How We Conducted This Study
The 2020 Commonwealth International Health Policy Survey was conducted by U.S. survey research firm SSRS and contractors in other countries from February to June 2020. The COVID Supplement was part of this larger survey and was in the field from March 30 to May 25, 2020, and administered to a nationally representative sample of adults ages 18 and older in Australia, Canada, France, Germany, the Netherlands, New Zealand, Norway, Sweden, the U.K., and the U.S.
A total of 8,259 interviews were completed for the COVID-19 Supplement Survey; final country samples ranged from 405 to 1,266. In the U.S., African Americans and Hispanics were oversampled in order to stratify the analysis by race and ethnicity. Interviews were completed either online or using computer-assisted telephone interviews.
In Australia, Canada, France, Germany, the Netherlands, New Zealand, the U.K., and the US, samples were generated using probability-based overlapping landline and mobile phone sampling designs; both mobile and landline telephone numbers were included to improve representativeness. Standard within-household selection procedures were used to increase the likelihood of reaching an eligible respondent for landline samples.
In the U.S., in addition to the overlapping dual-frame design, an address-based sampling (ABS) frame was incorporated to produce a representative sample of respondents. These ABS respondents were recruited via postal mail and invited to participate in an online or phone version of the survey. In Norway, respondents were randomly selected from a listed registry and interviews were completed via landline and mobile phones. In Sweden, respondents were selected via nationwide population registries, recruited via postal mail, and invited to participate in an online or phone version of the survey.
A common questionnaire was developed, translated, adapted, and adjusted for country-specific wording as needed. Not all questions were asked in each country (noted in the exhibits). Interviewers were trained to conduct interviews using a standardized protocol. Response rates varied from 14 percent (U.K., U.S., New Zealand) to 30 percent (Sweden).
International partners joined with the Commonwealth Fund to sponsor surveys, and some countries supported the use of expanded samples to enable within-country analyses. Data were weighted to ensure that the final outcome was representative of the adult population in each country. Weighting procedures considered the sample design, probability of selection, and systematic nonresponse across known population parameters including region, sex, age, education, and other demographic characteristics deemed consistent with standards for each country. In the U.S., the weighted variables also included race and ethnicity.
The margin of sample error for the COVID-19 supplement is approximately plus or minus 3 percent for Germany, plus or minus 4 percent for Australia and the U.S., plus or minus 5 percent for Canada, New Zealand, Norway, the U.K., and Sweden, and plus or minus 6 percent for France and the Netherlands at the 95 percent confidence interval.
We conducted bivariate analyses using Stata, version 16. Exhibits show weighted frequencies by country. To test for statistical significance, pairwise between-country differences were tested using unadjusted logit regressions.
The authors thank Jen Wilson and Chris Hollander, of the Commonwealth Fund, as well as Robyn Rapoport, Sarah Glancey, and Rob Manley, of SSRS.