No single mode reaches every audience well. Online panel brings scale and efficiency; telephone reaches the low-incidence, senior, and hard-to-engage respondents an online frame misses. This is a vendor-neutral methodology for combining CATI and online into one coherent, mode-flagged dataset — the design patterns, the integration layer, and the discipline that keeps the two comparable.
CatalystMR Research Team. (2026). Mixed-Mode Sample: Combining Telephone and Online for Coverage & Feasibility. CatalystMR Methodology Papers. https://www.catalystmr.com/insights/methodology-papers/mixed-mode-sample/
@techreport{catalystmr_mixed_mode_sample,
author={{CatalystMR Research Team}},
title={Mixed-Mode Sample: Combining Telephone and Online for Coverage & Feasibility},
institution={CatalystMR}, year={2026}, type={Methodology Paper},
url={https://www.catalystmr.com/insights/methodology-papers/mixed-mode-sample/}
}TY - RPRT AU - CatalystMR Research Team TI - Mixed-Mode Sample: Combining Telephone and Online for Coverage & Feasibility PB - CatalystMR PY - 2026 UR - https://www.catalystmr.com/insights/methodology-papers/mixed-mode-sample/ ER -
For many B2B and healthcare studies, no single mode is sufficient. Online panel offers scale, speed, and cost-efficiency for accessible segments; telephone (CATI) reaches the senior, niche, and low-incidence respondents who rarely engage with a panel. Forced to choose one, a researcher either accepts a coverage gap or pays for telephone where it is not needed. Mixed-mode design refuses the choice — it uses each mode where it is strongest.
But combining modes is not the same as running two surveys. The value is realised only when the two streams converge into one coherent dataset. This paper sets out the four mixed-mode design patterns; the integration layer — a single master screener, harmonised quotas, and a mode-flagged data file — that holds them together; the harmonisation rules that keep the modes comparable; the mode effects you accept and must make visible; and a closing mode-effects watchlist of diagnostics to run before trusting the merged file. It builds on Paper No. 139 (when to use CATI) and No. 140 (screen-sharing CATI).
Mixed-mode is not a compromise or a fashion; it is a response to a coverage problem. Any one mode reaches some of a target population better than others — and for studies whose audience spans both the easily reachable and the hard-to-reach, insisting on a single mode means accepting that part of the sample will be under-covered or unaffordable. Combining modes lets each do what it does best.
Reaches accessible segments quickly and cost-effectively, at volumes telephone cannot match — ideal for the broad, digitally reachable core of a study.
Engages senior, niche, and low-incidence respondents through human outreach — the part of the sample an online frame under-covers or misses entirely.
The decision is therefore about coverage and feasibility, not preference. Use mixed-mode when online alone cannot reliably reach the whole audience, when CATI alone would be unnecessarily expensive across the easy segments, or when the design needs both scale and high-touch recruitment. The goal is total coverage at a defensible cost — not the purity of a single method.1
Mixed-mode is not a single design but a family of them, differing in how the sample is split between online and telephone. The four patterns below cover most B2B and healthcare studies; the right one depends on where the hard-to-reach part of the audience sits and what the instrument needs. The bars show the typical online / CATI balance — illustrative, not prescriptive.
The accessible core is fielded online; CATI tops up the quotas online cannot fill — the most common design.
Telephone leads on the hard-to-reach quotas first; online fills the accessible remainder where it is efficient.
Both modes run at once under a shared screener and quota plan, each drawing the respondents it reaches best.
An online core, with screen-sharing CATI carrying the visual task for the segment that needs it (see No. 140).
What makes a mixed-mode study one study rather than two is the layer between the modes and the data. Both modes enter through a single master screener with one quota plan; both exit into one data file that records, for every respondent, which mode produced the record. That convergence — not the fact of using two modes — is the methodology.
Same qualification logic · comparable wording · one quota plan
If each mode runs its own screener, the two halves qualify different people and the dataset cannot honestly be combined. One master screener guarantees that an online respondent and a CATI respondent met the same bar — the precondition for treating the merged file as a single sample.
Recording the mode on every record is what lets an analyst later check for, and if necessary adjust for, differences between the modes. A combined file without a mode variable has thrown away the one piece of information needed to test whether the join is sound.2
A clean integration layer is necessary but not sufficient; the two modes also have to be built comparably from the start. Harmonisation is the design-time discipline that makes the post-field merge defensible — four rules that keep an online complete and a CATI complete measuring the same thing.
Harmonise the inputs — screener, wording, quotas — at design time, and tag the outputs — source and mode — at capture. Do both, and the merged file is one sample; skip either, and it is two datasets wearing one cover sheet. Comparable measurement across modes is precisely how total-survey-error discipline keeps the join honest.1
Mixing modes introduces an honest complication: respondents reached by different modes can differ, and the difference has two distinct sources that are easily confused. A credible mixed-mode design names both, and uses the mode flag to keep them in view rather than buried in a combined average.
The two are easily confounded — which is exactly why they must be separated rather than reported as one blended number.2
Because the mode variable rides on every record, an analyst can compare the modes directly — and, where a comparable single-mode benchmark exists, begin to separate selection from measurement. The point is not that mode effects can always be eliminated; it is that an honest study measures and reports them rather than letting a combined average hide them.
Ask: “How will mode effects be checked and reported, not just assumed away?” A credible answer names selection and measurement separately — and points to the mode flag as the means.
Before a combined dataset is delivered or analysed as one, a short set of by-mode diagnostics tells you whether the join is sound. Each compares a quality signal across the two modes; a large divergence is not automatically fatal, but it is a flag to investigate before the modes are pooled. Run these as standard, not only when something looks wrong.
Mixed-mode sampling earns its place when no single mode can cover an audience affordably: online for scale, telephone for the hard-to-reach part, combined for total coverage and feasibility. But the value lives entirely in the join. A mixed-mode study succeeds only when the modes enter through one master screener, are built comparably, exit into one mode-flagged dataset, and have their mode effects measured rather than assumed away. Done with that discipline, the result is genuinely one sample drawn from two channels; done without it, it is two surveys sharing a cover sheet. Recognised survey-quality and conduct frameworks let buyers ask for the difference in consistent terms.3,4
CatalystMR is a global market-research panel and fieldwork partner specialising in hard-to-reach B2B, healthcare, and niche audiences. We run CATI and online panel sample from a single point of contact, applying one master screener and a common quality standard across both modes — and mode-flagging every record so a combined file can be analysed as one.
For mixed-mode work we harmonise screener logic, question wording, and quotas at design time, and monitor completion quality by source so the merge is defensible, not assumed.
Compliance posture: our methodology is aligned to the ESOMAR Code and Guidelines and the ISO 20252 framework, and we are certified under the EU–U.S., UK, and Swiss Data Privacy Frameworks, with personal data siloed from response data.