The
Santa Barbara County Community Health Assessment Survey reports,
including all data tables, charts, and maps, are based on
data that have been weighted and post-stratified so that the
sample data are as projectable as possible to all county residents.
The survey consisted of a final random sample of 1050 households
in Santa Barbara County, California. Within each household,
one adult resident 18 years of age or older was interviewed
about his or her own health. In addition, data were collected
on a randomly selected child in each household that had one
or more children 17 or younger living in the household. Either
the randomly selected adult served as a "proxy" for the child
(answering our questions on behalf of the child), or another
proxy adult was interviewed about the health of the child,
if the adult respondent believed that the other adult would
be better able to provide more accurate information about
the child's health and risk factor status.
The
Sampling Frame
The survey used an equal probability of selection (EPSEM),
random digit dialing (RDD) design. The sample frame was all
residential telephone numbers ringing into households that
fall into known banks of working residential numbers in the
County. The frame was provided by Survey Sampling, Incorporated
(SSI). The EPSEM RDD method uses all possible area code, exchange,
and working block combinations known to contain at least one
working household number. As a result, all telephone households
in the geographic sampling frame are given an equal probability
of selection, within the limits of available data. Based on
studies using these samples, SSI has found that the general
pattern of these EPSEM list-assisted RDD samples closely resembles
those derived from well known and respected national random
digit dialing telephone surveys conducted at the Survey Research
Center (SRC) at the University of Michigan.
Once
the sample was purchased from SSI, sample control procedures
began by generating a series of random replicates (or sub-samples)
from the total sample of numbers for each stratum, with each
replicate containing 7 phone numbers.
At
the beginning of the study, a moderate number of replicates
were released to the system, and additional replicates were
introduced as the initial replicates were resolved. Interviewing
continued until all numbers in all replicates released to
the system had been disposed of, and did not stop when a target
number of completed interviews was reached. This guaranteed
that the final sample was a genuine probability sample of
the population, rather than a quota sample.
When
a residential number was successfully contacted, the household
members 18 years of age or older were listed according to
Michigan Behavioral Risk Factor Study (MBRFS) procedures:
men were listed from oldest to youngest, and women were then
listed in the same way. A random number was then generated
by ITS' CATI system based upon the total number of adults
living in the household, and this number was used to choose
one adult from the list to be interviewed. This procedure
was duplicated for the random selection of one child from
those children currently living in the household.
Weighting
and Post-Stratification
Weights were calculated to adjust for disproportionate probabilities
of selection, and the data were post-stratified so that the
results from the survey can be generalized to the population
of Santa Barbara County. Weighting, data management and risk
factor categorizations were performed using SPSS, SAS and
ITS' proprietary CATI software.
