Sample Selection and Characteristics

Sample Selection - Landline

Respondents were selected using a two-stage random digit dialing (RDD) sample design. In the first selection stage, a disproportionate stratified sample of telephone numbers was selected from two telephone number strata. One stratum consisted of blocks of numbers containing one or more listed numbers and presumed by the sampling firm to contain a high density of residential telephone numbers. The other stratum consisted of blocks of telephone numbers that include one or more unlisted telephone numbers and are presumed to contain a medium density of residential numbers. Both strata include only numbers that begin with area codes and exchange prefixes specific to Pennsylvania.

For the sake of efficiency, under the disproportionate stratified sample design, a larger proportion of the sample is selected from the stratum presumed to contain a high density of residential households.

Sampling continued as the selected telephone numbers were called to determine whether targeted numbers belonged to households with adult residents. Nonresidential telephone numbers were discarded from the sample. Residential numbers were subjected to a second stage of sampling wherein an adult was randomly selected as the respondent from a list of adults residing in the household. The person who answered the telephone generated this list.

Sample Selection - Cell Phone

Reliance on cell phones has increased steadily in Pennsylvania over time. Accordingly, the proportion of cell phones in the Pennsylvania sample has increased.

2011 18%
2012 19%
2013 25%
2014 35%
2015 50%
2016 67%
2017 70%

Cell phone usage differs from that of landline telephones. Accordingly, the management of the cell phone sample has its own characteristics.

Cell phone sample is selected randomly from blocks of numbers dedicated to cell phones. No subsequent within-household selection of a respondent is made if an adult is reached. Numbers selected as part of the landline sample for the survey -- that is those ported to a cell phone -- are transferred to the cell phone sample and called. Interviews of respondents selected from cell phone sample for other states but who are residents in Pennsylvania at the time of the survey are added to the Pennsylvania sample. These interviews will not contain responses to optional modules or state-added questions. Similarly, interviews of respondents from the Pennsylvania cell phone sample who were found during the interview to have moved to another state are transferred to the BRFSS sample of the other state.

Cell phone respondents are immediately asked whether they are driving or otherwise in an unsafe place. If so, they are re-called at another time, but some interviews are lost when this happens. Fewer follow-back calls are made for cell phones than for landline telephones. Unlike landline calls, a message identifying the survey is left at the first call attempt if there is no answer.

Sample Characteristics

The following table compares selected characteristics of the final interview sample for the 2017 BRFSS to the 2016 population estimates for the adult population of Pennsylvania. The observations used to calculate the estimates presented in the main report were weighted to account for differences between the population and the distribution of age, sex, race, Hispanic origin, marital status, home ownership, type of telephone and education characteristics of the sample.

Distribution of 2017 Pennsylvania BRFSS Survey Sample and 2016 Pennsylvania Adult Population Estimates for Selected Characteristics

2017 BRFSS Survey Sample 2016 Population Estimates
Number Percent Number Percent
All adults 6,542 100.00 10,109,382 100.00
Sex Male 3,040 46.47 4,892,729 48.40
Female 3,502 53.53 5,216,653 51.60
Race White 5,437 83.11 8,367,168 82.77
Black 592 9.05 1,051,858 10.40
Other 397 6.07 690,336 6.83
Unknown/refused 116 1.77 N/A N/A
Hispanic origin Yes 270 4.13 588,147 5.82
No 6,204 94.83 9,521,235 94.18
Unknown/refused 68 1.04 N/A N/A
Age 18-24 512 7.83 1,187,407 11.75
25-34 841 12.86 1,674,043 16.56
35-44 769 11.75 1,481,365 14.65
45-54 1,129 17.26 1,737,627 17.19
55-64 1,338 20.45 1,807,179 17.88
65-74 1,159 17.72 1,239,574 12.26
75+ 718 10.98 982,187 9.71
Unknown/refused 76 1.16 N/A N/A

Note1: Race data include Hispanics.
Note2: Population estimates allocate unknowns, so they are included in demographic categories. This is further indicated by the use of "N/A" or not applicable for the 2016 population estimate "Unknown/refused" entries.