Technical Information and Data
Numbered Memos
BRFSS POLICY MEMO 2003.2
Sample Provision Policy
BSB has developed a mechanism by which BRFSS projects can obtain sample
records from a commercial sample provider through BSB at no direct cost to
them. Projects are encouraged to obtain BRFSS sample through this mechanism.
The advantages for the projects and for the BRFSS as a whole are the
assurance of uniformly and appropriately generated sample and the inclusion
of a large number of fields on the sample records useful for improving data
collection as well as methodological and substantive analyses. BSB will
continue to provide sample records for all projects which request them. BSB
will have the sample records generated quarterly but the sample records will
be provided monthly. The provisions of this policy apply to contractors
collecting data for projects as well as to projects collecting their own
data.
Any sample obtained through BSB is for BRFSS data collection only. In
particular, BSB will not provide sample for BRFSS-like studies, even if they
are using the BRFSS questionnaire and the data are being collected at the
same time as BRFSS.
BSB will routinely review sample usage by projects. If a project is
consistently not releasing most of the sample it receives or if there is
strong evidence that the project is deviating from BRFSS protocol in such a
way that it could significantly increase its sampling efficiency (completion
rate) by adhering to BRFSS protocol, BSB may decrease sample provided during
a data collection year. In every case, BSB will consult extensively with the
project coordinator before decreasing sample.
Because the contract with the sample provider is a federal contract, BSB
is required to place the orders for sample. Thus, BSB will serve as
intermediary for ordering sample and for resolving issues regarding sample.
Projects should contact BSB staff, not the sample vendor, about any
questions or issues regarding their sample.
Projects are responsible for reporting any change in their designated
sample recipient to BSB prior to the effective dates of change.
Requests for additional sample: Projects may request additional
sample during a data collection year either to meet or to increase their
target monthly number of completes using the following guidelines:
- A project not meeting the monthly target(s) may request additional
sample if most of its sample is being released and the project is using it
in accordance with BRFSS protocol. (Projects with multiple geographic
strata will have a target for each stratum. For those projects, each
stratum will be treated separately.)
- The total quantity of increases in sample sizes that BSB is able to
accommodate during a data collection year is limited. Although BSB plans
for adjustments of sample sizes during a data collection year, the general
level of funding is based on target sample sizes provided by projects in
the Sampling Design Characteristics Form in October preceding the data
collection year. Approval of a large increase in target number of
completes by a project or several projects during a data collection year
could lead to a subsequent inability to provide more sample to a project
that is not meeting its original target number of completes. For this
reason, BSB must consider requests from projects for additional sample for
purposes of increasing target sample size on a case-by-case basis.
Projects should initially contact their Project Officers with such
requests.
Project Purchase of Sample: Coordinators or contractors who are
provided sample by BSB may purchase additional sample for BRFSS at their own
expense if prior, written agreement of the Survey Operations Section Chief
or the BSB Branch Chief has been obtained. Data from sample purchased
without the prior, written agreement of the Survey Operations Team Leader or
the BSB Branch Chief will not be processed as part of the BRFSS annual file.
Any additional sample purchased by the project must meet the same standards
as that purchased by BSB. Projects may not generate additional sample for
BRFSS independently.
Implementation of Sample Provision Policy
Requesting Change in Sample Size During a Data Collection Year
A project wishing to obtain additional sample during a quarter should
submit an e-mail request to the Sample Coordinator at least three weeks
prior to the month in which additional sample is needed (e.g., May 10 for
June 1 sample). A project wishing to increase or decrease sample size at the
start of a quarter should submit an e-mail request to the Sample Coordinator
at least one month prior to start of the new quarter. (e.g. June 1 for third
quarter sample). (Prior quarter's sample amount will be e-mailed to a
project=s BRFSS coordinator for review one month prior to the start of the
new quarter.)
Provide the following information by stratum to support your request:
- Sample Record Request—
Stratum Number |
1 |
First Month Sample Needed |
July |
Current Sample Amt. |
1000 |
Additional Amount (in 50's) |
200 |
New Total Amount |
1200 |
- Prior Month(s) History by Stratum—
Stratum |
Month |
Target Completes |
Actual Completes |
1 |
January |
250 |
236 |
|
February |
250 |
195 |
|
March |
250 |
207 |
|
April |
250 |
186 |
- Identify any other factors that may have influenced number of
completes.
E-mail call history files to BSB.
- Ci3 CATI Windows projects should include a zipped file of all
files in their study directory for the three most recently completed
months in the calendar year.
- Other projects should contact Survey Operations Team Leader to
discuss how to send call history files.
Review of Sample Requests
BSB will review requests as they are received and follow through with the
BSB sample provider to ensure timely receipt of sample. The review of a
request to increase sample size will involve obtaining answers to at least
the questions below. Other questions may also be addressed depending on the
situation of an individual project. Additional questions may also be
addressed as BSB gains more experience with requests for additional sample
records and expanded quality control procedures.
- Has the project routinely released almost all replicates in the strata
for which additional sample records are being requested and has the
project released all to-be-called numbers in the released strata?
- Has the project substantially followed the BRFSS callback rules, that
is, obtained in CATI an appropriate final disposition for almost all
numbers, including the calling of unresolved numbers at least 15 times,
with at least 3 calls on weekdays, 3 on weeknights, and 3 on weekends, and
over at least 5 calling occasions?
- Does the pattern of final disposition codes indicate that the project
may be missing significant numbers of households in its sample. For
example, does the project have extremely large percentages of unresolved
numbers among listed one-plus household numbers and not-listed one-plus
block numbers?
- Does the pattern of final disposition codes indicate that the project
is experiencing extremely low cooperation or response rates?
The intention of BSB is to approve all requests from projects for which
the answers are ‘Yes’ to the first two questions and ‘No’ to the last two
questions. Answers of ‘No’ to the first two questions and ‘Yes’ to the last
two questions could imply the existence of substantial numbers of additional
completes in the current sample and thus the possibility of an extreme
amount of bias in the data and the non-comparability of the data to those of
other BRFSS projects. In such cases, BSB may ask a project to make
appropriate changes in its practices before approving a request for
additional sample records.
For states which do not have any of the problems identified above, BSB
will estimate the number of sample records needed, by stratum, based on the
following formula:
Total number of replicates needed per month = Target monthly number of
completes divided by Expected number of completes per replicate, where
Expected number of completes per replicate = [The average number of
listed household numbers in previously generated replicates (for up a year
previously) times the completion rate among listed household numbers (for up
a year previously)] plus [the average number of not-listed one-plus block
numbers in previously generated replicates (for up a year previously) times
the completion rate among not-listed one-plus block numbers (for up a year
previously)].
The resulting Total number of replicates needed per month will be
routinely rounded up to the next whole number. If the fractional part of the
initial calculation is greater than .75, the Total number of replicates will
be rounded up an additional whole number. An additional adjustment may be
made in agreement with the project under special circumstances.
For example, suppose a project has two geographic strata. The total
number of replicates needed for each stratum has to be calculated
separately. Suppose that the replicates of 50 contain an average of 11.2
listed household numbers and that the completion rate among the listed
household numbers is .55. Suppose further that the replicates in the stratum
contain an average of 10.5 not-listed one-plus-block numbers and that the
completion rate among these numbers is .30. The expected number of completes
per replicate would be (11.2 x .55) + (10.5 x .30) = 9.31. Next, suppose
that the monthly target number of completes in the stratum is 75. Then, the
total number of replicates needed per month for that stratum would be 75 /
9.31 = 8.06, which would be rounded up to 9. The number of replicates needed
per month for the second stratum would be calculated separately in the same
way. If a state has only one geographic stratum, the above procedure is used
to calculate the single stratum.
This formula takes into account previous project experiences with sample
records and with outcome rates. BSB recommends that projects receiving
sample from BSB in the previous year and not experiencing any of the
problems indicated above use this formula to calculate their annual sample
requests.
For states with extremely low completion rates, a rate higher than the
observed rate might be substituted.
Purchase of Additional Sample Records by Projects, Including
Contractors, With Their Own Funds
Projects or their contractors wishing to purchase additional sample
records from BSB's sample provider may do so if (1) they meet the criteria
specified above for requests for additional sample, including agreement of
BSB, (2) the sample records contain the same fields in the same format as
sample purchased through BSB, and (3) they make arrangements for BSB to
receive the additional sample records. BSB will process data records from
sample purchased directly by projects only if the Survey Operations Team
Leader or the BSB Branch Chief has agreed to do so in writing beforehand and
they meet these three criteria.
Return to Top
|