services; Center for the Study of Human Resources at the Lyndon B. Johnson School
of Public Affairs at the University of Texas, Austin; and the Texas Legislative Council.
Interviews were conducted from June through December 1998.
II. Data systems for Social Programs
A. Names of Data systems
AFDC/TANF System for Application, Verification, Eligibility, Referral and
Reporting (SAVERR)
Food Stamps System for Application, Verification, Eligibility, Referral and
Reporting (SAVERR)
Medicaid Eligibility System for Application, Verification, Eligibility, Referral and
Reporting (SAVERR)
JOBS (JOBS successor) (CHOICES) Data system
Medicaid Claims Claims Subsystem on the Texas Medicaid Management Information
System (MMIS)
Child Protective
Services
Child and Adult Protective System (CAPS)
Foster Care Child and Adult Protective System (CAPS)
Unemployment
Insurance
Earnings Records
Wage record Subsystem
Child Support Texas Child Support Enforcement System (TCSES)
Child Care Subsidy Child Care Management System (CCMS)
JTPA The Workforce Information System of Texas (TWIST)
B. Description of Data systems
System for Application, Verification, Eligibility, Referral and Reporting (SAVERR)
The primary information system for welfare programs in Texas, SAVERR, is maintained by the
Texas Department of Human Services (DHS). This system contains information on
AFDC/TANF, Food Stamps, and Medicaid eligibility for the entire state. SAVERR has been in
place for almost two decades, first operated in the late 1970s. It is the system to which
caseworkers enter information on clients and cases.
The same system assigns a unique client number to individual who comes in contact with the
programs on the system. Each individual should have one SAVERR identification number over
time and across programs. This number is used only in SAVERR; no other state systems use the
number. SAVERR also generates a unique case number for each case in each program type.
However, in Texas, Medicaid is still linked to TANF, so a family's Medicaid case and TANF
case will have the same case number, while that family's Food Stamps case number will be
different. On the other hand, there may be multiple Medicaid case numbers in a family because
children of different ages are often in different Medicaid programs. Other identifying
information in SAVERR includes the Social Security numbers of everyone in a case, first and
last name (in one field), date of birth, sex, and ethnicity/race. Geographic information includes
the county of residence, but this is not a reliable data element because it sometimes reflects the
location where the recipient was served rather than where the recipient lives, or it is not updated
when the recipient moves. Mailing zip codes are collected, which do not necessarily reflect
where clients live.
In SAVERR, overwriting data fields takes place regularly. For example, the county of residence
is overwritten if it changes, so there is no historical information on that variable. The date of
Food Stamps certification and TANF approval are overwritten if a client returns after a period
off the program. However, there is a Medicaid history file that does provide information on an
individual's receipt of Medicaid over time, and it is found to be a good proxy for AFDC/TANF
receipt over time as well. This file contains just Medicaid eligibility information; Medicaid
claims information is on another system – MMIS (see below). The Medicaid history file can be
queried online, record by record, using the client identifier. However, if an individual was off
Medicaid for more than three years, earlier information on that client will have been purged.
Food Stamp case and client information is purged every 12 months. AFDC case information was
purged every two years and client information every three years, but currently TANF client
information is not being purged because of the need to track clients for time limits. In general,
much more client information is kept than case information. Much of the information being
purged has already been retained through regular archiving.
Data on SAVERR are regularly archived through the creation of "stripped files" each month.
These stripped files are the primary source of SAVERR data for state staff who analyze welfare
programs for administration, budgeting, forecasting, and research. It is much more common for
DHS staff to access and save these monthly files than to actually go into the SAVERR system
itself. These stripped files, containing case and client information, are available for at least the
last eight years for TANF.
Data from SAVERR are provided to other state or federal agencies as needed for each agency's
function. For example, the Texas Workforce Commission receives information on TANF and
Food Stamps, but not on Medicaid. Outside agencies receive data from SAVERR if they have a
contract with the agency (and thus are an extension of that agency) or if the law requires a study
of a program. For example, the Center for the Study of Human Resources at University of Texas
at Austin is under contract with DHS to study various welfare issues. Texas A&M is under
contract with the state legislature to develop a special database. (Both of these projects are
described in the "Linked Databases" section below.) If data are provided in other cases, they are
first "de-identified," which may also require aggregating data from smaller counties. Requests
for raw data are very rare, but requests for summary or aggregate data are much more common.
CHOICES Data system
In 1996, Texas reformed and renamed its JOBS program, calling it CHOICES. The original
JOBS data system, implemented by the Texas Department of Human Services (DHS) in 1990,
was updated in 1996 to reflect the new program rules. Although responsibility for the new
CHOICES program and its data system has been turned over to the newly created Texas
Workforce Commission (TWC), the system itself is still housed at DHS. The CHOICES data
system is a statewide system into which client data are entered by workers at the local board
offices that administer the CHOICES program.
The primary identifier used for individuals in the system is the SAVERR client number, which is
the same as the number used to identify individuals receiving DHS services such as TANF and
Food Stamps. Because it is an individual-level program, the CHOICES data system does not
assign a case identification number. It does, however, collect and maintain the case numbers that
are used by SAVERR. In addition to SAVERR identification numbers, the following individual
identifying information is collected in the CHOICES data system: Social Security number, first
and last name, date of birth, sex, race, and individual mailing and residential address, including
county codes for each address.
Many of these identifying data items are transferred through an interface with the SAVERR
system. The nightly batch process allows the CHOICES data system to be populated with new
applicants who are eligible to participate in Texas's CHOICES program. A monthly batch
process allows both CHOICES and SAVERR to be routinely updated with specific individual
program participation information. CHOICES also receives data from the Texas Education
Agency, which provides TWC with GED test scores for participants in the CHOICES program.
With the exception of SAVERR, the only other routine sharing of data from CHOICES is for
Texas's welfare waiver evaluation. The data set being constructed for the waiver evaluation (the
ACT Welfare Waiver Evaluation Database) is discussed in greater detail in the "Linked
Databases" section below.
In addition to the interfaces described above, data from CHOICES are being collected in a data
warehouse maintained by the TWC. The data warehouse contains information from the
CHOICES, TWIST (for JTPA and other employment support and training programs), Child Care
Management, Unemployment Insurance wage record, and other data systems maintained by the
TWC. The warehouse is discussed in greater detail in the "Linked Databases" section below.
Records from the CHOICES data system are purged every two years. At the time of the
interview, CHOICES data were routinely archived by DHS. These data, although they date back
to the system's inception in 1990, are not easily accessible. Requests for archived data require a
contract with DHS. In addition, there are technical challenges to accessing the data that were
stored prior to 1996. There are very few data fields that are overwritten when data for the field
change. Address and appointment data are two such examples. Other information in the
CHOICES system, including services received, client demographic information, and monthly
eligibility data, is not overwritten.
TWC is currently incorporating the functions of the CHOICES Data system into The Workforce
Information System of Texas (TWIST), which will allow for easier exchange of data and better
coordination of services across the programs overseen by the TWC. In addition, TWIST will
provide enhanced case management facility for CHOICES caseworkers. The TWC anticipates
that full conversion of CHOICES into the TWIST system will take place in the next two years.
Claims Subsystem on Texas Medicaid Management Information System (TMMIS)
The Texas Department of Health (TDH) contracts with four different companies to administer
the Medicaid program, also called the Texas Medicaid Administrative System, or "TMAS,"
which is transitioning to managed care. The National Heritage Insurance Corporation (NHIC),
the "claims administrator," is the primary contractor that processes claims and maintains the
central data systems. Birch and Davis administers the primary care and case management
(PCCM) program in the rural areas. MAXIMUS, the enrollment broker, assigns recipients to the
various health plans, and THQA (Texas Health Quality Alliance), performs quality control
monitoring.
NHIC maintains the two major data systems, the Texas Medicaid Management Information
System ("Texas TMMIS") and the "TexMednet." TexMednet serves as a clearinghouse, with a
library of information on the network of Medicaid programs. The other three contractors have
files on TexMednet to which they send information, and they can access the "library" on
TexMednet as they need. For example, Birch and Davis sends claims and contact data from the
PCCM program to TexMednet, where it is processed by NHIC and then sent to Texas MMIS for
storage. All claims and contact are stores in the "claims subsystem" on the Texas MMIS.
Much of the information on the TMMIS comes from SAVERR in the Department of Human
Services. This information is sent by tape to NHIC, as there is no online interface between
SAVERR and Texas MMIS. The Texas MMIS has an "eligibility subsystem" and "enrollment
subsystem," which contain the information from SAVERR's Medicaid eligibility file and
Medicaid enrollment file. SAVERR's Medical ID system prints the monthly Medicaid card for
clients, and SAVERR's Premium Payable system calculates premiums, and although these two
systems are considered part of the Medicaid data system, they are not copied onto Texas MMIS.
All client information in the Texas MMIS comes from SAVERR and is sent to NHIC by a
monthly tape. Texas MMIS uses the identifiers from SAVERR and does not have its own
receipt numbering system. It uses the client identifier from SAVERR to identify individuals and
also collects and maintains each client's "case number." Also captured from SAVERR are first
name, last name, birth date, Social Security number, sex, race, street address, zip code, and
county. Although the SAVERR system has a nightly update process as new client information is
received, Texas MMIS client information is only updated monthly through the tape of SAVERR
information sent to NHIC.
Child and Adult Protective System (CAPS)
The CAPS (Child and Adult Protective System) is maintained by the Texas Department of
Protective and Regulatory Services (DPRS). CAPS is the information system for most of the
programs run by DPRS, including child abuse and neglect investigation, child protective
services, and foster care. CAPS is a relatively new system, implemented in 1996.
Enhancements to CAPS continue as the smaller DPRS programs are integrated into the system.
CAPS is an online, active database that allows caseworkers in local offices to enter and access
case information. The data are then accessible to other authorized Protective and Regulatory
Services workers at the state office or in other local offices. In contrast to the previous system,
CANRIS (designed to track only the most basic of program statistics), CAPS was designed
mainly to help facilitate case management.
The primary identifier for cases in the system is an ID number that is unique to CAPS. All
individuals in a case are tied to this primary identifier. All individuals in the system are
identified by another, CAPS-specific number. Social Security numbers for all children and
adults in the case are collected routinely and consistently. When they are available,
TANF/AFDC, Food Stamps, and Medicaid case numbers are collected in CAPS. Other
identifying information contained in the CAPS system for both children and others includes first
and last name, sex, mailing address, county of residence, and race.
CAPS does not currently have any interfaces with other Texas social services data systems. The
key informant noted that there used to be an interface with the SAVERR system. This interface
allowed DPRS to determine client eligibility for other services. However, because data linking
between CAPS and SAVERR was difficult and did not typically yield a sufficient number of
matches, eligibility for other services is now determined directly through CAPS. The system
does use data from some other types of information systems. For example, criminal record
checks are made for prospective foster parents, and school records are occasionally accessed to
obtain additional child information.
No data in the CAPS system are overwritten, unless it is a field such as race or Social Security
number for which any change is the result of a previous data entry error. When data from other
fields (such as mailing address) change, the old information is written into an online transaction
file. There is no limit to the number of changes that can be recorded in the transaction file.
DPRS is currently implementing a new data purging plan. (Prior to this time, data purging was
not necessary because CAPS was so new.) Cases will be purged at different times, depending on
the outcome of the investigation or of the placement. For example, investigated cases that are
ruled with no risk are purged after 18 months, those with risk are purged after three years. Foster
care cases that end with adoption are kept indefinitely and for five years if there is no adoption.
Prior to purging, some data will be stripped of identifiers and placed into "tables" stored in the
online CAPS system. In addition, some data from the previous CANRIS system, dating back to
the early 1980s, are also available through tables in CAPS. The key informant reported that,
from a technical perspective, the data in tables are not difficult to access. However, there are
careful safeguards that restrict access to only a small number of people.
DPRS has memorandums of understanding regarding data sharing with other state agencies,
including the Department of Criminal Justice, the Texas Workforce Commission, and the
Department of Human Services. Most of these data-sharing agreements were legislated in order
to improve day-to-day program operation. For example, the Department of Human Services and
DPRS share case information so that caseworkers in each agency are aware of the other services
families are receiving and so that DPRS can ensure that families in their caseload are receiving
all the human services for which they have a need and are eligible.
Data from CAPS are fairly inaccessible to researchers outside state government. Sharing
agreements are made on a case-by-case basis. Those who have easiest access to CAPS data are
"partners" with whom DPRS has contracted to provide services. Special-use data sets are
typically constructed for these partners. Other requests from outside of government are subject
to great scrutiny. They must be made to the Executive Division of DPRS, which, if appropriate,
passes the request along to the Legal Services Division, which evaluates whether sharing the data
violates any privacy laws. Very few such requests for data have been approved.
However, DPRS carries out a great deal of in-house analysis using administrative data. The
department seems to be far advanced in terms of effectively extracting and using the wealth of
information contained in its child welfare system. Key informant interviews revealed that DPRS
has created numerous data sets to answer a variety of research questions and information
requests. The interviews further revealed that these examples are not isolated but that the DPRS
has a structure, process, and plans for continuing this type of work into the future.
An entire division – Forecasting and Program Statistics – is devoted to creating and analyzing
usable data sets extracted from CAPS and to providing data sets to other authorized persons in
the department. For example, field units can request data to carry out unit-specific evaluations.
Nongovernment service providers with whom DPRS contracts may also obtain relevant data
extracts. Data sets are used internally for analyses such as forecasting caseloads and predicting
lengths of service. In addition, to facilitate all of this work , DPRS is building several data
warehouses. The data warehouse and another innovative data set, the Work Measure Project
Data Set, are described in the "Linked Databases" section below.
Wage Record Subsystem
Unemployment Insurance wage records in Texas are maintained in the "Wage record
Subsystem" of the state's Tax Data system. The Unemployment Insurance program is run by,
and the Wage record Subsystem is owned and maintained by, the Texas Workforce Commission
(TWC), which also has responsibility for the state's JOBS successor (CHOICES), child care
subsidies, and a number of other work support and work training programs. The Wage record
Subsystem which is currently in use, was originally implemented in the late 1980s and was
substantially rewritten in 1996.
The primary client identification number used in the Wage record Subsystem is Social Security
number. The only other individual identifying information is collected in the Wage record
Subsystem is first and middle initial and the first 20 characters of the last name. Because it
supports a person-based program, case numbers are not collected as part of the Wage record
Subsystem.
No data fields are overwritten in the Wage Record Subsystem. The online system currently
stores data for prior quarters in two areas. A "wage detail table" contains data for the most
recent six quarters. A "history table" contains data for no fewer than the most recent 16 quarters.
The TWC is currently moving toward a quarterly purge process. However, because the last
purge of the Wage record Subsystem took place in 1996, there were more than 16 quarters of
data in the system at the time of the interview. The TWC is also setting up a process that will
archive Wage record Data prior to their purge.
The TWC routinely provides wage record extracts to a number of state agencies . In nearly all
cases, the agencies provide the TWC with a list of Social Security numbers that the TWC crossmatches
with the Wage record Subsystem to extract data for the relevant individuals. For
example, DHS uses the data to verify client eligibility for TANF and Food Stamps. The Texas
State Occupational Information Coordinating Committee uses the data to follow up on the
outcomes of various job training and education participants, a project that is discussed in greater
detail in the "Linked Databases" section below. The biggest challenge to these data-sharing
arrangements is that the sheer size of the Wage record Subsystem (over 350,000 employers) is
creating performance problems in completing the extracts, a challenge that will be addressed as
the state implements a new data system in the near future.
Texas Child Support Enforcement System (TCSES)
Child support enforcement is administered under the Texas Office of the Attorney General, Child
Support Division. Child support payments and collections have been tracked through the Texas
Child Support Enforcement System (TCSES) since September 1997. This system is "memberbased,"
the member being the custodial parent – that is, the one receiving child support
payments. The primary identifier is Social Security number. A case consists of the member and
his or her children on whose behalf support is ordered, and the system generates a case number
for each case and also keeps the "cause number" from the court order. A member may have
more than one case number.
Child Care Management System (CCMS)
In 1991, Texas implemented the statewide Child Care Management System (CCMS) for data on
the state's child care assistance program. CCMS is made up of a number of "applications," or
subsystems, that carry out different functions. Among them are a local service delivery
application, a budget and payment application, and a quality assurance and reporting application.
The system was originally implemented by the Department of Human Services in 1991, is owned
and overseen by the Texas Workforce Commission (TWC), and is implemented at the local level
by 28 local service delivery contractors.
Through the "local service delivery application," these contractors each maintain a local database
containing information about their clients. The local systems are not networked with one another
and only certain portions of the databases are extracted to files and the applications used by the
TWC. The TWC receives extracts monthly and uses these data for making payments and for
meeting federal and state reporting requirements.
The primary identifier used for individuals in the CCMS is the Social Security number. Cases in
the system are made up of a parent or caretaker and all children in care. CCMS uses a systemgenerated
case number to identify cases. Parts of CCMS also contain data about other children
and adults in the household. In addition to Social Security number, the following data elements
for all individuals in the household are collected: first and last name, date of birth, race, sex,
mailing address, and county of residence. However, for the most part, these data elements are
only extracted to the statewide files for the caretaker and for children in care. For others in the
household, these data elements are only maintained in the local databases.
Because child care assistance was previously administered through the DHS, CCMS collected
SAVERR case numbers until mid-1997. Currently, no case numbers used by other state data
systems are collected in CCMS.
Data are never purged from the local databases. As a result, local service delivery contractors
have maintained all the program data that have been generated since March 1991. TWC does,
however, purge its CCMS files as needed to expand system capacity. The TWC online system
typically contains (at least) the most recent three years of data. Some statewide CCMS data
(from September 1993 on) have been archived. However, the key informant noted that accessing
the archives is technically time-consuming and, therefore, not done very often.
Data from CCMS are routinely extracted and added to a database being used to evaluate Texas's
welfare waiver. This data file, the Achieving Change for Texans (ACT) (Welfare Reform
Waiver) Evaluation Database, is described in greater detail in the "Linked Databases" section
below. In addition, CCMS data are being collected in a TWC data warehouse, also discussed in
greater detail below. Other than this, CCMS data are primarily used by TWC to meet federal
and state reporting requirements.
III. Linked Databases
A. Integrated Database Network (IDBN)
The Integrated Database Network (IDBN) links data within the Health and Human Services
Enterprise, which consists of eleven service agencies and the oversight agency, the Health and
Human Services Commission (HHSC). These agencies have separate information systems using
a variety of different operating systems. In 1992, the commission established guidelines for the
functionality of each of the systems, and then developed the IDBN to test the ability to retrieve
and merge data from the different systems. The IDBN was implemented in February 1995. At
this point, IDBN successfully links and provides access to information from the four agencies
that initially participated in the project. The other agencies may provide information to the
system at a later time. The HHSC, Department of Information Resources, is the administrative
entity responsible for the operation of the IDBN.
The IDBN was developed by programmers from the state agencies and with a contractor,
Electronic Data systems (EDS). Both HHSC and EDS subcontracted with Catapult Systems in
Austin for additional programmers as specific skills were needed. Contracting with an outside
agency was cited as an important factor in the success of system development because it allowed
HHSC to comply with an aggressive time frame without having to work around the schedules of
each agency. The IDBN server is currently located and maintained at the MIS division of the
Department of Human Services. DHS has an interagency contract with HHSC to maintain the
database as EDS did initially.
At this point, data are fed from the systems of four of the eleven agencies: the Department of
Human Services (DHS), Department of Mental Health and Mental Retardation (DMHMR), the
Texas Department of Health (TDH)'s immunization tracking and WIC system, and the Texas
Rehabilitation Commission (TRC)'s Disability Determination System. The bulk of data come
from the DHS's UNISYS system, SAVERR, which contains information on all public assistance
recipients (AFDC/TANF, Food Stamps, Medicaid, SSI) in the state. The actual IDBN database
is primarily demographic information on clients in various programs administered by those
agencies, although IDBN can also be used to access SAVERR and the immunization tracking
system directly. It is a snapshot of the most current information and does not retain historical
data.
The IDBN was developed for two distinct purposes: to assist workers in the field to rapidly
collect information on clients necessary for case processing, and to assist state agency staff in
statistical and management reporting. State administrative staff use the database to forcast
caseload for planning and budgeting or to answer questions from the legislature, such as how
many clients are receiving services from more than one agency. Before IDBN, state analysts
had to pull and merge data themselves, which resulted in duplicate or missing data. With IDBN,
they now have one source of data that is already in usable form. State reporting and caseworker
inquiries remain the current users of the system. Although they do not yet feed their own data
into the IDBN, the Texas Workforce Commission (TWC) accesses the IDBN and imports data
into their own system, TWIST, for the use of their caseworkers, and they use IDBN as a tool to
get into the SAVERR system. Accessing the SAVERR or immunization tracking systems
through IDBN requires a higher security clearance.
Caseworkers and state administrative staff have different access to the system. State
administrative staff can access the information on all clients, and retain extracts of the system at
previous points in time. Caseworkers access current information on an individual client, after
receiving permission from that client through a signed release form. Workers receive a half day
of training on the system, which includes how to explain to clients the benefits of releasing
information, such as not having to provide information.
The HHSC expanded use of IDBN to two local service sites that provide a combination of local
and state services. However, both of these sites are primarily local services, and the information
in IDBN is primarily state services, so the system has not been found to be particularly useful. It
remains at only one of these sites currently, the Casey Project in Houston. The HHSC has not
yet received any outside requests, such as from researchers, for data from IDBN; however, if
they did, they would have a state analyst perform a query or piece together extracts from the
system rather than allow online access to the system.
The IDBN is a cross-sectional data set, containing only current information on clients in any of
the programs tracked by these systems (see above). Because its fed from different systems, the
data are real-time but updated at various times according to each system. SAVERR updates
daily, DMHMR updates weekly, TDH updates immunization information daily, and TRC
updates disability information monthly. The records are organized at the individual level,
although case information is also available for each individual. The IDBN is a SYBASE
relational database on a UNIX server, which can pull data from all types of mainframes,
including UNISYS (SAVERR) and IBM (DMHMR). All clients in the state in these programs
are on the IDBN, with zip code the smallest geographic identifier. When the system was
implemented in February 1995, the unduplicated count of individuals in the database totaled 6.5
million. The caseload has dropped since then.
In general the IDBN contains demographic information as well as an indication of the
program(s) the individual is in and the start and end date of those programs. The programs
include AFDC, Food Stamps, Medicaid, SSI, Mental Health and Mental Retardation programs,
immunization and WIC, and Disability Determination. Any other programmatic information
beside dates and case composition would be accessible only by going into the agency's system
itself, primarily SAVERR. Information captured in SAVERR would be displayed in the
SAVERR query screens. Neither foster care, child abuse and neglect, nor information or job
training information is available because the Department of Protective and Regulatory Services
and the Texas Workforce Commission have not yet provided information to IDBN, although
they have discussed with HHSC doing so sometime in the future. Child support enforcement is
under the Attorney General's Office and not part of the Health and Human Services Enterprise;
therefore, child support payment information is not on the system. The IDBN assigns each
individual its own unique client identifier, and also maintains identifiers from the other systems.
For example it keeps the SAVERR client number and case number.
Several years from now IDBN will be subsumed under a new system called the Texas Integrated
Enrollment and Services System (TIESS). The HHSC has been mandated by the Texas
Legislature to streamline eligibility and service delivery in three agencies: DHS, TDH, and
TWC. This effort will involve integrating both information and client processing. The
functionality of IDBN will remain the same, but it will be used through this new system.
B. Achieving Change for Texans (ACT) (Welfare Reform Waiver) Evaluation Database
In mid-1996, Texas implemented a welfare waiver program called ACT (Achieving Change for
Texans). The waiver had four main provisions, two of which are currently being evaluated by
the Center for the Study of Human Resources (CSHR) within the Lyndon B. Johnson School of
Public Affairs at the University of Texas at Austin. The first provision involves the
implementation of time limits on receipt of benefits. The second provision being evaluated is
called RER (Responsibility, Employment, and Resources) and involves various components
including a personal responsibility agreement for clients as well as changes to resource limits on
eligibility for services. The waiver and evaluation are scheduled to continue through the year
2001. The first interim report was due in October 1998. The final evaluation report is due in
May 2002.
To carry out the evaluation, CSHR is constructing a database using information for a sample of
people from at least ten state information systems, using data from at least 13 human service
programs.
• From the SAVERR system, CSHR is obtaining data about AFDC/TANF, Food Stamps,
and Medicaid eligibility.
• From the Texas Workforce Commission, CSHR is obtaining data about JTPA,
JOBS/CHOICES, child care, and Unemployment Insurance wages. Data for these four
programs are currently maintained in separate databases but there are plans to integrate
JTPA, JOBS/CHOICES, and child care into a single system called TWIST, from which
CSHR will continue to receive data about these programs.
• From the Office of the Attorney General's Child Support Enforcement Data system,
CSHR is obtaining data about child support collections and distributions.
CSHR is also in the process of negotiating with other state agencies to obtain the following data
for the sample of clients in the database.
• From the Higher Education Coordinating Board, CSHR plans to obtain participation and
completion data for both public four-year and community colleges.
• From the Texas Department of Regulatory and Protective Services, CSHR hopes to
obtain child abuse and neglect and foster care data.
• From the Texas Department of Health, CSHR is working to obtain immunization data.
• From the Texas Education Agency, CSHR hopes to use child education data, including
attendance, test scores, and drop-out data.
The database, which is under construction, will contain longitudinal data on the population of
welfare recipients from eight Texas counties (Bexar, El Paso, Ector, Nueces, Jefferson, Medina,
Caldwell, and Walker (spellings?)). Since the start of the waiver project in 1996, all new and
existing welfare recipients from these counties have been randomly assigned into control and
experimental groups. Approximately half of the welfare client population in each county is
assigned to each group. Although welfare reform has since been implemented statewide in
Texas, the welfare rules relating to the two provisions under study were not implemented among
the control groups. However, the control groups have been subject to some other provisions of
welfare reform.
The database currently contains information on approximately 45,000 caretakers and their
children. The sample size will grow as additional people in the counties are added to the public
assistance rolls. Records are organized at both the client and the person level and are housed in
SAS data sets on Windows-based PCs. All identifying information is maintained from the source
data, and the smallest, consistently maintained geographic identifier is county of residence.
In general, data from each system are being collected from one or two years prior to the start of
the waiver in 1996 and will be collected through December 2001. There are a few exceptions to
this plan: data for child care subsidy and child abuse and neglect are only available from 1996
and education data will only be collected through the year 2000. For the most part, data are
provided in the form of monthly tapes. However, because CSHR requests and obtains these tapes
only as the data are needed for subsequent stages of the evaluation, there is no regularly
scheduled process for updating the database.
There are a number of interesting items that are included in the data set. AFDC/TANF
recommended grant amount is measured monthly. CSHR noted that, in a few cases, the
recommended grant amount might be different from the actual grant amount (for reasons such as
recouped of overpayments). Although the AFDC/TANF data contain a "denial code" describing
why a client stopped receiving benefits, the CSHR researchers do not believe this code will be
very useful as it often is coded as "failed to show up for an appointment," no matter the reason
for exit from the program. CSHR plans to develop a code for "reason for exit from aid" using a
combination of data from Unemployment Insurance wage records, child support enforcement,
and other sources. In addition, the AFDC/TANF data do or do not contain a reliable code to
indicate recipients who have been sanctioned and how.
The data set contains information about child care receipt and amount, as well as information on
reason for eligibility, funding source, and type of payment made for child care. There are a
number of measures of participation in job training included in the data set. These include data
about participation in JTPA and JOBS/CHOICES, and plans to include information about
participation in higher education. Adult education is the only widespread form of job training
about which data will not be available. Education level as reported by clients at entry to
AFDC/TANF is included as part of the data from the SAVERR system. However, because
increases in education level after entry may or may not be recorded in the system by
caseworkers, CSHR may try to update education level using JOBS and JTPA data. Data about
employment wage rate and hours of employment are included for only JTPA and
JOBS/CHOICES participants. Occupation is only reliably available for JTPA participants.
Linking data across the various systems is accomplished using different client identifiers. First,
when it is available, data are matched according to the SAVERR client identification number.
Second, when this number is not included as part of the data set, Social Security number is used.
In some cases, Social Security number is used to check the matches completed according to
SAVERR client identification number. Finally, in a few cases (in particular for the
immunization records) CSHR anticipates that they will have to match according to first and last
name and other key demographic variables.
CSHR credits the success of this project to the strength of long-standing relationships and to
prior work with the social service agencies from which they are collecting data. The key
informants emphasized that it has been critical for CSHR to be very familiar with the programs,
the data systems, and the contact people at the various social service agencies. This point is
illustrated by the fact that CSHR is still negotiating access to data from agencies with whom they
have not previously worked closely. Legal issues regarding confidentiality present the largest
barriers to accessing the data. Although issues of understanding the programs and data and
insufficient agency staff time for data extractions are also barriers, the CSHR informants
believed that these issues were more surmountable, and less frequently encountered, than that of
confidentiality.
CSHR's contract with the Texas Department of Human Services specifies that the evaluation
data set be made public. To this end, CSHR is creating unique identification codes for clients in
the sample. These codes cannot be traced back to any information that would allow other
researchers to identify specific people in the sample. Other details about making the data set
public (such as who will own and distribute it and in what format) have not yet been determined
and will probably not be addressed until the end of the evaluation process.
C. Texas JOBS Program Evaluation Data Set
In the early 1990s, the Texas Department of Human Services (DHS), collaborating with the
Texas Education Agency, the Texas Department of Commerce, and the Texas Employment
Commission, contracted with the Center for the Study of Human Resources (CSHR), a research
center within the Lyndon B. Johnson School of Public Affairs at the University of Texas at
Austin, to conduct an evaluation of the state's Job Opportunities and Basic Skills (JOBS)
program. CSHR used administrative data from a number of social service programs to carry out
an assessment of JOBS participant outcomes in the areas of employment, earnings, and exits
from AFDC.
Longitudinal data for several cross-sections of AFDC recipients were collected from the
following systems/agencies.
• AFDC spell and demographic data from the SAVERR system, JOBS program
participation data from the JOBS data system, and DHS-funded child care and
transportation data from the AE payment system, all maintained by DHS.
• JPTA program participation data from the JTPA Management Information System
maintained by the Texas Department of Commerce.
• Employment Service program participation and employment referral data from the
SAMS data system and earnings data from the Unemployment Insurance Wage record
System, both maintained by the Texas Employment Commission.
Data from these systems were linked using Social Security number. The only exception to this
were the child care and transportation data, for which Social Security number was not directly
available. In these cases, data were linked using SAVERR identification number, which allowed
Social Security number to be linked from the SAVERR system to the child care and
transportation data.
The JOBS evaluation had a quasi-experimental design. A treatment group was randomly
selected from the statewide cohort of female AFDC caretakers who entered the JOBS program in
each of the seven quarters from November 1990 through March 1992. Around 3,300 new JOBS
entrants were selected in each quarter, yielding a total treatment group of around 20,000 women.
A control group of similar size was selected through use of a "nearest neighbor" match. The
selection process was designed to yield a comparison group, made up of non-JOBS AFDC
recipients, who were as similar as possible to the AFDC recipients who were JOBS participants.
The Texas JOBS Program Evaluation Data Set was organized at the individual level, with the
unit of analysis defined as the adult caretaker of the case. The data set was organized as flat files
and was analyzed using SAS on a mainframe computer. With the exception of Unemployment
Insurance wage records, data for each cohort were collected from the period of sample entry
(November 1990 through March 1992) through February 1993. Wage records were collected for
the eight quarters prior to sample entry through the first quarter of 1993. Analysis of the data
was undertaken for the period four quarters after the sample selection quarter and, in cases where
the cohort was chosen early enough, data were collected for six quarters after the selection
quarter.
There are a large number of specific data elements that were collected and analyzed for the JOBS
Program Evaluation Data Set. Some of these are listed below.
• AFDC data that were measured included: total time on AFDC as of the beginning of the
quarter; client's age at first AFDC spell; whether the client had more than one child;
whether the client had been sanctioned during the quarter; whether the client left AFDC
for the entire quarter; whether the client returned to AFDC during the quarter (after
having exited); and length of time between exit from and return to AFDC.
• Job training data that were collected included: number of hours and type of JOBS
services received; whether the JOBS participant was engaged in JOBS activities during
the quarter; whether the client received job-finding services from the Employment
Security Commission and frequency of receipt of these services; type and hours of JPTA
training, education and job placement.
• Educational activities related to JOBS and/or JTPA participation were measured in terms
of the hours of education completed. Educational attainment at application or recertification
was also included, although CSHR researchers caution that, in many cases,
this is an unreliable variable.
• Receipt of child care was included as a dummy variable.
• Earnings data from the Unemployment Insurance Wage record System were taken
quarterly.
The data set (with additional data) was used for a subsequent, second-year JOBS impact report.
However, the data set is no longer in use and is not accessible to any researchers outside CSHR.
There were no public use provisions for the data set. In fact, the confidentiality agreements
specified that CSHR could only provide the collaborating agencies with modified versions of
their own data. The collaborating agencies were prohibited from gaining access to any data set
containing information from other collaborating agencies.
CSHR reports that the effort to collect, link, and analyze data from the various state agencies for
the JOBS evaluation was successful. The results were used to develop future legislation that
modified the JOBS program and were also used in developing Texas's plan for welfare reform.
One factor that contributed to this success was that CSHR had done previous work linking data
from all of the systems except JOBS. In addition, it was critical that each of the collaborating
agencies supported and had ownership of the evaluation, both conceptually and financially. The
biggest challenge CSHR reported facing was with developing an appropriate procedure for
selecting a comparison group. This challenge was successfully addressed and a comparison
group that looked very much like the treatment group was selected.
D. Texas Workforce Commission (TWC) Data Mart
The Texas Workforce Commission (TWC), the state agency responsible for a wide variety of
work training and support programs and related data systems, is in the process of constructing a
data warehouse that assembles data from a number of state systems, including some which are
managed by other state agencies. The project was initiated in response to a legislative mandate
requiring the TWC to improve coordination of services across multiple programs and to bring
data from the various programs together to more effectively evaluate program outcomes. The
key informants reported that, thus far, the Data Mart has been successful in meeting the
legislative mandate.
The TWC Data Mart is a relational database that allows the data to be organized in any number
of configurations, including at the case and at the individual levels. The Data Mart contains
statewide data on the population of program participants since April 1996. Data continue to be
added to the Data Mart, which, at the time of the interview, was around 40 to 50 gigabytes in
size. There are no plans to purge information from the Data Mart.
For the most part, data from the various systems are loaded into the Data Mart monthly.
Unemployment Insurance wage records are loaded from the Wage record Subsystem quarterly.
Other systems from which data are collected include: SAVERR, (AFDC and Food Stamps),
CHOICES (JOBS successor), the TWIST (JTPA and other employment training programs), the
Child Care Management System, and the Child Support Information System. Some of the
specific data elements included are: TANF grant amount, size of family, participation in job
training or education, reason for exit from aid, work activity of program participants, education
level, and receipt of child care assistance. Data from the Child Support Information System
provide information regarding child support orders and receipt of child support payments.
Access to the individual records in the Data Mart is limited to authorized TWC staff and to local
workforce development boards responsible for the local administration of TWC programs. State
TWC staff have access to software that allows ad hoc queries to be made. Local boards receive
access to the data through "power cubes," which, though limiting the extent to which the data
can be rearranged and manipulated, facilitate analysis of the data. Others may only request
summary-level data. These requests must be made through the Data Release section of the
TWC.
E. The Workforce Information System of Texas (TWIST)
The Texas Workforce Commission (TWC) is currently constructing a data system that will
combine the separate data systems currently being used for JTPA, Employment Services, the
Child Care Subsidy, Food Stamp, Employment and Training, and CHOICES (the JOBS
successor) programs. This data system, The Workforce Information System of Texas (TWIST),
is being implemented in three phases, scheduled to be completed in the next two years. Since its
initial implementation in 1996, data for the JTPA and Employment Services have been converted
into the system. TWIST carries out a number of major functions, including case management,
following participant outcomes, and program reporting and evaluation.
TWIST was developed to assist the implementation of a new model for delivering employment
services in Texas, referred to as "One Stop." This model allows a person seeking services to
obtain assistance from a single caseworker, no matter for which program they are (or become)
eligible. TWIST facilitates the use of the One Stop model by allowing caseworkers to use one
system (rather than six) for managing clients in each of the different employment services
programs. TWIST was constructed using Sybase and, at the time of the interview, was
approximately six gigabytes in size. All historical data from the previous systems are converted
and maintained in TWIST.
TWIST uses Social Security number as the primary individual identifier. In addition, TWIST
assigns a case number for each program in which an individual participates. The system
routinely and consistently collects a number of other pieces of identifying information including:
first and last name, date of birth, sex, race, and mailing and residential address with county and
zip code.
Some demographic data fields (such as address) are overwritten when data for the field change.
However, TWIST does maintain a record of who changed the information and when the change
took place. No participation or services tracking data are ever overwritten. Because the system
is so new, no data have been purged from TWIST. However, TWC does plan to eventually
implement a purging process. Similarly, although there are plans to archive data, at the time of
the interview, no TWIST information had been archived.
There are some data from other state information systems that are automatically entered into
TWIST. Both the TANF and Food Stamp (SAVERR) and the JOBS successor (CHOICES) data
systems have nightly interfaces through which TWIST extracts demographic and case
participation data for clients listed in the TWIST system. In addition, these interfaces populate
the TWIST system with data for other individuals who might be eligible for TWIST-related
services. TWC plans to also extract Unemployment Insurance wage records for TWIST clients.
The only sharing of data from TWIST is with the SAVERR system. Through the automated
interface, TWIST passes information regarding employment program participation and suggested
case actions to TANF administrators.
F. Automated Student and Adult Learner Follow-Up System
The Automated Student and Adult Learner Follow-Up System, implemented by the Texas State
Occupational Information Coordinating Committee (SOICC), is a service that facilitates the
evaluation of a number of Texas job training and education programs. Through the Automated
Student and Adult Learner Follow-Up System, the Texas SOICC routinely provides data to a
number of private and government agencies on the post-exit results achieved by their former
education and job training participants. The Follow-Up System facilitates linkages of records
from a variety of data systems to state Unemployment Insurance wage records. In addition, the
Follow-Up System obtains data from a number of other federal and state databases likely to
contain information on the post-exit experiences of former education and job training
participants. While responsible for carrying out the data linkages, the Texas SOICC typically
turns the linked data back to the requesting agencies for evaluation and analysis.
The Texas SOICC has collaborated with (among others):
• the Texas Workforce Commission to generate follow-up data for JTPA, Food Stamp,
Employment and Training, Job Corps, and JOBS/CHOICES participants
• the Texas Education Agency to follow-up on senior high school exit cohorts and on exit
cohorts from a variety of specialty schools
• the Texas Higher Education Coordinating Board to follow-up on exit cohorts from all the
state's technical and community colleges as well as a large number of volunteer four-year
colleges and universities, and
• the Texas Department of Criminal Justice to follow-up on prison-based vocational
education participants.
These organizations provide administrative data on cohorts of participants to the Texas SOICC.
Using Social Security numbers, the Texas SOICC links the individual participant records to the
state's Unemployment Insurance wage record files. The Unemployment Insurance wage records
provide data on a number of items, listed below.
• Employment status
• Industry of employment
• Quarterly earnings
• Employer contact information
Data from the Unemployment Insurance wage records are supplemented with data from other
systems to more completely determine the post-exit status of former program participants. Texas
Department of Criminal Justice records are searched to identify former participants who have
been incarcerated. Federal employment records are used to identify former participants who
have found federal employment (which is not included in Unemployment Insurance wage
records). Texas Higher Education Coordinating Board records are searched to identify program
participants who have enrolled in public postsecondary education in Texas. GED files from the
Texas Education Agency are used to identify former participants who complete the GED
requirements.
In addition, the Texas SOICC is negotiating with other agencies to gain access to more data
systems. Vital statistics would allow the Texas SOICC to identify former participants who are
deceased. Education and Unemployment Insurance wage records from other states or federal
data sets would allow the Follow-Up System to determine the post-exit status of participants who
have moved from Texas.
Since the start of the Follow-Up System, the Texas SOICC staff have linked over 2 million
Social Security numbers to post-program achievement data. The Texas SOICC has created
annual databases with linked data since 1991 for JTPA participants, since 1996 for JOBS
participants, and since 1996 for Food Stamp Employment and Training participants. The
databases are created when, at the end of each program year, the administering agencies provide
exit cohort seed records to the Texas SOICC. These seed records contain information on the
entire population of program exiters from the previous program year (typically July through
June). The Texas SOICC insists that data on all exiters (including those who successfully
completed programs as well as those who did not) be contained in the seed records. For each
exiter, the seed records include Social Security number, an identifier for the program in which
the client participated, information about any occupational-specific training or education, exit
status, reason for exit, and exit date. In addition, the data provided by the administering agencies
also include relevant program-specific information. For example, the JOBS seed records include
fields from which length of time the exiter has received public assistance can be constructed.
With the Automated Student and Adult Learner Follow-Up System, the Texas SOICC uses
Social Security numbers to link these seed records to the Unemployment Insurance wage records
and other databases described above. Wage record data are added to the data set for two
quarters: the quarter prior to receipt of services and the final quarter of the calendar year in
which the program year ended. The "quarter prior to receipt of services" is identified for each
exiter using individuals' program start dates. When program start dates are not included in the
seed records, they are imputed using exit date and program guidelines that describe the expected
length of service.
JTPA Title III participants are the only ones for whom additional wage record data are linked.
Because this program specifically targets dislocated workers, wage record data are collected for
the entire year prior to receipt of services, rather than for just one quarter. This ensures that data
on wages prior to dislocation can be compared in more meaningful ways with wages after receipt
of services.
There is typically a lag of several months before each set of linked data is ready for use. The
Texas SOICC usually gains access to the fourth quarter wage record data in May of the
following year. Ultimately, the linked data are used by the requesting agencies to answer a
number of important client follow-up status questions.
The data have been used to study (among other things) earnings gains that result from program
participation, employment retention of program exiters, and job mobility (specifically the
distance participants must travel to find a job in the field in which they received training). In
addition, aggregate data that compare the achievements of exiters across training and education
programs are included in career planning software produced by the Texas SOICC.
In some cases, the linked records are supplemented with data from employer surveys. The
Survey Research Center at the University North Texas identifies the employers to be surveyed
using data from relevant Unemployment Insurance wage records and obtains data from the
employers about the occupations of former participants. This information is used by service
providers to assess how closely related specific types of training /education are to actual post-exit
occupations.
The Texas SOICC will continue generating data sets for new exit cohorts in all of the programs
described above. In addition, the Texas SOICC has a number of planned enhancements to the
data sets generated using the Automated Student and Adult Learning Follow-Up System.
Reduced welfare dependency will be explored as the "flipside" of employment. This will be
done by linking welfare payment data from a Texas Workforce Commission data warehouse
(under construction) to the other data routinely linked by the Texas SOICC. In addition, the
Texas SOICC is working on a number of pilot data sets that will assess the feasibility, usefulness
and costs of linking wage record data three and five years after program exit.
The Texas SOICC maintains the data sets on servers running the Windows NT operating system.
The data management architecture is a relational database constructed using Visual Fox Pro
software. There are plans to move to Oracle software for database management. Statistical
analyses are carried out using SPSS, although there are plans to move to SAS for statistical
analyses.
The data sets are jointly owned by the Texas SOICC and the agency that requests the data. The
data sets are not accessible to anyone, under any circumstance, who is not duly authorized.
According to the Texas SOICC, the innovative collaboration between agencies that administer
training and education and the agency that maintains Unemployment Insurance wage records is
based on strict understanding and adherence to privacy laws. The data are shared among
government agencies according to formal data exchange agreements that specify that the data
may only be used by public officials in carrying out their official duties leading to program
improvement. The agreements prohibit any participating agency from doing secondary releases
of the exchanged data.
The Automated Student and Adult Learner Follow-Up System was initiated in the early 1990s
with follow-up for graduates of one community college. The community college was facing
accreditation by an agency which wanted to know if there was a connection between fees
charged and wages in occupations for which the community college provided training. The
college had been conducting alumni surveys to collect these data but the surveys were expensive
and had a low (20%) response rate. As a result, the college approached the Texas Employment
Commission (now the Texas Workforce Commission) and asked them to link student and
Unemployment Insurance wage records. The success of this work led to the expansion of the
project to all community colleges in the state.
Expanding the use of wage records for follow-up of JTPA and other job training programs came
about as program funders (primarily legislators) demanded more rigorous accountability studies,
carried out by objective, neutral, third-party investigators. Success of the project can be
attributed to the commitment of the legislature to ask for and fund this type of follow-up study as
well as to the desire of various government entities to create a system that standardizes core
performance measures and research methods. In addition, the importance of adherence to the
data confidentiality rules cannot be overemphasized. The misuse of linked data by any
participating agencies would end the unique collaboration that makes the Automated Student and
Adult Learner Follow-Up System possible.
Although none of the data sets is available to outside researchers, the Texas SOICC staff are
willing to share their experiences in creating the Follow-Up System. The Texas SOICC also has
a number of written resources that can help other states implement a follow-up system similar to
the one that has been implemented in Texas. One of these resources, Anderberg and Pfeiffer's A
Field Guide to Automated Follow-Up, can be downloaded from the Texas SOICC's web site at
"http://www.soicc.state.tx.us".
G. Texas Legislative Information Data Set
The Texas Legislative Council is in the final stages of constructing a longitudinal data set
containing information about the population of TANF, Food Stamps, and Medicaid recipients.
The data set will be built using tape files extracted from the Texas Department of Human
Services System for Application, Verification, Eligibility, Referral and Reporting (SAVERR).
In addition, the Legislative Council is looking into, but has not finalized, plans for obtaining
administrative data from other systems to include in the data set. The Legislative Council will
use the data set to answer legislative requests for information about these social services. It is
expected that the data set containing SAVERR data will be available for use by the Legislative
Council by the end of 1998.
The data set will be stored in a relational database, primarily organized at the individual level.
SAS will be used to store, format, and analyze the data and will run on Unix and Windows NT
operating systems. Selected data from the monthly SAVERR tape files provided by the
Department of Human Services will be rolled into the database. However, the tape files will
remain accessible in the event that the Legislative Council needs additional data from these
tapes. Assistance in constructing the data set was provided by a research center at the
Department of Rural Sociology at Texas A&M University.
The Legislative Council currently has SAVERR data from September 1995 through the present
and will continue to add new data monthly, approximately one and a half months after the data
have been entered into the SAVERR system. Data will be kept for 12 to 24 months in the online
data set, with historical data to September 1995 available on demand. The smallest geographic
identifier data set will be county. The Legislative Council is still in the process of deciding
whether city and zip code will be included.
The Legislative Council signed an agreement with the Department of Human Services, ensuring
that the Council would keep the SAVERR data confidential by keeping it in a secure location
and by not releasing individual-level data to others outside the Council. The key informant noted
that future agreements regarding the use of data from other systems will probably require that
identifying information be encrypted. Under no circumstances will anyone outside of the
Legislative Council gain access to the Legislative Information Data Set.
The data set will contain, in addition to many others, the following specific data elements.
• TANF grant amount;
• Size of family;
• Participation in a job training program;
• Reason for exit from aid;
• Work activity; and
• Educational activity.
All of these data elements will be measured as they are in the SAVERR system. The Legislative
Council is exploring the possibility of including information from additional systems in the data
set but has not yet established any definitive plans for doing so.
H. Child and Adult Protective System (CAPS) Data Warehouses
Most data from the Child and Adult Protective System (CAPS) will be also be stored in at least
one of 15 data warehouses or "data marts" currently under construction by the Information
Technology Division of the Department of Protective and Regulatory Services (DPRS). The
warehouses are designed to simplify access to data from the CAPS system in order to facilitate
the data analysis, forecasting, and trending functions that are performed by DPRS staff. The key
informant estimated that the data in the warehouses provide the information needed to answer
80% to 90% of questions that might be asked. Obtaining answers to the remaining questions will
require directly accessing CAPS.
Each warehouse contains data specific to one aspect of DPRS's services. For example there are
permanency planning, family preservation, adoption, investigation, and intake warehouses, as
well as a "child subcare," warehouse in which each child is treated as his or her own case, and a
"family stages" warehouse that contains family-level data that can be linked to and from
information in the other warehouses.
Data are usually moved from CAPS into the data warehouses six days after the end of each
month. All identifying data associated with the various records are included in the warehouses.
Each warehouse has a basic "fact table" with the main data for the warehouse and a series of
dimension tables that store details for the information contained in the fact tables. For example,
most fact tables include dimension tables for history of services and persons associated with a
case. For a period of five months after initial entry, the data warehouses will accommodate
changes to time-specific data. This enhances the ease and accuracy with which figures such as
monthly caseloads can be calculated.
The data warehouses contain data for all cases that were active during the most recent 36
months. After that time, cases are purged from the warehouses. The warehouses are relational
databases that use Oracle, running on a Unix machine, for their reporting and storage functions.
ODBC and Microsoft Access can both be used to access the data contained in the warehouses.
I. Work Measurement Project Data Set
The Division of Forecasting and Program Statistics in the Texas Department of Protective and
Regulatory Services (DPRS) is planning to couple administrative data from the Child and Adult
Protective System (CAPS) with survey data to measure and analyze the dynamics of workloads
among department workers. This data set, containing information from November 1997 through
March 1998, will be used to create a "1998 Work Measurement Report," which is used in the
management of DPRS programs. Specifically, data from the report will provide guidance to
administrators in making important budgeting and staff allocation decisions.
The bulk of the Work Measurement Project Data does not come from CAPS but from task logs
that a random sample of workers fill out for the specified period. Data from the task logs will be
matched with case data contained in CAPS by using client names. Because the Work
Measurement Project Data Set is managed using the relational database capabilities of Microsoft
ACCESS, all of the data in CAPS (currently dating from 1996 to the present) are available on an
as-needed basis for use with other Work Measurement Project data. Case record data that are
likely to be used to carry out the Work Measurement analysis include: type of case, case
duration, and outcome and timing of specific case actions.
Carried out in response to legislative mandate, Work Measurement Projects were also completed
in 1990 and 1994. Because of the confidential nature of the data, no one except DPRS
researchers have access to the data set.
J. Post-employment Services Demonstration
This is a federal demonstration project being conducted in 4 sites in the country, one being San
Antonio, Texas. Texas Department of Human Services (DHS) applied to have a site for this
project, to test and study the effectiveness of post-employment services. Mathematica is the
contractor conducting the demonstration project in all four sites. State agencies in Texas
prepared data for Mathematica, which came primarily from SAVERR and data held at Texas
Workforce Commission. The project is now managed by TWC, rather than DHS, because the
project relates to employment.
IV. Contacts
Name:
Agency:
Address:
Phone:
Fax:
E-mail:
Expertise:
Marc Anderberg, Director of Follow Up
Texas State Occupational Information Coordinating Committee (SOICC)
Whitney Jordan Plaza #103B
9001 North Interstate 35
Austin, TX 78753
(512) 491-4943
(512) 491-4966
anderberg@coicc.state.tx.usStudent and Adult Learner Follow-Up System
(no access to data)
Name:
Agency:
Address:
Phone:
Fax:
E-mail
Expertise:
Jake Esterline, Lead Forecasting Analyst
Forecasting and Program Statistics
Texas Department of Protective and Regulatory Services
Mail Code E-661
P.O. Box 149030
Austin, TX 78751
(512) 438-4802
(512) 438-4853
esterlja@auste654c.aust.tdprs.tx.usChild and Adult Protective System (CAPS) Data Warehouses
Work Measurement Project
(no access to non-summary data for outside researchers)
Name:
Agency:
Address:
Phone:
Fax:
Expertise:
James Hermon, Unit Supervisor
Forecasting and Program Statistics
Texas Department of Protective and Regulatory Services
Mail Code E-661
P.O. Box 149030
Austin, TX 78751
(512) 438-4072
(512) 438-4853
Child and Adult Protective System (CAPS)
(no access to non-summary data for outside researchers)
Name:
Agency:
Address:
Phone:
Fax:
E-mail:
Expertise:
Cindy Peck, Associate Commissioner
Information and Resources Management
Texas Health and Human Services Commission
4900 North Lamar
Austin, TX 78751
(512) 424-6516
(512) 424-6591
cindy.peck@hhsc.state.tx.usIntegrated Database Network (IDBN)
Name:
Agency:
Address:
Phone:
Fax:
E-mail:
Expertise:
Deanna Schexnayder, Associate Director
Center for the Study of Human Resources
Lyndon B. Johnson School of Public Affairs
University of Texas at Austin
Suite 3.100
3001 Lake Austin Boulevard
Austin, TX 78712
(512) 471-2193
(512) 471-0585
dschex@uts.cc.utexas.eduAchieving Change for Texans (ACT) Welfare Waiver Evaluation Database
(no current access to data…plan for public use data set to be available after 2001)
JOBS Program Evaluation Data Set
(no access to data)
Name:
Agency:
Address:
Phone:
Fax:
Expertise:
Don Warren, PhD
Director, Statistics and Demographics
Texas Legislative Council
Capitol Station
P.O. Box 12128
Austin, TX 78711-2128
(512) 463-1143
(512) 936-1020
Texas Legislative Information Data Set
(no access to non-summary data for outside researchers)
--
Kenedeno & Associates
Inbox
