Dillman’s Tailored Design Method: The Contact Sequence Behind Higher Survey Response Rates

A survey with a well-designed questionnaire can still fail if too few of the sampled people ever answer it. Don A. Dillman’s Tailored Design Method (TDM) — the successor to his earlier Total Design Method, first published as the fourth edition of Internet, Phone, Mail, and Mixed-Mode Surveys: The Tailored Design Method (Dillman, Smyth & Christian, 2014, Wiley) — is the most widely cited framework for engineering response rates up rather than hoping for the best. It treats non-response as a design problem with a design solution: a coordinated contact sequence, not a single mailing, plus a handful of respondent-facing design choices that together change whether a sampled person answers at all. This guide covers the mechanics researchers actually need to plan — the sequence itself, why it works, what the evidence says about each piece, and how it adapts to email/web administration.

The theory behind the method: social exchange, not persuasion

TDM is not a collection of response-rate “tricks.” It is built on social exchange theory: a person is more likely to complete a survey when the perceived rewards of responding outweigh the perceived costs, and when the requester has established enough trust that the respondent believes those rewards will actually materialize. Dillman’s framework translates that into three design levers researchers control directly:

• Reduce the cost of responding — a shorter, clearly organized, easy-to-follow questionnaire; a postage-paid return envelope; questions ordered so early items are easy and non-threatening.
• Increase the reward — explaining concretely how the respondent’s answer will be used, token (not contingent) financial incentives, and social validation (the study matters, the respondent’s perspective is needed).
• Establish trust — identifiable, credible sponsorship (a university letterhead, a named investigator with contact information), personalization instead of “Dear Participant,” and consistency between what a contact promises and what the respondent actually experiences.

Every element of the contact sequence below exists to serve one of these three levers, which is also why swapping in a single element from TDM (a reminder postcard, say) without the surrounding design usually produces a smaller effect than the published response-rate figures suggest — the method’s evidence base is for the coordinated whole, not any one piece in isolation.

The contact sequence

The core of TDM is a sequence of coordinated contacts with every sampled person, not a single request. The classic mail-survey sequence, and the version most frequently cited in methods sections, has five contacts:

1. Pre-notice — a short letter or postcard, sent a few days before the survey itself, that tells the recipient a request is coming, briefly states why, and signals legitimacy. Its only job is to make the second contact recognizable rather than a cold, easily discarded request.
2. The survey itself — the questionnaire, a personalized cover letter explaining the study’s purpose and importance, a postage-paid return envelope, and (where used) a small token incentive included with the mailing itself, not promised contingent on completion.
3. Thank-you / reminder — sent about a week after the main mailing, addressed to everyone (not just non-respondents, since researchers rarely have real-time confirmation of who has returned a mail survey yet). It thanks those who already responded and gently reminds those who have not.
4. Replacement questionnaire — sent a few weeks after the main mailing to apparent non-respondents, with a fresh cover letter (not a copy of the first) noting the survey may not have arrived or been misplaced, and including another copy of the questionnaire and return envelope.
5. Final contact — a last reminder, sometimes switched to a different mode (a phone call, a priority/certified mailing, or in modern practice a final email) to signal that this really is the last request and to reach anyone the earlier mode may not have.

Two design choices matter as much as the count of contacts:

• Timing, not just repetition. Each contact is spaced to arrive as an earlier one would still be fresh but not so soon that it reads as nagging — roughly a pre-notice a few days ahead, the main survey, a thank-you/reminder about a week later, a replacement several weeks after that, and a final contact some weeks after the replacement. The gaps, not just the total number of touches, are part of the tested design.
• Every contact restates the request in different words — a verbatim repeat of the same letter reads as an automated nag; a varied restatement (different framing of why the study matters, an acknowledgment that time has passed) reads as a genuine, ongoing effort to reach the recipient.

What the evidence actually shows

Dillman’s own research center at Washington State University (the Social & Economic Sciences Research Center, which he helped found) describes early large-scale applications of the method achieving mail and telephone response rates of roughly 60–70% from general populations — substantially above the response rates typical of single-contact mail surveys. A separate, frequently cited evaluation using U.S. Decennial Census test forms in the 1990s tested sixteen candidate design factors individually; five improved response rates measurably: a pre-notice letter, a reminder postcard, a replacement questionnaire, a more respondent-friendly questionnaire layout, and a printed notice that response was mandatory — that combination lifted response from roughly 20–30% (a bare single mailing) to around 65%. Applied studies report a wide range depending on population and topic salience: one published account of the method applied to a survey of family physicians reported a 92.8% return rate, near the top of what has been documented; general population mail surveys more typically land in the 60–70% range Dillman’s own center cites, and any single figure should be read as an upper bound achievable under favorable conditions (a salient topic, a well-defined sampling frame, adequate contact information) rather than a guarantee.

The mechanism behind the two highest-leverage elements is worth stating plainly for a methods section: a pre-notice works mainly by making the real request recognizable rather than novel, which measurably raises the odds it gets opened rather than discarded; and small prepaid token incentives (a nominal amount included with the request, not a larger reward promised only on completion) reliably outperform larger contingent incentives of equal or greater expected value, because the prepaid token operates through the trust/reciprocity mechanism social exchange theory predicts, while a promised-on-completion incentive is functionally a payment for labor and does not build the same trust before the respondent has decided whether to engage at all.

Adapting the sequence for email and web surveys

TDM predates near-universal email and web-survey administration, and Dillman’s later editions (through the 2014 fourth edition, co-authored with Jolene Smyth and Leah Melani Christian) extend the same logic rather than replacing it. The structural translation researchers use in practice:

• Pre-notice becomes a short email (or, where a mailing address is available and the population skews toward lower email engagement, a physical postcard) sent a few days ahead of the survey link.
• The survey contact becomes an email containing the personalized invitation and a direct link, with the same cover-letter content (purpose, sponsorship, time estimate, confidentiality) that a paper cover letter would carry.
• Postage-paid return envelopes have no web equivalent, but the underlying cost-reduction principle carries over as minimizing respondent burden: a short, mobile-friendly, correctly branched questionnaire (see CASRAI’s guide to skip logic and branching), because in a web survey the “cost” that matters most is time and friction, not postage.
• Reminder and replacement contacts become follow-up emails, still varied in wording contact to contact, sent to the full list (since web platforms can in principle track individual completion, some researchers do suppress reminders to confirmed completers — a real advantage over paper that TDM’s original mail design could not use).
• Mixed-mode designs — offering a paper option alongside a primary web survey, or vice versa — introduce mode effects (systematic differences in who selects a mode, and in how the same respondent answers depending on mode) that are a distinct, separate concern from the contact sequence itself; see CASRAI’s guide to survey mode effects and mixed-mode design for how to manage that trade-off.

What does not change across mode: the sequence-not-single-contact principle, the social-exchange logic behind incentives and personalization, and the discipline of varying each contact’s wording rather than repeating a template.

Where TDM stops and other survey-design concerns start

TDM is a framework for maximizing response once the survey exists and the sample is defined — it does not tell a researcher how to word a question, how many response options a scale should have, or how to define a sampling frame. Those are covered separately: see CASRAI’s guide to questionnaire design for item wording and response-option construction, and the sampling methods reference for how the population a contact sequence reaches is defined in the first place. A perfectly executed five-contact sequence cannot rescue a poorly worded questionnaire or a badly drawn sample — the three are complementary decisions, not substitutes for one another.

Frequently asked questions

Do all five contacts need to happen for TDM to work?

The published response-rate gains are for the coordinated sequence, not any single contact in isolation. The Census factor-testing research found value in the pre-notice, the reminder, and the replacement questionnaire individually, but the largest gains came from combining them with a respondent-friendly questionnaire design — treat the sequence as a coordinated system, not a menu to pick one item from.

Do larger incentives always produce a higher response rate than a small prepaid token?

Not reliably. TDM’s evidence base favors a small, unconditional, prepaid incentive included with the request over a larger incentive promised only upon completion, because the prepaid token engages trust and reciprocity before the respondent decides whether to participate, while a completion-contingent reward functions more like payment for labor and does not carry the same trust-building effect.

How does TDM apply to a purely web-based survey with no mailing address?

The same contact-sequence logic applies with email substituting for physical mail at each stage (pre-notice, survey invitation, reminder, replacement, final contact), and postage-paid return envelopes are replaced by minimizing the time/friction cost of the questionnaire itself (see the skip-logic and mobile-friendly design considerations above). The underlying social-exchange principles — reduce cost, increase reward, establish trust — carry over unchanged.

Is a 60–70% response rate realistic for every study?

No. That range comes from Dillman’s own reporting of early large-scale general-population applications and is best read as achievable under favorable conditions rather than guaranteed for any given study. Response rate also depends heavily on topic salience, sponsor credibility to the specific population, and how well-defined the sampling frame is — a full TDM contact sequence applied to a low-salience topic or a poorly defined frame will not automatically reach that range.

See also CASRAI’s guides to questionnaire design, survey mode effects and mixed-mode design, and skip logic and branching in questionnaire design, and the dictionary entries for survey research methods, research questionnaire, sampling methods, and pilot study.