14
METHODOLOGY
The results are based on polling conducted by Politico and Harvard T.H. Chan School of Public
Health. Representatives of the two organizations worked closely to develop the survey
questionnaires and analyze the results of the polls. Politico and Harvard T.H. Chan School of
Public Health paid for the surveys and related expenses.
The project team was led by Robert J. Blendon, Sc.D., Richard L. Menschel Professor of Health
Policy and Political Analysis at Harvard T.H. Chan School of Public Health, and Joanne Kenen,
Executive Editor, Health Care at Politico/Politico Pro. Harvard research team also included John
M. Benson, Logan S. Casey, and Justin M. Sayde.
Interviews were conducted with a nationally representative sample of 1,007 randomly selected
adults, ages 18 and older, via telephone (including cell phones and landlines) by SSRS of Glen
Mills, Pennsylvania. The interviewing period was February 21 – 25, 2018. The data were
weighted to reflect the demographics of the national adult population as described by the U.S.
Census.
When interpreting these findings, one should recognize that all surveys are subject to sampling
error. Results may differ from what would be obtained if the whole U.S. adult population had
been interviewed. The margin of error for the full sample is ±3.7 percentage points. For
questions asked of half-samples, the margin of error is ±5.2 percentage points.
Possible sources of non-sampling error include non-response bias, as well as question wording
and ordering effects. Non-response in telephone surveys produces some known biases in survey-
derived estimates because participation tends to vary for different subgroups of the population.
To compensate for these known biases and for variations in probability of selection within and
across households, sample data are weighted by household size, cell phone/landline use and
demographics (sex, age, race/ethnicity, education, and region) to reflect the true population.
Other techniques, including random-digit dialing, replicate subsamples, and systematic
respondent selection within households, are used to ensure that the sample is representative.