Dedoose vs. Taguette: Free and Low-Cost Qualitative Analysis Software

Most qualitative data analysis (QDA) software comparisons stop at NVivo, ATLAS.ti and MAXQDA — three capable but expensive, seat-licensed products aimed at well-funded labs and institutions with site licences. That leaves a real gap for the graduate student on a fixed stipend, the single-PI grant with a thin materials budget, or the short pilot project that does not justify a multi-year software commitment. This guide covers that tier: Dedoose, Taguette, and briefly QualCoder and Delve — what each is actually for, what Dedoose’s unusual per-active-month pricing means in practice, what you give up by going free and open source with Taguette, and the data-governance question almost no software comparison addresses: where does your IRB-restricted qualitative data actually live once you upload it to a cloud tool, and what does that mean for your approved protocol.

If your project has an institutional site licence, a large multi-coder team, or genuinely needs the deepest feature set (advanced query tools, framework matrices, literature-review integration), see our NVivo vs. ATLAS.ti vs. MAXQDA comparison instead — this page assumes budget is a real constraint, not a preference.

Quick answer: which one fits your project

If you…	Use
Need to code documents for free, with no institutional budget line at all	Taguette
Want a free tool but need it fully under your own institutional server, not a vendor’s cloud	Taguette (self-hosted) or QualCoder (fully local)
Are combining qualitative coding with quantitative/demographic variables (mixed methods)	Dedoose
Have a distributed team that needs to code the same transcripts concurrently on a modest budget	Dedoose
Need advanced query tools, framework/matrix analysis, or an institutional site licence	NVivo, ATLAS.ti, or MAXQDA — see the full comparison

Dedoose: built for mixed methods, priced by the active month

Dedoose (dedoose.com) is a browser-based, cloud-hosted QDA tool developed originally at UCLA. Its real differentiator against both the expensive incumbents and the free tools is that it was built from the ground up for mixed-methods research — combining qualitative coding of text, audio, video and images with quantitative descriptors (demographics, survey scores, experimental conditions) attached to the same cases, so you can run cross-tabulations of code frequency against a quantitative variable (for example, comparing how often a code appears across age group or study arm) inside the same tool rather than exporting to a separate stats package. NVivo and MAXQDA both have mixed-methods features layered on; Dedoose was designed around the combination from the start, which shows in how naturally case-level quantitative fields sit alongside the code tree.

Pricing model. Dedoose does not sell a perpetual licence or a flat annual seat. It bills per user on an active-month basis: you are only charged for a given user in a billing period if that user actually logs in during it. As of this writing, the standard individual rate is in the high-teens per user per active month, with a lower per-user rate for small groups (2-5 users, billed based on who was active in the prior 30-day period) and a discounted student rate. Larger group accounts move to a different model where users are billed per month regardless of login, though administrators can move users in and out of the group or temporarily disable accounts to avoid charges. Because pricing tiers and exact figures change, verify current rates directly on Dedoose’s own pricing page before budgeting a grant line — but the structural point that matters for planning holds: on a project with a defined field-work window and quiet stretches between waves of data collection, active-month billing can cost meaningfully less over a multi-year grant than a flat annual seat licence charged whether or not anyone logs in that month. The trade-off is unpredictability — a burst of concurrent coding by a large team produces a correspondingly larger bill that month — which makes it easier to budget for a short, intense coding sprint than for a steady multi-year subscription with a fixed annual line item.

Collaborative coding and inter-coder agreement. Dedoose supports concurrent, real-time multi-user coding on the same project since it is cloud-native by design — there is no file-locking or merge step the way there can be with a locally-hosted project passed between coders. It includes a built-in training/test tool that calculates inter-rater reliability (via Cohen’s kappa and pooled kappa across code applications) so a team can establish and document coding consistency before splitting up a full transcript set, which matters for any study reporting a coding-reliability figure in its methods section.

Taguette: free and open source — and what you actually give up

Taguette (taguette.org) is a free, open-source QDA tool released under a BSD-3 licence, built specifically as an accessible alternative for researchers without a software budget. It is genuinely no-cost: no trial period, no seat limit, no feature paywall. You can run it two ways — hosted for free at app.taguette.org, or self-hosted on your own or your institution’s server (a meaningful option in its own right for the data-governance reasons covered below). It imports common document formats (PDF, Word, plain text, HTML, EPUB, ODT, RTF), lets you highlight passages and tag them with codes you define, and exports your project, documents, highlights and codes back out in full — the project’s stated design principle is that your data stays yours.

What you give up relative to Dedoose or the paid incumbents is real, not cosmetic:

• No auto-coding or AI-assisted coding. Every segment is highlighted and tagged manually; there is no pattern-matching, sentiment tool, or auto-code-by-keyword feature.
• Limited quantitative integration. Taguette is a text-coding tool, not a mixed-methods platform — there is no case-level quantitative-variable framework for cross-tabulating codes against demographic or numeric data the way Dedoose supports natively.
• Simpler export and analysis output. Exports are straightforward (codebook, highlights, source documents) rather than the matrix queries, framework tables, or visualization suites NVivo/ATLAS.ti/MAXQDA build in.
• No built-in inter-rater reliability statistics. If your study needs a reported kappa or percent-agreement figure, you will be calculating it outside Taguette from exported coding data.

None of that matters for a large share of qualitative work. A single-coder thematic analysis, a small interview study without a quantitative arm, a course assignment, or a pilot study meant to inform a later, better-funded main study are all cases where Taguette’s feature set is genuinely sufficient — the missing features are ones the project never needed, not corners cut. Where it stops being sufficient is exactly where Dedoose’s differentiators are load-bearing: multi-site concurrent coding at scale, a documented inter-coder-reliability statistic required by a journal or funder, or a genuinely mixed-methods design where qualitative codes need to be cross-tabulated against quantitative variables inside the analysis tool itself.

QualCoder and Delve, briefly

QualCoder is a free, open-source desktop application (not browser-based) that runs entirely on your own machine, with no cloud component at all unless you deliberately sync project files yourself. That makes it the most straightforward option when the governance answer you need is “the data never leaves a device I control” — see the next section. It supports text, audio, image, and video coding, memos, and basic mixed-methods case attributes, with a more traditional desktop-software interface than Taguette’s browser-first design.

Delve is a paid, cloud-based QDA tool positioned between Taguette’s free tier and the full incumbents — subscription-priced, browser-based, with AI-assisted coding suggestions as a notable feature the free tools lack. It is worth a look if auto-coding assistance specifically is the feature driving your decision and full Dedoose-level mixed-methods case management is not required, but verify current pricing and features directly before committing, as this is a smaller, faster-moving product than the others on this page.

Data governance: where does your IRB-restricted data actually live?

This is the section most software comparisons skip, and it is often the more consequential decision than price. Dedoose, Delve, and Taguette’s hosted option (app.taguette.org) are all cloud-hosted — your interview transcripts, field notes, or coded audio are stored on a third party’s servers, not on your institution’s infrastructure. Before you upload human-subjects data to any of them, work through three questions:

• Does your approved IRB protocol and consent language actually permit third-party cloud storage? A protocol that specifies data will be stored on institutional, password-protected servers, or that does not mention cloud vendors at all, may not cover uploading transcripts to a commercial QDA platform without an amendment. This is a common and easily missed gap: the software decision gets made by the research team independently of the protocol language, and the two go out of sync.
• Is there a data use agreement or business-associate relationship with the vendor? If your data includes protected health information or other regulated identifiers, your institution’s data governance or privacy office may require a signed Data Use Agreement (DUA) or, for HIPAA-covered data, a Business Associate Agreement with the vendor before any upload — treating the QDA vendor as the third-party data processor it functionally is, not as a neutral piece of desktop software. Check the vendor’s own security and compliance documentation for what commitments (encryption at rest/in transit, data residency, breach notification, retention/deletion on account closure) they will actually sign to, rather than assuming a cloud tool is compliant because it is popular in your field.
• What does de-identification look like before upload, not after? Where the protocol allows it, uploading transcripts that have already had direct identifiers removed is a materially different risk profile than uploading raw recordings and identified transcripts to a cloud vendor. See our guide on de-identification for what that actually requires under HIPAA Safe Harbor and equivalent frameworks.

The practical decision rule: if your protocol or consent form is silent on cloud storage, or restrictive by default, the compliant move is either to get an amendment approved before adopting a cloud QDA tool, or to default to a locally-hosted option — self-hosted Taguette or QualCoder — where the data genuinely does not leave institution-controlled infrastructure. Choosing software before checking the protocol is a common and avoidable way to end up technically out of compliance with your own approved research plan, discovered only at continuing review or audit.

Export, portability, and lock-in

All four tools let you export coded data, but the completeness of that export varies. Taguette and QualCoder, as open-source and (in QualCoder’s case) fully local tools, export in straightforward, non-proprietary formats with no vendor dependency — there is no scenario where you lose access to your own coded work because a subscription lapses. Dedoose and Delve, as subscription cloud services, generally allow full project export (codebook, coded excerpts, memos, case data), but confirm exactly what is and is not exportable, and in what format, before committing a multi-year project to either — and budget for the fact that if the subscription lapses, ongoing access to the live project (versus your last export) typically lapses with it. For any project with a data-retention obligation beyond the life of the grant or the subscription, treat “export everything, on a schedule, to a location you control” as a standing task, not a one-time step at project close.

Coded output and your data availability statement

If your study is subject to a funder or journal data-sharing requirement, the coded output from any of these tools — the codebook, coding frequencies, and de-identified excerpts — is what typically feeds the qualitative portion of a data availability statement. Qualitative data availability statements are usually more restrictive than quantitative ones by necessity (raw transcripts often cannot be shared openly even de-identified, given re-identification risk in small, richly-described samples), so the statement more often points to the codebook and analytic framework being available on request or in a repository, with underlying transcripts restricted. Whichever tool you use, export a clean, well-documented codebook as a standing deliverable, not an afterthought reconstructed at submission time — see our guide on writing a data availability statement for reproducibility for how to phrase the qualitative-data portion of that statement.

Choosing: the ladder from free to institutional

1. Choose Taguette if: cost is a hard constraint, you are coding alone or with a small team, you do not need auto-coding or built-in mixed-methods case variables, and either the hosted or self-hosted option satisfies your protocol’s data-storage language.
2. Choose QualCoder instead of Taguette if: your governance answer requires data to never leave a device or server you directly control, and you are comfortable with desktop rather than browser software.
3. Choose Dedoose if: your design is genuinely mixed-methods (qualitative codes analyzed against quantitative case variables), you have a distributed team that needs true concurrent coding, you need a documented inter-coder-reliability statistic, and month-to-month active billing fits your budget better than a flat licence — and your protocol/DUA situation permits cloud storage with this vendor.
4. Choose Delve if: AI-assisted coding suggestions are the deciding feature and you do not need Dedoose’s depth of mixed-methods case management.
5. Move up to NVivo, ATLAS.ti, or MAXQDA if: you need advanced query and matrix tools, an institutional site licence is already available to you, or your project’s scale and complexity outgrows what any of the tools on this page are built for — see the full incumbent comparison.

Frequently asked questions

Is Dedoose actually cheaper than NVivo or ATLAS.ti?

Usually, yes, for individual researchers or small teams, because Dedoose bills per active month rather than as an annual seat licence — but the comparison depends entirely on your usage pattern. A team that logs in steadily for a full year can spend more on Dedoose’s monthly model than they would on a discounted academic annual licence for one of the incumbents; a project with short, concentrated coding sprints separated by quiet months usually comes out ahead on Dedoose. Compare based on your project’s actual timeline, not a generic monthly-versus-annual list price.

Can I use Taguette for a mixed-methods study?

You can code the qualitative side in Taguette, but it has no built-in framework for attaching quantitative case variables to coded segments or cross-tabulating the two, the way Dedoose does. Many mixed-methods teams on a Taguette budget manage the quantitative side separately (in a spreadsheet or stats package) and merge the two analytically by hand — workable for a smaller study, more cumbersome as sample size and variable count grow.

Does using a cloud QDA tool automatically violate my IRB protocol?

Not automatically, but it is not automatically compliant either — it depends entirely on what your approved protocol and consent language actually say about data storage and third-party processors, and whether your institution has a data use or business-associate agreement in place with the vendor. Check before uploading, not after; an amendment is generally straightforward if the storage plan needs updating, but using an unapproved storage location is a compliance gap regardless of intent.

What happens to my coded project if I stop paying for Dedoose?

Export your full project (codebook, coded excerpts, memos, case data) before or immediately after any lapse — confirm current export completeness and format on Dedoose’s own documentation, since subscription cloud tools generally do not guarantee ongoing access to the live, interactive project once a subscription lapses, only to whatever you have already exported.

Is QualCoder as capable as Taguette?

They cover similar ground — free, open-source, manual coding of text, audio, image and video — but QualCoder is a local desktop application with no hosted option, while Taguette is browser-first with both a free hosted version and a self-hosted option. Choose based on your governance requirement (local-only versus hosted-with-self-host-option) and interface preference rather than assuming one is strictly more capable than the other.