ACADEMIC WRITING GUIDE
How to Establish Trustworthiness in Qualitative Research (With Worked Examples)
You have interviewed participants, analysed their accounts, and developed themes. Your supervisor then asks: “How can readers trust your interpretation?” Simply saying that the interviews were recorded, your supervisor checked the work, or the findings were “valid and reliable” does not answer that question. You need to show how the research process and the evidence support your conclusions.
Trustworthiness is built through coherent design, ethical data generation, transparent analysis, reflexive consideration of the researcher’s role, and appropriately qualified reporting. This guide explains how to plan those safeguards and write them into a dissertation without inventing procedures or applying a checklist mechanically.
All sample studies and findings below are hypothetical teaching examples. Do not claim to have completed an interview, member check, independent coding exercise, or audit unless it actually happened.
What does trustworthiness mean in qualitative research?
Trustworthiness refers to the grounds on which a reader can judge whether a qualitative account is plausible, well supported, appropriately situated and transparent about how it was produced. A widely used framework associated with Lincoln and Guba distinguishes credibility, transferability, dependability and confirmability. These are helpful questions for many studies, not four certificates to be earned by ticking boxes. Some qualitative traditions use different quality concepts and procedures.
| Criterion | The question it answers | What convincing evidence might look like |
|---|---|---|
| Credibility | Are the interpretations plausible in light of participants’ accounts and the whole dataset? | Careful interviewing, attentive interpretation, contradictory cases, suitable participant engagement |
| Transferability | Can readers assess whether insights may be relevant in another setting? | Thick, responsibly de-identified description of setting, participants and circumstances |
| Dependability | Is the research process understandable and its changes traceable? | Documented recruitment, interview-guide revisions, analysis decisions and dates |
| Confirmability | Can readers trace the claims back to evidence rather than unexplained preferences? | Analytic memos, extracts linked to themes, reflexive notes, defensible interpretive decisions |
Credibility does not mean that every participant must agree with the researcher. Transferability is not statistical generalisation. Dependability does not always mean that another researcher must reproduce an identical interpretation. Confirmability does not require pretending the researcher has no perspective. The meaning and relevance of each criterion depend on your methodology.
Trustworthiness, reliability, validity and ethics: how are they different?
In many quantitative designs, reliability concerns measurement consistency and validity concerns how well evidence supports an intended interpretation. Some qualitative researchers use those terms, while others prefer trustworthiness, authenticity or methodological coherence. The distinction is not an invitation to claim that qualitative work needs less rigour; it needs rigour appropriate to how knowledge is generated.
The key starting point is the design itself. Our research methodology guide explains why the question, research approach, evidence and analysis must fit together. Trustworthiness strategies must likewise fit the chosen approach, rather than being added because a generic template requires them.
Research ethics is related but distinct. Protecting participants, obtaining required approvals and reporting consent truthfully are essential, but ethical compliance by itself does not demonstrate that an interpretation is analytically credible. Conversely, an insightful interpretation does not excuse poor data protection or coercive recruitment.
Nine steps for designing and writing a trustworthy qualitative study
Step 1 — Name your qualitative approach and its assumptions
State whether you are conducting, for example, a qualitative descriptive study, case study, grounded theory study, phenomenological inquiry or a particular form of thematic analysis. Explain why that approach fits the research question. A study seeking to describe service experiences may justify different evidence and analytic decisions from one examining how identities are constructed in conversation.
Do not borrow techniques from an incompatible approach without a reason. For example, reflexive thematic analysis does not generally treat consensus coding or inter-rater reliability as universal signs of quality; interpretive engagement and coherence with the stated analytic approach matter. If you do use multiple coders, explain what their involvement accomplished rather than simply calling it “validation.”
Step 2 — Make recruitment and context transparent
Explain who was eligible, how participants were approached, why the selected setting matters and whose voices may be missing. If you use purposive recruitment, describe the logic and its limits. Give sufficient contextual detail for readers to assess possible relevance elsewhere, while suppressing details that could expose participants.
For example, “ten students were interviewed” leaves readers guessing. A stronger account indicates the type of university, year of study, recruitment process, broad participant characteristics relevant to the question, and any important exclusions. Do not include a uniquely identifying combination of role, age, department and incident.
Step 3 — Collect accounts that can genuinely answer the question
Develop questions and follow-up probes that invite concrete experiences rather than rehearsed yes/no answers. Our qualitative interview-guide article shows how to translate a research question into participant-facing questions and probes without leading participants toward the answer you hope to hear.
Keep contemporaneous notes about changes in the guide, interview conditions, interruptions, and unexpectedly difficult questions. Recordings can preserve speech, but they cannot by themselves prove that the questions were fair, the context understood, or the resulting interpretations sound.
Step 4 — Identify the credibility checks that suit your research
Possible strategies include comparing accounts across relevant participants, examining negative or deviant cases, sustained engagement with data, careful use of contextual documents, peer discussion, and—when suitable—asking participants to respond to summaries or interpretations. Each has limits. Participants may disagree for good reasons; institutional records may reflect different priorities; a peer may challenge an interpretation without independently certifying it.
If you propose member checking, specify what participants will review, when, how disagreement will be handled, and whether the process is ethically workable. Do not write “all participants validated the findings” unless that happened and you can explain what their responses meant.
Step 5 — Keep an audit trail of important decisions
An audit trail is a usable record linking decisions to dates, data and reasons. It need not be a huge binder of screenshots. A small decision log may be enough to show how the interview guide changed, why a code was revised, or why two candidate themes were merged. It should not contain unnecessary personal information or reveal confidential raw data.
| Decision point | Hypothetical record | Why it matters |
|---|---|---|
| Recruitment | Week 2: widened invitations beyond student club officers after early interviewees were mostly highly involved | Shows how selection bias was noticed and addressed |
| Interview guide | Week 3: replaced “Did the tutor support you?” with “Can you describe what happened when you sought help?” | Shows a move away from a leading prompt |
| Coding | Week 5: split “avoidance” into “fear of judgment” and “time constraints” after comparing extracts | Makes interpretive refinement traceable |
| Theme development | Week 6: retained an account that contradicted the prevailing theme | Shows that inconvenient evidence was not hidden |
Step 6 — Practise reflexivity, not a ritual declaration of neutrality
Reflexivity means examining how your experiences, role, assumptions and relationships may affect recruitment, questioning and interpretation. A researcher who is also a student representative, an employee of the clinic studied, or a supervisor of participants has a position worth disclosing and considering.
Instead of “the researcher remained completely unbiased,” explain what you actually did: recorded expectations before interviewing, documented moments when an account challenged those expectations, adjusted probes to avoid prompting agreement, or reviewed interpretations against disconfirming material. Reflexivity is ongoing work, not an admission that the research is inherently defective.
Step 7 — Link interpretations to enough supporting evidence
Interpretations need a traceable route from data extracts to codes, broader patterns and analytic claims. Our thematic-analysis guide explains the movement from transcripts to meaningful themes. Here the additional concern is demonstrating why your chosen interpretation is defensible, including what did not fit neatly.
In the findings chapter, use quotations purposefully: introduce the claim, provide a suitably anonymised excerpt, and explain how the excerpt supports or complicates that claim. One vivid quotation is not proof that everyone felt the same. Where appropriate, discuss variation across participants and explain how the full dataset shaped the theme.
Step 8 — Protect participants while providing adequate detail
Rich descriptions and illustrative quotations can accidentally reveal identities, especially in small organisations or sensitive studies. See Acadelyra’s ethical-considerations guide for planning consent, confidentiality, secure data handling and limitations of anonymity. In a trustworthiness account, explain how the integrity of the evidence was balanced against privacy rather than publishing identifying detail to appear transparent.
If you edit quotations for clarity or privacy, use a consistent and honest convention; never alter the substantive meaning. Explain relevant transcription, translation, anonymisation and data-retention choices as required by your programme and approvals.
Step 9 — Report both safeguards and their limits
Write what was actually planned, what happened, and what it can support. Some procedures may be infeasible: participants may be unreachable for a follow-up, a single researcher may conduct the analysis, or an institution may restrict sharing of raw records. These facts are not automatic disqualifications. They should be described and their consequences considered.
Our limitations-of-a-study guide explains how to state the consequence of a limitation, not merely name it. For instance, a single-site sample may provide strong context-specific insight while limiting the basis for readers to transfer the findings to institutions with different characteristics.
Worked example 1 — Student belonging interviews
Hypothetical research question: How do first-year commuting students at one university describe experiences that helped or hindered their sense of academic belonging?
Design: A qualitative descriptive interview study using purposive recruitment across several faculties, semi-structured interviews, and an explicitly described qualitative analytic approach.
Initial risk: The researcher belongs to the university student-support team. Participants might provide favourable accounts, and the researcher may assume that official support services were beneficial.
Planned safeguards: Recruit outside the support team’s direct caseload; make participation voluntary; use neutral interview prompts; maintain reflexive notes; compare supportive and critical accounts; record theme-development decisions; provide de-identified context.
Model dissertation paragraph — student belonging
“Trustworthiness was addressed through transparent recruitment, ongoing reflexive documentation and systematic examination of contrasting accounts. Eligible first-year commuting students were invited through faculty channels rather than through the researcher’s own support-service caseload, reducing direct pressure to participate. During analysis, the researcher recorded initial assumptions about campus support, checked candidate interpretations against the full set of transcripts, and retained accounts in which students described support services as inaccessible or unhelpful. A dated decision log recorded revisions to codes and theme boundaries. The study setting, recruitment route and relevant participant characteristics were described without disclosing identifying details, enabling readers to judge possible relevance to similar commuter settings. These procedures strengthened the transparency of the analysis but did not eliminate the possibility of social-desirability effects or fully represent students who declined participation.”
Why this works: Each claim names an actual kind of action and the concern it addresses. The paragraph does not pretend that reflexive notes mechanically remove bias, that themes are independently true, or that ten or twelve interviews establish a population-wide prevalence.
Adaptation test: Replace the recruitment channel, researcher position and analytic strategy with your real procedures. If no negative-case review or dated log existed, remove those claims.
Worked example 2 — Employee experiences of hybrid work
Hypothetical research question: How do employees in a medium-sized organisation describe the practical tensions between flexibility, availability and collaboration in hybrid work?
Design: A bounded organisational case study drawing on voluntary interviews and, if authorized, policy documents to understand the organisational context.
Initial risk: The researcher is employed by the organisation; the available interviewees are mostly managers, and reporting detailed job titles could identify dissenting employees.
Planned safeguards: Invite variation across roles, avoid line-manager recruitment, record missing perspectives, compare interview claims with policy documents without assuming documents tell the whole truth, keep an analytic decision log, and generalise role descriptions in quotations.
Model dissertation paragraph — hybrid work
“Credibility was supported by purposive recruitment across broad job categories and by examining accounts that differed from the dominant description of hybrid work as flexible. Where access was permitted, formal work-arrangement policies provided contextual material against which employees’ experiences could be interpreted, rather than independent proof that any particular account was correct. The researcher documented their employee role and used reflexive memos to consider how organisational familiarity shaped questioning and interpretation. An audit log recorded the inclusion of contradictory examples and the development of analytic categories. To protect confidentiality in a small workforce, quotations omitted exact job titles and non-essential event details. Readers should therefore understand the findings as context-specific interpretations shaped by the organisation’s structure and the voices available, not as a representative survey of employee opinion.”
Why this works: The paragraph distinguishes data-source triangulation from simplistic “three sources prove the result” thinking; it reports a real protection for participants and names the transferability limit. A corporate policy may describe official intentions while interviews describe lived practice.
What would weaken it: Writing “two coders achieved 100% agreement, proving objectivity” when no such coding was conducted, or using cross-source agreement as the only criterion without asking whether sources capture different parts of the issue.
Worked example 3 — Patient experience of clinic appointment changes
Hypothetical research question: How do adult patients describe the practical and emotional consequences of repeated outpatient appointment changes at one clinic?
Design: An interview-based qualitative study sampling patients with different appointment experiences, subject to approval and appropriate safeguards.
Initial risk: If recruitment occurs only among patients who attend their next appointments, those who stop attending may be systematically missing. Clinic staff may also influence whether patients feel free to criticise care.
Planned safeguards: Where approval and lawful contact arrangements permit, invite a broader range of eligible patients; separate care decisions from recruitment; offer clear consent and withdrawal procedures; record the consequences of inaccessible perspectives; preserve both positive and negative accounts; anonymise unusual clinical details.
Model dissertation paragraph — clinic appointments
“To strengthen the credibility of the analysis, interview invitations were designed to include patients with varied experiences of appointment rescheduling, including those whose subsequent attendance differed, where approved recruitment routes permitted this. Interviewers clarified that participation would not affect care, and neutral prompts encouraged descriptions of both helpful and difficult experiences. The analyst compared accounts of inconvenience with accounts in which rescheduling enabled access to necessary care and documented why those extracts modified the developing categories. An analytic log linked final interpretations to anonymised transcript excerpts and recorded decisions about excluded identifying details. Description of the service context was retained at a level sufficient for assessment of applicability without revealing individual clinical circumstances. Patients who could not be contacted may still have had different experiences; therefore, the findings should not be interpreted as establishing the frequency of these experiences among all clinic patients.”
Why this works: Recruitment and ethics are presented as real constraints, not promises. The passage distinguishes a plausible account of experiences from an estimate of how common an experience is. A student must not use this paragraph as if recruitment or ethics approval actually occurred.
From weak claims to evidence-backed writing
| Weak wording | Why it is insufficient | Stronger approach |
|---|---|---|
| “The research was valid because interviews were recorded.” | Recording captures what was said, but not necessarily the credibility of questions or interpretation. | Specify transcript checks, context, interpretive comparisons and limits as applicable. |
| “Bias was completely eliminated.” | The researcher’s position and judgment cannot simply be switched off. | Explain reflexive decisions and how interpretations were challenged. |
| “Participants validated all themes.” | This may be false, overly broad or methodologically inappropriate. | State exactly what feedback was sought, from whom, and how it was interpreted. |
| “Results are generalisable to all universities.” | A context-specific qualitative sample is not usually a probability sample. | Describe context richly and explain the limits of transferability. |
How to write the trustworthiness section of a dissertation
Your department may place this material under “Research rigour,” “Trustworthiness,” “Quality,” “Methodological considerations,” or another heading. Some require a separate section; others weave quality decisions into recruitment, data collection and analysis. Follow your programme’s expectations while preserving a coherent explanation.
A practical section usually moves from methodological orientation to the specific risk, actual safeguard, evidence of implementation and remaining limitation. Use past tense only for activities that have genuinely happened; before data collection, describe plans in future tense. Keep the section proportional to the study rather than repeating every procedure from the methods chapter.
An adaptable paragraph template
“This [qualitative approach] study addressed trustworthiness through [specific procedures] chosen to respond to [main risks]. To support credibility, [what was done or will be done and how interpretations were examined]. To support dependability and confirmability, [how decisions, data links and reflexive considerations were documented]. Transferability was addressed by [what non-identifying contextual information was provided]. These procedures [realistically described benefit], although [remaining constraint] limits [specific inference or application].”
Treat this as a structure, not a ready-to-submit claim. If your chosen tradition evaluates quality differently, adapt the labels and rationale. Never insert invented interviews, audit records or participant responses to fill a template.
Common mistakes to avoid
-
Listing the four criteria without identifying any procedure relevant to the study.
-
Calling interviews “valid” merely because an audio recorder was used.
-
Claiming triangulation where only one source exists, or assuming agreement across sources automatically proves truth.
-
Presenting member checking as obligatory or claiming it happened when it did not.
-
Using inter-rater reliability as a universal test of quality in interpretive approaches where it is inappropriate.
-
Writing that the researcher was “completely objective,” instead of discussing positionality and reflexivity.
-
Treating thick description as permission to publish information that could identify participants.
-
Ignoring participants or extracts that complicate a preferred interpretation.
-
Calling findings representative of a whole population without an appropriate basis.
-
Claiming an audit trail but retaining no dated or understandable decision record.
-
Repeating generic assurances rather than showing the link from evidence to interpretation.
-
Confusing ethics approval with proof of analytical trustworthiness.
Frequently asked questions
Are credibility, transferability, dependability and confirmability mandatory?
They are widely used criteria, not universally mandatory labels. Your qualitative approach, discipline and supervisor may favour different quality frameworks. Explain the criteria and procedures you actually use and why they fit your research.
Do I need both triangulation and member checking?
No. Neither is inherently required in every qualitative study. Use a strategy only when it makes methodological and ethical sense, and report what it did and did not establish.
Can one researcher conduct a trustworthy study?
Yes. Trustworthiness does not automatically require a second coder. Clear analysis decisions, reflexivity, critical examination of the dataset and transparent reporting can be especially important in single-researcher work. Whether additional review is useful depends on the design.
Where do I report trustworthiness in my dissertation?
Usually in the methodology chapter or distributed across sampling, data generation and analysis procedures, with relevant limitations explained later. Follow the department’s requirements and avoid duplicating whole paragraphs in several chapters.
What is the difference between an audit trail and a reflexive journal?
An audit trail documents important procedures, changes and analytic decisions. A reflexive journal records how the researcher’s role, assumptions, relationships and reactions may affect the work. They can overlap, but neither substitutes for showing how findings are supported by data.
Does COREQ guarantee a trustworthy dissertation?
No. COREQ helps authors report important aspects of interview and focus-group studies. Reporting completeness is not a certificate that a study was well conducted. SRQR likewise supports transparent reporting, not automatic validation.
Trustworthiness checklist before submission
-
Is the qualitative approach and its quality logic clear?
-
Do the proposed safeguards fit the question, participants and context?
-
Have I explained who was recruited and whose accounts may be missing?
-
Have I documented important interviewing or data-generation choices?
-
Have I considered competing interpretations and contrary evidence?
-
Do interpretations connect clearly to extracts and the wider dataset?
-
Are reflexive considerations specific to my role rather than generic claims of neutrality?
-
Is there a usable record of significant analysis decisions?
-
Have I provided contextual detail while protecting confidentiality?
-
Have I described any member feedback, triangulation or coding review precisely and truthfully?
-
Have I avoided claiming universal transferability or statistical representativeness?
-
Do my limitations state what remains uncertain?
-
Are future plans distinguished from completed procedures?
-
Have I followed my institution’s ethics and reporting requirements?
Final takeaway
Trustworthiness is not a decorative paragraph in which a dissertation claims to be credible, transferable, dependable and confirmable. It is the documented connection between the research question, how accounts were generated, how the researcher interpreted them, and how cautiously the findings are presented. A convincing section lets the reader see the work behind the claims—without overstating what any single technique can prove.
Need guidance with your methodology section?
If you need help organising a coherent methods chapter, explaining qualitative quality criteria or reviewing the clarity of your own writing, request academic help from Acadelyra. Share the brief and institutional requirements so any support remains grounded in your work and consistent with academic integrity.
References and further reading
Editorial note: All worked examples are hypothetical writing models. The suggested procedures must be adapted to the research approach, institutional rules, and work actually planned or completed.