Choose CATI when you need representative reach into hard-to-cover groups, complex probing, or sensitive topics that require a trained interviewer’s judgment. Choose online when speed, scale, and cost-per-complete matter more than universal coverage. Most research teams land somewhere between the two, using mixed-mode designs to control cost without sacrificing representativeness. The sections below walk through the trade-offs and give you a working checklist.

TL;DR:

  • CATI is essential for reaching populations without reliable internet access and improves response and completion rates among older or hard-to-reach respondents.
  • Online surveys are faster and cheaper but risk undercoverage of certain demographics, with higher item nonresponse and lower response rates, especially for sensitive topics.
  • Response rates are significantly higher in CATI, with a roughly 42 percentage point advantage over online methods, leading to better data quality and coverage.
  • Cost-effectiveness depends on response rates; CATI can be more economical per complete when online response rates drop below a certain threshold, especially with difficult samples.
  • Mixed-mode designs, typically web-first with phone follow-up for nonrespondents, optimize coverage and cost but require careful planning to account for mode effects and sequencing.
Veridata insights
Build a Research Design That Fits
Veridata Insights supports quantitative and qualitative research, from methodology and data collection through analytics and visualization.

Explore research support

Table of Contents

What Is CATI and How Does It Work?

CATI stands for computer-assisted telephone interviewing. A live interviewer calls respondents from a sample list, reads questions from a script displayed on screen, and enters answers directly into survey software that manages skip logic, quotas, and call scheduling automatically. The interviewer isn’t just reading a script. They’re managing the conversation.

That’s the whole appeal, and also the whole cost.

  • Interviewer role: probes vague answers, clarifies confusing questions, and keeps respondents engaged through longer instruments.
  • Quota management: real-time dashboards let supervisors close cells as targets fill, avoiding overshoot on any single demographic.
  • Monitoring: call centers typically record and spot-check a percentage of interviews for quality control, something no online platform can replicate.

The trade-off is straightforward. CATI wins on completion quality and reach into landline-and-mobile populations that skip online panels entirely, but it costs more per hour of labor and introduces interviewer effects, small shifts in how people answer when a stranger is listening. It’s also slower to scale past a certain sample size, since you’re bound by how many trained interviewers you can staff on a shift.

What Are Online Surveys and How Do They Work?

Online surveys are self-administered instruments delivered through email invitations, panel platforms, or open links, with no interviewer present at any point. That absence of a human moderator is precisely what makes the mode fast and cheap, and precisely where its weaknesses start.

Three deployment models dominate the field:

  • Probability-based online panels, recruited through address-based or random-digit sampling, which preserve some of the representativeness CATI offers at a lower run cost.
  • Standing commercial panels, built from opt-in volunteers and reused across studies, fast and inexpensive but prone to professional respondents.
  • Open links, shared broadly through social media or client databases, useful for directional feedback but weakest on sample control.

Online research wins on marginal cost and turnaround. You can field a national sample in days, not weeks. But coverage error is real: people without reliable internet access, older respondents, and certain hard-to-reach professional or clinical audiences are underrepresented no matter how large the panel claims to be. Item nonresponse, where people skip questions rather than answer honestly, also tends to run higher online than by phone.

Which Mode Wins on Response Rate, Cost, and Data Quality?

The evidence here is more decisive than most procurement conversations acknowledge. A randomized trial comparing phone and online completion found respondents assigned to the phone arm were roughly 42 percentage points more likely to finish the survey, with the gap widening further among women and older adults. That single figure should reshape how you budget fieldwork for hard-to-reach demographics.

Statistic callout: Phone respondents completed at rates far above their online counterparts in a controlled trial, and the same study found CATI reduced careless, rushed answering, meaning higher completion didn’t come at the expense of data quality.

Coverage tells a related story. General-population studies aiming for representative inference often need more than an online-only design, since coverage error and nonresponse bias concentrate in the same populations panels struggle to reach. Phone sampling, by contrast, reaches people who never touch a panel invitation.

Measurement quality diverges too, and not randomly. Pew Research Center found an average gap of about 5.5 percentage points between phone and web answers across 60 tested questions, with the largest mode effects showing up on sensitive or evaluative items, the kind where a live interviewer’s presence nudges people toward socially acceptable answers, while a screen invites more candor. A PMC review of CATI and web-based collection adds that CATI typically produces fewer skipped items, meaning cleaner datasets with fewer missing-data headaches downstream.

Cost is where intuition misleads people most. Online almost always wins on sticker price per interview. But when outreach costs are high and completion rates online run low, the math on cost per completed interview can flip in CATI’s favor, because you’re not paying to recruit five times the panel just to backfill dropouts.

Quick summary:

  • Response/completion: CATI leads, sometimes by a wide margin, especially for older and harder-to-reach respondents.
  • Coverage: CATI reaches non-internet populations; online risks systematic gaps.
  • Measurement: online reduces social-desirability bias on sensitive questions; CATI reduces item skipping.
  • Cost: online wins on unit price; CATI can win on cost-per-complete when response rates diverge sharply.

How Do You Decide Which Mode to Use?

Run through this before you write a single questionnaire line:

  1. Define the population. If a meaningful share lacks reliable internet access or skews older, weight your design toward CATI or a phone-inclusive mixed mode.
  2. Assess topic sensitivity. Health behaviors, income, or anything socially loaded performs better self-administered, where no interviewer is listening.
  3. Estimate probing needs. Open-ended, exploratory, or complex-routing questionnaires benefit from a human who can clarify on the spot.
  4. Set the sample size and timeline. Large samples on tight deadlines usually favor online or web-first mixed-mode; small, precise samples favor CATI.
  5. Build the weighting plan before fielding, not after, so quota targets match your intended adjustment strategy.

Pro Tip: Write your RFP questions before you pick a vendor. Ask how they handle quota fills for underrepresented cells, what their average CATI completion rate looks like for your population, and how they sequence mixed-mode follow-up. Vague answers here predict vague data later.

As a rule of thumb: go web-first sequential mixed-mode when budget is the binding constraint but you still need broad coverage. Go CATI-only when the population is hard to reach digitally or the topic demands real-time probing.

Decision paths for choosing survey collection mode

How Should You Structure a Mixed-Mode Study?

Sequential designs, typically web-first with phone follow-up for nonrespondents, tend to control cost better than running both modes simultaneously, because you’re only paying for the expensive channel where the cheap one failed. Concurrent designs, offering both at once, can boost overall response but complicate mode-effect analysis since you can’t cleanly separate who chose which channel and why.

A few operational habits make sequential designs work:

  • Tailor invitation language by subgroup rather than sending one generic blast.
  • Use incentives strategically to lift web uptake before escalating to phone follow-up.
  • Route only the specific quota cells that are underfilled to CATI, rather than re-contacting everyone by phone.
  • Treat phone follow-up as a nonresponse reminder rather than announcing a mode switch upfront, since flagging the follow-up mode in advance can depress initial web response.

Weighting gets more complex once you combine modes. You’ll need to account for mode as a variable in your adjustment model, not just demographics, or you risk baking mode effects into your topline numbers. A primer on defensible survey weighting walks through how to build that adjustment plan before fielding starts.

What Does Fieldwork Actually Cost and How Long Does It Take?

Online studies typically field in one to two weeks once programming is done; CATI studies run three to six weeks depending on sample difficulty and quota complexity. Build in extra lead time for translated instruments or specialty screening.

  • Cost drivers: interviewer hours and call attempts for CATI; panel incidence rates and screening length for online.
  • Recruitment: hard-to-reach B2B and healthcare quotas often need targeted custom recruitment paired with CATI follow-up rather than reliance on standing panels alone.
  • Questionnaire design: CATI scripts need built-in probe language; online instruments need mobile-optimized matrix items and shorter grids, since effective question design matters more once there’s no interviewer to rescue a confusing item.

Veridata Insights: Built to Run Whichever Mode Your Study Needs

Here’s the practical problem with most method debates: they assume you’re locked into one mode. Veridata Insights isn’t, and neither should your project be. Our full-service market research team scopes CATI, online, and mixed-mode studies side by side, then recommends whichever combination actually fits your population and budget instead of whichever mode we happen to sell.

We handle respondent recruitment for B2B, healthcare, and other hard-to-reach audiences where panel-only sampling falls short, plus survey programming, weighting, and data visualization once fieldwork wraps. Mixed-mode sequencing lives or dies on recruitment expertise. That’s the part a Scrappy-Doo sized team can get wrong just as easily as a giant one, and we know exactly which cells need a phone call instead of a panel invite.

There are no project minimums, and support is available every day of the year to scope your next study. Consultations are available to help build a research design around your population, not any preferred channel.

Sources

FAQ

What Does CATI Stand For?

CATI stands for computer-assisted telephone interviewing, where a live interviewer conducts the survey by phone using software that manages scripting, routing, and quotas in real time. It differs from CAPI, which is the in-person equivalent using a tablet or laptop.

What Are the Disadvantages of Online Surveys?

Online surveys carry coverage error, since people without reliable internet or in certain hard-to-reach groups get underrepresented no matter how large the panel is. They also show higher item nonresponse than CATI in comparative reviews, and open-link designs offer little control over who actually completes the survey.

What Are the Key Differences Between CAWI and CATI?

CAWI (computer-assisted web interviewing) is self-administered online with no interviewer present, while CATI involves a live phone interviewer guiding the respondent through the questionnaire. CATI tends to produce higher completion and fewer skipped items; CAWI tends to reduce social-desirability bias on sensitive questions, per Pew Research’s comparative findings.

How Do You Interpret a Mode-Effect Difference Between CATI and Online Results?

A mode-effect difference means the survey channel itself, not just the underlying opinion, shifted how people answered, most often on sensitive or evaluative questions. Pew’s research found gaps averaging 5.5 percentage points across 60 questions, so any comparison across modes should account for that before drawing conclusions about real change.

When Should You Use Mixed-Mode Instead of a Single Method?

Mixed-mode makes sense when you need both broad coverage and cost control, typically through a web-first design with phone follow-up for underfilled quota cells. Veridata Insights can help scope which sequencing fits your population before fieldwork begins.