Methodology
This study is based on a telephone survey of the Illinois adult
population, aged 18 and older. Interviewers at the Public Opinion
Laboratory at Northern Illinois University completed 1309 Interviews
in November of 2004 through December of 2004. Interviews were
conducted and responses recorded using a computer assisted telephone
interview system.
The Illinois Policy Survey uses a disproportionately stratified
sample, divided into six geographical areas: Chicago, the balance
of Cook County, five collar counties (DuPage, Kane, Lake, McHenry
and Will), the rest of Northern Illinois, Central Illinois and
Southern Illinois. Sample sizes in the geographical areas ranged
from 201 to 242.
A two-stage sampling strategy was used. For the first stage,
telephone numbers were drawn with a known probability of selection
within regions. Both listed and unlisted numbers were included
in the sample at this stage. The second stage of sampling was
carried out when an interviewer reached an eligible household.
An eligible household was a non-institutional residence in Illinois
which was the home of at least one adult
18 years of age or older. After reaching an adult informant
in a household, interviewers gave a brief explanation of the
survey and the purpose of the selection process. The household
member to be interviewed was randomly selected from among the
adults who were at home.
Data were weighted to adjust for disproportionality in the
sampling. Unequal probabilities of selection at each stage of
sampling were accounted for in the initial weight. A post-stratification
adjustment to the initial weight was calculated by using a six
by two by six groupings of age groups by sex, by region using
current population
estimates from EASI data. The weighted sample was adjusted back
to the original sample size for analysis.
The sampling and weighting procedures were designed to provide
a representative sample of the adult population of Illinois
that is accessible by telephone. This year's survey was conducted
in both English and Spanish. Thirty-seven of the final interviews
were conducted in Spanish. The conservative 95% confidence interval
for a sample of 1309 individuals is plus or minus 3%. This means
there is a 95% likelihood the
population percentages will fall within 3 percentage points
of estimates based on the sample percentage. Confidence intervals
for subgroups of the sample will be larger, and will vary with
the size of the subgroup.