After more than a decade of the Colorado Health Access Survey, CHI set out to modernize how the survey reaches households and what it reports.

Colorado's legislature, state agencies, county health departments, and community organizations decide where to direct health resources, and those decisions turn on questions national data sources cannot answer: for example, how many residents of a given county lack coverage, who is skipping care because of cost, and how food and housing costs shape health. The Colorado Health Institute (CHI) created the biennial Colorado Health Access Survey (CHAS) in 2009 to address this, and has built it into a resource state and local leaders use to set priorities and gauge the effects of policy change.

County officials needed figures for their own populations rather than for multi-county regions. Reaching a representative sample of Colorado households was getting harder as phone-based methods declined. And any changes to the survey had to preserve comparability with prior rounds, or the trend lines that give the survey its value would break. CHI brought in NORC as a survey partner to make those changes.

NORC replaced the CHAS phone sample with an address-based design, then added county-level estimates in 2025.

NORC has served as CHI’s survey partner since 2021, fielding the 2023 and 2025 CHAS cycles. Our first structural change was the sample itself: we moved the survey off random-digit dialing to an address-based design that covers every Colorado household, mailing invitations to randomly selected addresses and offering the questionnaire online or by toll-free phone in English and Spanish. The design allows oversampling by age, income, race and ethnicity, household size, and insurance status, so estimates hold up for groups that a general-population sample would otherwise capture too thinly.

In 2025 we collected more than 10,000 interviews and built survey weights that produce representative estimates statewide and for each of Colorado’s 21 health status regions. New that year, we applied small area estimation to extend selected health access measures down to the county level, giving local officials figures for their own jurisdictions rather than for a multi-county region.

For the first time, we added a follow-up survey of respondents who agreed to further contact, letting CHI track changes between rounds. Separately, we test methods to inform future rounds: incentive experiments to lift participation and reduce nonresponse bias, and a prepaid cell phone sampling pilot aimed at households that standard designs reach less well.