Finding the 1 in 10 who insure themselves

The group most directly affected by the implementation of the Affordable Care Act's insurance markets - those who buy insurance on their own directly from an insurer or through one of the exchanges - is a difficult group to survey.

Those who purchase their own insurance make up just 1 in 10 people under the age of 65. Their demographic profile resembles that of the typical person who skips out on taking surveys - younger, more male, less income, less education. In the end, most surveys of the American public fail to capture enough people from this group to adequately assess their opinions, or even to properly identify them.

A survey from the Kaiser Family Foundation combined results from three separate samples in order to generate a representative survey of 742 adults who fall within this group. Most interviews (409) were conducted by cellphone, and 333 were completed on a landline telephone.

To be included in the survey, a respondent must have met several criteria.

They had to be between the ages of 18 and 64 and covered by a health insurance plan which they purchased. Researchers probed further to rule out those whose insurance plans were offered by an employer, extended via COBRA, managed by the government (such as Medicare or through the Department of Veterans Affairs), applied only on a college campus, covered other nonrelated employees, was subsidized by a trade association, or which the respondent said they purchased anywhere other than from an insurer or via healthcare.gov or a state exchange.

Interviews, conducted from April 3 through May 11, 2014, drew from three separate samples.

A group of 377 interviews came from traditional random national samples of telephone numbers. A second batch of 179 came from a random national sample that was selected so that those living in lower-income areas were more apt to be dialed, since income is a key predictor of insurance status. The final 186 had been reached using a traditional random national sample for a previous poll during which they indicated that they purchased their own insurance or were uninsured at the time. They were called back and had to meet the same screening criteria as those dialed without prior contact.

Results of the survey were weighted to adjust for the different sample sources and to account for differing probabilities of selection, and then weighted again to reflect the approximate demographics of this group for age, gender, education, race, ethnicity, region, population density, marital status and phone use.

Results for the full sample have a margin of sampling error of plus or minus 4 percentage points. It is larger for subgroups.