About The Survey

HOW THIS SURVEY WAS CONDUCTED

The Commonwealth Fund Affordable Care Act Tracking Survey, March–June 2017, was conducted by SSRS from March 28–June 20, 2017. The survey consisted of 15-minute telephone interviews in English or Spanish and was conducted among a random, nationally representative sample of 4,813 adults, ages 19 to 64, living in the United States. Overall, 1,198 interviews were conducted on landline telephones and 3,615 interviews on cellular phones.

This survey is the fifth in a series of Commonwealth Fund surveys to track the implementation and impact of the Affordable Care Act. The first was conducted by SSRS from July 15 to September 8, 2013, by telephone among a random, nationally representative U.S. sample of 6,132 adults ages 19 to 64. The survey had an overall margin of sampling error of +/- 1.8 percent at the 95 percent confidence level.

The second survey in the series was conducted by SSRS from April 9 to June 2, 2014, by telephone among a random, nationally representative U.S. sample of 4,425 adults ages 19 to 64. The survey had an overall margin of sampling error of +/- 2.1 percent at the 95 percent confidence level. The sample for the April–June 2014 survey was designed to increase the likelihood of surveying respondents who were most likely eligible for new coverage options under the ACA. As such, respondents in the July–September 2013 survey who said they were uninsured or had individual coverage were asked if they could be recontacted for the April–June 2014 survey. SSRS also recontacted households reached through its omnibus survey of adults who were uninsured or had individual coverage prior to the first open enrollment period for 2014 marketplace coverage.

This third survey in the series was conducted by SSRS from March 9, 2015 to May 3, 2015 by telephone among a random, nationally representative U.S. sample of 4,881 adults, ages 19 to 64. The March–May 2015 sample was also designed to increase the likelihood of surveying respondents who had gained coverage under the ACA. SSRS recontacted households reached through their omnibus survey of adults between November 5, 2014 and February 1, 2015 who were uninsured, had individual coverage, had a marketplace plan, or had public insurance. The survey has an overall margin of sampling error of +/- 2.1 percentage points at the 95 percent confidence level.

The fourth survey in the series was conducted by SSRS from February 2, 2016 to April 5, 2016 by telephone among a random, nationally representative U.S. sample of 4,802 adults, ages 19 to 64. The February–April 2016 sample was also designed to increase the likelihood of surveying respondents who had gained coverage under the ACA. Interviews in wave 4 were obtained through two sources: 1) stratified RDD sample, using the same methodology as in waves 1, 2 and 3; and 2) households reached through the SSRS omnibus where interviews were previously completed with respondents ages 19 to 64 who were uninsured, had individual coverage, had a marketplace plan, or had public insurance. The survey has an overall margin of sampling error of +/- 2.0 percentage points at the 95 percent confidence level.

The March–June 2017 sample was also designed to increase the likelihood of surveying respondents who had gained coverage under the ACA. Interviews in wave 5 were obtained through two sources: 1) stratified RDD sample, using the same methodology as in waves 1, 2, 3 and 4; and 2) households reached through the SSRS omnibus where interviews were previously completed with respondents ages 19 to 64 who were uninsured, had individual coverage, had a marketplace plan, or had public insurance.

As in all waves of the survey, SSRS oversampled adults with incomes under 250 percent of poverty to further increase the likelihood of surveying respondents eligible for the coverage options as well as allow separate analyses of responses of low-income households.

The data are weighted to correct for oversampling uninsured and direct purchase respondents, the stratified sample design, the overlapping landline and cellular phone sample frames, and disproportionate nonresponse that might bias results. The data are weighted to the U.S. 19-to-64 adult population by age, by state, gender by state, race/ethnicity by state, education by state, household size, geographic division, and population density using the U.S. Census Bureau's 2015 American Community Survey. Data are weighted to household telephone use parameters using the CDC's 2014 National Health Interview Survey.

The resulting weighted sample is representative of the approximately 190 million U.S. adults ages 19 to 64. Data for income, and subsequently for federal poverty level, were imputed for cases with missing data, utilizing a standard regression imputation procedure. The survey has an overall margin of sampling error of +/- 2.1 percentage points at the 95 percent confidence level. The landline portion of the main sample survey achieved a 16.5 percent response rate and the cellular phone main-sample component achieved a 9.7 percent response rate. The overall response rate, including the prescreened sample, was 9.6 percent.

Read the complete survey findings and analysis