Questions & Answers

Byleron QDA brings source management, manual coding, memos, multimedia transcripts, analysis, AI-assisted review, and results preparation into one research project. You can work with different file types without creating a separate analysis environment for each format. Codes, evidence, document metadata, and memos remain connected as you move from reading to analysis and reporting.

What to do

1. Create a project and describe its research purpose.
2. Add research materials and check that each file is readable.
3. Build or import a codebook, then code precise evidence and write memos.
4. Use Analysis to examine patterns and Results Builder to assemble selected outputs.

Select “+ New,” give the project a clear title, and add a short description of what the study examines. Project Guidance can hold the research aim, questions, study context, method, key concepts, evidence preference, audience, and language choices. This context helps you work consistently and gives the AI Assistant appropriate framing without treating project notes as evidence.

What to do

1. Create the project before uploading files.
2. Complete only the guidance fields you can justify; leave uncertain fields blank.
3. Set the project/source language to match the research material.
4. Choose an AI output language only when AI writing should use a different language.

The Sample project is a safe practice project containing example sources, a starter code hierarchy, coded evidence, metadata, and memos. Use it to explore manual coding, Analysis, AI review, and Results Builder before working with sensitive or irreplaceable research material.

What to do

1. Select Sample in the project toolbar.
2. Open files and inspect how evidence is connected to codes.
3. Try changing, adding, and retrieving codes before starting your own project.
4. Delete the Sample project when you no longer need it.

Begin by checking source quality and clarifying the research question. Read a meaningful sample, create provisional codes, define them, and revise the hierarchy before coding the full corpus. Write memos when your interpretation changes. After coding, inspect evidence and patterns rather than relying on frequencies alone. Add only defensible tables, figures, and quotations to Results Builder.

What to do

1. Prepare and verify the source material.
2. Pilot the codebook on a varied subset of documents.
3. Refine definitions, inclusion guidance, and exclusions.
4. Code the corpus and preserve negative, ambiguous, and contradictory evidence.
5. Review patterns with the original context before reporting findings.

Choose the method from the research question, epistemological position, source material, and reporting expectations—not from a software feature. The detailed methods guides explain how Byleron QDA can support content analysis, thematic analysis, codebook and template analysis, framework analysis, comparative work, and visual or multimodal workflows while keeping the methodological decisions with the researcher.

QDA stands for qualitative data analysis. Byleron QDA is a browser-based environment for organizing sources, coding evidence, writing memos, examining analytical patterns, and preparing traceable results for qualitative research.

Select + New, enter a distinctive project title, and create the project. A newly created project becomes the active project. Use the project list to move between studies and the project edit controls to rename or delete one. Each project has its own files, codebook, coding, metadata, memos, Analysis state, and Results Builder draft.

What to do

1. Use one project for sources that belong to the same study and analytical framework.
2. Check the active project name before importing, coding, or opening Analysis.
3. Export a backup before deleting an important project.

Project Guidance records the analytical frame used throughout the study. The description states what the project is about; the aim states what the study seeks to understand; research questions define the inquiry; context identifies the setting and participants; method and analytic approach describe how material will be interpreted; key concepts identify sensitizing ideas; writing style, interpretation level, evidence preference, audience, and language guide later assistance and reporting.

What to do

1. Write only information supported by the research design, protocol, or researcher decisions.
2. Update the fields when the design changes and record important analytical changes in a memo.
3. Distinguish project context from findings: guidance frames interpretation but is not participant evidence.

Byleron QDA supports researchers conducting coding-based qualitative work across text, documents, spreadsheets, PDFs, images, audio, and video. Its strongest fit is qualitative content analysis, thematic analysis, codebook or template analysis, framework analysis, comparative and cross-case analysis, document analysis, visual qualitative analysis, and multisource or multimodal analysis. It can also help organize work associated with grounded theory, narrative, discourse, phenomenological, and conversation-oriented approaches, but those methods require methodological decisions and close researcher interpretation that software cannot automate.

What to do

1. Choose the method from the research question, epistemological position, data, and reporting expectations—not from a software feature.
2. Set up a suitable project description, codebook, cases or metadata, and memo practice before broad coding.
3. Use Analysis and Results Builder to inspect and report the work, while retaining your methodological audit trail.

For qualitative content analysis, code meaningful segments into defined categories and systematically compare the resulting evidence. Categories may be deductive, inductive, or mixed. For thematic analysis, use coding as a step toward developing, reviewing, defining, and interpreting broader themes; a list of generated codes alone is not a completed thematic analysis. Byleron QDA supports both workflows through precise coding, code hierarchies, code definitions, memos, evidence retrieval, comparative views, and editable reporting.

What to do

1. For directed or deductive content analysis, create definitions and inclusion/exclusion guidance before applying the codebook.
2. For conventional or inductive content analysis, develop provisional codes from close reading, compare them, and revise the hierarchy transparently.
3. For thematic analysis, use memos and evidence review to test candidate themes against the full dataset, not only frequent codes.
4. Use AI suggestions as reviewable drafts and make final coding and theme decisions yourself.

Codebook or template analysis begins with an initial set of concepts, applies it consistently, and revises definitions, boundaries, and hierarchy as the study develops. Framework analysis adds systematic charting across cases and domains. Comparative and cross-case analysis examines similarities, differences, convergence, and divergence across selected documents, interview questions, cases, or meaningful metadata groups. Byleron QDA supports these workflows through code definitions, parent/subcode structures, retrieved evidence, metadata-aware comparisons, question-organized comparative coding, and Results Builder tables and figures.

What to do

1. Document the starting template and every meaningful revision in code or project memos.
2. Use metadata only when it represents a meaningful comparison and check group sizes and missing values.
3. For framework-style work, review each case against the same analytical domains rather than relying solely on aggregate counts.
4. For comparative work, inspect evidence from every compared group before reporting a difference or convergence.

Use visual content analysis to code visible features in images, drawings, scanned material, or video frames; use visual thematic analysis when interpreting broader meaning patterns. Audio and video can be transcribed for textual analysis while retaining timestamps or coded ranges that link findings back to the original media. A multisource or multimodal workflow can bring together an interview, drawing, observation, photograph, or recording for the same case and examine convergence, complementarity, discrepancy, or absence across sources.

What to do

1. Use text coding when a source has reliable machine-readable text; use region or range coding for visual or time-based evidence.
2. Code each source in a way that respects its modality before comparing them across a case.
3. Use document metadata or consistent naming to identify sources that belong to the same participant, event, or case.
4. Keep the original media available when timing, gesture, visual context, or voice quality matters to interpretation.

You can work with plain text, DOCX documents, PDF files, XLSX spreadsheets, common image formats, audio, and video. The available coding tools adapt to the material: text selections for machine-readable text, rectangular regions for visual material, transcript selections for recorded speech, and time or frame ranges for media.

What to do

1. Open the project and select “+ File.”
2. Choose the correct file and review any import options.
3. Open the imported file and compare it with the original before coding.

Use a clear header row. Put participant or case identifiers and demographic information in metadata columns, and place each interview question or open-ended survey question in its own response column. Consistent question headings allow Comparative Document Coding and question-based Chat with Data to group answers correctly.

What to do

1. Remove decorative title rows, merged cells, and completely empty columns where possible.
2. Keep one participant or case per row.
3. Use the same question wording or stable question key across related files.
4. Review the import preview and identify metadata and question columns accurately.

The workspace extracts a stable text representation for reading, searching, coding, and project exchange. Headings and paragraphs are retained where possible, but the workspace is not a word processor and does not reproduce every page-layout or styling detail from Microsoft Word.

What to do

1. Open the imported document and compare the paragraph order with the original.
2. Check tables, text boxes, footnotes, and unusual multi-column layouts carefully.
3. Keep the original DOCX as part of your research archive.

Use text coding when the PDF contains a reliable machine-readable text layer. Select the exact words you need and assign codes or memos. Use Draw Region for photographs, charts, handwriting, scanned passages, complex layouts, or any visual area that should be analyzed as an image. Both forms of evidence remain part of the same project and can appear in Analysis and Results Builder.

What to do

1. Try selecting and copying a short sentence to check the text layer.
2. Use text selection for quotations that must remain searchable.
3. Use Draw Region for visual meaning or when text selection is unavailable or unreliable.
4. Review the highlighted area before finalizing the code.

The PDF may contain only scanned page images, may have a damaged or inaccurate text layer, or may restrict text access. You can still code rectangular page regions. If you need searching, exact quotations, or text-based AI assistance, run OCR in a trusted document-processing application, verify the recognized text, and then import the corrected PDF.

What to do

1. Test whether text can be selected in another PDF reader.
2. If it cannot, use Draw Region or prepare an OCR version.
3. After OCR, copy a sample into a text editor and compare it with the visible page.
4. Keep the original scan as the archival source.

First confirm that the original file opens normally on your computer and is not password-protected or damaged. Then check the extension, file size, and browser compatibility. Re-exporting the source to a standard DOCX, PDF, XLSX, PNG, MP3, MP4, or plain-text format often resolves unusual encoding or codec problems.

What to do

1. Refresh the workspace once and reopen the project.
2. Try the file in a current desktop browser.
3. Re-export a copy using a common format without password protection.
4. If the problem continues, report the file type, approximate size, browser, and visible error message without attaching sensitive research material unless requested.

The number of files allowed per project depends on your plan (see “How do the plan limits compare?”). Large PDFs, high-resolution images, and long audio or video files take longer to import and prepare because the workspace extracts text, generates previews, or transcribes the recording before it becomes codeable.

What to do

1. Check your plan’s file limit before importing a large batch.
2. Allow extra time for large PDFs, images, and media files to finish preparing.
3. Keep the browser tab open until an import or transcription confirms completion.

Create concise labels that represent analytically meaningful ideas, not entire questions or vague topics. Add a definition explaining what the code means, inclusion guidance describing when it applies, and exclusion guidance distinguishing it from nearby concepts. Use parent codes for broader analytic domains and subcodes for distinct meanings within them.

What to do

1. Select “+ Code” and enter a clear label.
2. Write a definition that another researcher could apply consistently.
3. Add subcodes only when they represent reusable, distinguishable meanings.
4. Pilot the codebook and merge synonyms or separate overloaded codes.

Select the smallest passage that preserves the meaning you need, then choose Assign Code to Selection. You can assign one or several existing codes, create a code, or attach a memo. The evidence preview should match the selected words before you finish.

What to do

1. Drag over the exact sentence or phrase in the document or transcript.
2. Open the coding dialog or drag the selection onto a code.
3. Select every relevant code; multiple codes may share the same evidence.
4. Choose Done only after checking the quotation and assigned codes.

Open the coded segment and choose Adjust Range or Adjust Evidence. Expand or reduce the selection so it contains the intended meaning, then save it. The codes and memos remain connected to the corrected evidence.

What to do

1. Open the code label or evidence item that points to the wrong range.
2. Choose Adjust Range or Adjust Evidence.
3. Select an exact passage from the authoritative source shown in the dialog.
4. Save and reopen the evidence to verify the result.

After selecting evidence, use the drag handle to drop it onto a code in the code tree. You can also drag codes to reorganize the hierarchy. A drop target is highlighted before the change is applied, helping you distinguish coding evidence from moving a code.

What to do

1. Select evidence and begin dragging from the evidence handle.
2. Drop it on the intended code label to assign that code.
3. To reorganize the codebook, drag a code onto its intended parent.
4. Review the code tree and evidence count after the operation.

Use the pencil beside Codes to open Manage Codes. From there you can add a subcode or memo, edit the label and definition, change the parent, merge equivalent codes, or delete selected codes. Merging moves existing coding assignments to the chosen target; deleting removes the selected code and must be reviewed carefully.

What to do

1. Search or scroll to the code you need.
2. Use Edit for label, definition, color, or parent changes.
3. Use Merge only when two codes should become one analytical concept.
4. Before bulk deletion, check evidence counts and whether child codes are affected.

Memos preserve reasoning that cannot be represented by a code label alone. Project memos record design and reflexive decisions; document memos describe a source or case; code memos define and refine concepts; evidence memos explain a particular selection, region, or media range. Memos remain connected to their subject and can support later interpretation.

What to do

1. Choose the memo location that matches the scope of the note.
2. Give the memo a specific title and record the date or stage when useful.
3. Explain decisions, uncertainties, alternatives, and changes—not only summaries.
4. Review relevant memos before finalizing a codebook or results section.

Search files filters the current project’s file list by title. Search codes filters the visible code hierarchy by code label so you can find, retrieve, edit, or use a code without changing the codebook. Clearing the search restores the complete list.

What to do

1. Enter part of a file title or code label; exact spelling is not normally required.
2. Open the matching item or use its row actions.
3. Clear the query before assuming that a file or code is missing.

Use the workspace search tool to search machine-readable source text. Results identify matching passages that can be opened in their documents. Search is useful for checking terminology, locating known phrases, reviewing uncoded occurrences, and supporting transparent lexical exploration; it is not a substitute for reading context or interpreting meaning.

What to do

1. Choose an exact phrase or a focused search term relevant to the research question.
2. Open results in their sources and examine surrounding context.
3. Treat spelling variants, inflections, translations, and OCR errors as possible missed matches.

Select a code’s evidence count or retrieval action to open Retrieved Evidence. The table lists the source, evidence type, location, code, quotation or preview, and memo. You can include subcodes, open an item in its source, or unassign an incorrect code while preserving the source file.

What to do

1. Decide whether the review should include child codes.
2. Read evidence in context rather than relying only on short previews.
3. Correct ranges or unassign coding that does not fit the definition.

The definition states the analytical meaning of the code. Inclusion guidance identifies the kinds of evidence that qualify. Exclusion guidance identifies similar material that should not receive the code and may point to a neighboring code. Together these fields support consistent manual coding, clearer collaboration, better AI code matching, and a defensible audit trail.

What to do

1. Define the concept rather than merely repeating its label.
2. Add positive and boundary examples observed during pilot coding.
3. Revise guidance when disagreements or ambiguous cases reveal an unclear boundary.

Use a parent code for a broad analytical domain and subcodes for reusable, distinct meanings within it. A hierarchy is useful when it clarifies relationships and supports analysis at both broad and detailed levels. Keep codes at the top level when they do not genuinely belong under a broader concept, and avoid creating a hierarchy only to make the tree look organized.

What to do

1. Check that every child code answers “a type or aspect of what?” in relation to its parent.
2. Move a code by editing its parent or by using supported drag-and-drop hierarchy controls.
3. Choose No Parent (Top level) when the code should return to the root.

Choose the memo location that matches the scope of the thought. Project memos record design-wide and reflexive decisions. Document memos describe a source, case, or interview. Code memos explain a concept and its development. Evidence memos interpret one selected passage, region, frame, or time range. Interview-question memos record literature-informed priorities or instructions relevant to answers for that question.

What to do

1. Attach the memo to the narrowest subject that fully fits the note.
2. Use a title that states the decision, issue, or analytical idea.
3. Update or link later thinking instead of creating many indistinguishable notes.

Open the recording and choose Transcribe Audio or Transcribe Video. Select the available transcription style, such as plain, timestamped, or speaker-aware output. When processing finishes, review the transcript while listening to the original media, then edit speakers, question structure, names, numbers, and specialist terms.

What to do

1. Confirm that the recording plays correctly in the browser.
2. Choose the transcription mode appropriate to your analysis.
3. Keep the page open while preparation, upload, and transcription continue.
4. Use Edit Transcript to correct content and structure before coding.

Recordings that cannot be sent in one request are prepared as smaller parts. The workspace searches for quiet intervals near safe boundaries, cuts at silence rather than fixed sentence positions, uploads the parts sequentially, and restores transcript timestamps to the original recording timeline.

What to do

1. Start transcription normally and allow preparation to finish.
2. Do not close the page while browser preparation or part uploads are running.
3. If no safe silence can be found, prepare a shorter copy or divide the recording manually at a natural pause.

Include timestamps when the timing of speech, pauses, overlap, emotion, gesture, or media navigation is analytically important. They are also useful for audit trails and returning to long recordings. Hide them from evidence and written outputs when they make quotations difficult to read and timing is not part of the interpretation.

What to do

1. Use the Include timestamps control beside audio or video material.
2. Apply one project preference consistently unless the reporting purpose changes.
3. Check exported evidence and Results Builder quotations before submission.

Use Edit Transcript or Assign Speakers and Interview Structure. Give recurring speakers stable names or roles, mark interviewer questions and participant answers, and assign the same question key to equivalent questions across files. This structure allows question-based comparison without changing how the transcript appears as a document.

What to do

1. Correct speaker labels before assigning analytical roles.
2. Mark question headings and the answers that belong beneath them.
3. Reuse stable question keys across participants.
4. Preview detected questions before Comparative Document Coding.

Yes. Select a time range on an audio waveform, a visual or audio range on video, or a video frame region. This is useful for silence, tone, gesture, interaction, visual composition, or moments that are not adequately represented by transcript words.

What to do

1. Drag across the waveform or choose valid start and end times.
2. For video, choose whether the range represents audio or visual evidence.
3. Assign codes and memos, then reopen the evidence to check playback.

No AI suggestion becomes production coding merely because it was generated. Coding drafts remain provisional. You review the proposed code, hierarchy, evidence, definition, and rationale; accept, reject, change, recode, or adjust evidence; and then explicitly finalize accepted work. Results Builder AI also produces editable drafts that must be checked against selected evidence.

What to do

1. Check whether the evidence supports the proposed meaning.
2. Prefer an existing code when its definition genuinely fits.
3. Correct duplicate, vague, or overly broad code proposals.
4. Finalize only after reviewing the complete accepted set.

Project Guidance tells the Assistant what the study is about, which method and interpretation level are intended, how evidence should be used, and which writing language is preferred. The Assistant uses this information as framing, not as evidence. Suggest supported values can propose entries based on existing project information, but you should accept only values that reflect your actual design.

What to do

1. Complete the research aim, questions, context, and method first.
2. Review every suggested guidance value before applying it.
3. Update the guidance when the design or analytic purpose changes.
4. Reopen or refresh the Assistant after major project-context changes if the displayed overview has not updated.

Select a meaningful passage and choose AI Assistant. The Assistant can summarize, explain, suggest provisional codes or subcodes, compare the passage with the current codebook, and help examine alternative or contradictory interpretations. Suggested evidence must match the source before it can become coding.

What to do

1. Select enough context to make the passage understandable.
2. Choose the task that matches your analytic purpose.
3. Inspect suggested code definitions and hierarchy, not only labels.
4. Use Adjust Evidence when the suggested quotation is close but not exact.

The Assistant reads the document in manageable sections while preserving surrounding context. It checks the active codebook, proposes precise evidence, reuses suitable codes, suggests new codes or subcodes for genuine gaps, and harmonizes equivalent proposals across the document before opening the review draft.

What to do

1. Open the intended document and choose AI-Assisted Document Coding.
2. Allow long documents time to finish; processing may continue through several parts.
3. Review coverage, code hierarchy, quotations, and duplicate concepts.
4. Accept, revise, or reject proposals before finalizing.

Question-organized comparison treats each interview question as its own analytic domain. It can create or reuse a concise parent code, examine every supplied answer, divide answers into distinct meanings, and propose reusable subcodes across participants and processing batches. Whole-document comparison instead looks for harmonized themes across selected documents.

What to do

1. Choose Interview questions when the sources share a stable question structure.
2. Choose Whole documents when comparing cases or documents without equivalent questions.
3. Review question coverage and any “coded,” “partly coded,” or uncoded indicators.
4. Use question memos to provide literature-informed guidance or analytic priorities.
5. Check that recurring meanings reuse stable subcodes rather than synonyms.

The proposed quotation could not be aligned exactly with the authoritative source text at the expected location. The interpretation may still be useful, but the system will not silently create coding against uncertain evidence. Differences in punctuation, whitespace, OCR, transcript formatting, or a paraphrased AI quotation can cause this warning.

What to do

1. Open in document and check whether the intended passage exists.
2. Choose Adjust Evidence to select the exact source words.
3. Use Re-code when the proposal needs a new evidence search.
4. Reject the suggestion when no exact supporting passage exists.

Chat with Data answers questions from the selected scope: one document, selected documents, or all detected answers to an interview question. Supported answers include source-labelled quotations that can reopen the original document. Use it to explore themes, differences, contradictions, negative cases, and potential code structures—not as a replacement for systematic coding or source reading.

What to do

1. Choose Document, Compare, or Question before asking.
2. Select the correct documents or interview question.
3. Ask a focused question that can be answered from the chosen evidence.
4. Open cited evidence and verify the surrounding context.
5. When code suggestions are offered, review them before adding codes or assignments.

Keep the project and source unchanged, then resume or retry the saved draft when offered. Large documents and question sets are processed in recoverable batches, so a connection ending does not always mean earlier work was lost. Avoid repeatedly starting new drafts while one is still saved.

What to do

1. Read the error message and check whether Resume, Retry question, or Re-evaluate is available.
2. Confirm that the selected documents still exist and the codebook has not changed unexpectedly.
3. Resume the saved draft before creating another run.
4. If finalization reports that the source or codebook changed, re-evaluate the draft and review affected suggestions.

The Assistant can misread context, propose a quotation that does not exactly match the source (flagged as an invalid evidence anchor), miss meaningful passages, or produce different suggestions on repeated runs. It works from the scope you provide and general reasoning, not from your unstated intentions. Treat every suggestion as a draft that a researcher must verify against the source before it becomes coding, interpretation, or reported text.

What to do

1. Verify every AI-suggested quotation against the original source before accepting it.
2. Expect some variation in wording or emphasis if you rerun the same AI task.
3. Do not treat an AI summary as evidence of full corpus coverage; open the source for prominent, rare, and contradictory cases.
4. Disclose AI assistance in your methodology in line with your institution’s or journal’s policy, describing which tasks used it and how outputs were verified.

Suggest supported values examines the project information and available text-bearing sources, then proposes values only for empty guidance fields it can support. Suggestions remain editable and should be reviewed before saving. The Assistant should leave a field unresolved when the project does not provide a defensible basis rather than inventing a method, audience, or research decision.

What to do

1. Add representative sources or an accurate project description first.
2. Run Suggest supported values and review every proposed field in the panel.
3. Edit, accept, or leave blank any value that does not match the actual study design.

Begin with Overview and the retrieved coded evidence. Check corpus size, file types, code coverage, uncoded or lightly coded sources, and whether the active filters match the intended analytic scope. Frequencies and visual patterns are useful orientation tools, but interpretation should return to exact evidence and document context.

What to do

1. Confirm the correct project and document scope.
2. Review active metadata and code filters.
3. Open evidence behind prominent, rare, and contradictory patterns.
4. Record interpretations and uncertainties in memos before adding outputs to Results Builder.

Metadata filters restrict analysis to documents or cases with selected attributes, such as participant group, site, age band, or data-collection wave. Use them to compare analytically relevant groups and to select balanced evidence, not to imply statistical representativeness that the qualitative sample does not support.

What to do

1. Open Variables and inspect field types and missing values.
2. Select one or more values for the intended comparison.
3. Check the remaining document count before interpreting a chart.
4. Clear filters before moving to a different research question.

Advanced Evidence Query finds coded evidence that meets combinations of code, evidence-type, metadata, and proximity conditions. You can require all selected codes in a file, match any selected code, exclude files containing selected codes, include subcodes, require multiple evidence types, or look for compatible text or time anchors within a chosen distance.

What to do

1. Select the codes and evidence types relevant to the question.
2. Choose All, Any, or Exclude deliberately.
3. Use proximity only when character or time distance has a meaningful interpretation.
4. Run the query, inspect the returned evidence, and add a table only when the result is analytically useful.

The co-occurrence workspace shows codes appearing in the same documents and provides normalized similarity measures such as Jaccard, Dice, and overlap. Use the matrix, map, and pair list to identify candidate relationships, then open the evidence drawer to determine whether the codes describe the same passage, different parts of a case, contrast, sequence, or unrelated topics.

What to do

1. Choose a normalization and minimum tie appropriate to the corpus.
2. Select a pair and inspect shared documents and quotations.
3. Compare raw co-document counts with normalized scores.
4. Describe the relationship only after reading the underlying evidence.

No. Some visualizations are exploratory tools for noticing distribution, accumulation, prominence, trajectories, transitions, outliers, or memo relationships. A figure becomes reportable only when its measure, unit of analysis, ordering, filters, and limitations are appropriate for the research design and are explained in the accompanying text.

What to do

1. Read the method note and axis or score definition shown with the visualization.
2. Check whether document order has a defensible temporal or sampling meaning.
3. Inspect candidate outliers and negative cases in their source context.
4. Export or add the figure only after deciding what claim it can support.

The agreement view compares whether coders applied codes within comparable documents and highlights disagreements for review. It supports discussion and adjudication of code application. Unless a measure explicitly compares matching segment boundaries, document-level code-presence agreement should not be reported as segment-level intercoder reliability.

What to do

1. Confirm that coders worked from comparable source copies and codebook versions.
2. Review disagreements rather than relying only on an overall percentage.
3. Use Adjudication to document the final decision and reasoning.
4. Report the unit of analysis and calculation used.

Coded Evidence is the main evidence-level audit view. It brings coded text, PDF selections and regions, image regions, transcript passages, and media ranges into a comparable table with their source, location, code, memo, and preview. Use it to review coding quality, retrieve quotations, and move from a pattern back to its empirical basis.

What to do

1. Filter or sort to the codes, documents, or evidence types relevant to the current question.
2. Open evidence in its source before making a substantive claim.
3. Correct or unassign coding that does not fit the code definition.

Frequency summarizes how often codes were applied and, where shown, how many documents contain them. It is useful for orienting yourself to the codebook, identifying dominant and rare coding categories, checking uneven coding, and choosing evidence to inspect. Interpret frequency alongside code definitions, document coverage, segment length, and the study design.

What to do

1. Compare evidence counts with document counts so repeated coding in one source is visible.
2. Inspect both frequent and infrequent codes in context.
3. Use cautious language when documents, participants, or response opportunities differ.

Document / Code Distribution compares how coding is spread across sources and codes. Use it to detect heavily or lightly coded documents, see whether a code is concentrated in a few cases or distributed across the corpus, and check whether differences may reflect document length, response opportunity, sampling, or inconsistent coding.

What to do

1. Choose whether the question is about documents, codes, or both.
2. Compare raw counts with document coverage and available source length.
3. Open unusual concentrations and sparse cases before interpreting them.

The Code Distribution Model provides a structured view of how code activity is distributed rather than treating a single count as the whole result. Use it to compare code concentration, document reach, and the balance between frequent local coding and broader corpus coverage. It is most useful for codebook review and exploratory pattern detection.

What to do

1. Identify whether a code is broad across cases or intense in only a few documents.
2. Compare related parent and subcodes using the same scope.
3. Retrieve evidence before assigning substantive meaning to a distribution shape.

Segment Length Distribution shows how large or small coded evidence units are. It is useful for auditing coding granularity, finding unusually broad selections, comparing coding practices, and deciding whether some codes are being applied to precise meanings while others capture entire paragraphs or long transcript turns.

What to do

1. Inspect very long and very short segments rather than assuming either is wrong.
2. Compare lengths within similar evidence types because characters and time ranges are not equivalent units.
3. Adjust evidence when a selection is broader or narrower than the intended meaning.

A Treemap uses area to show the relative coded volume of codes or code groups. It is useful for a compact overview of a large codebook and for spotting imbalance between categories. Use it as an exploratory summary, then inspect the counts, document coverage, hierarchy, and evidence behind large or small areas.

What to do

1. Check what quantity controls rectangle area in the current view.
2. Compare related codes within the same scope and filter state.
3. Avoid interpreting area as conceptual importance without supporting evidence.

Portrait provides a compact visual profile of coding across documents or cases. Use it to compare the composition and concentration of codes, identify cases with distinctive profiles, and select contrasting documents for close reading. The view summarizes assigned coding and should be interpreted with source length, available questions, and metadata in mind.

What to do

1. Compare cases that had comparable opportunities to discuss the topic.
2. Open distinctive or apparently empty areas in their source context.
3. Use metadata carefully to explain the sample, not to claim unsupported group effects.

Maps visualize different relationships in the project. A hierarchy map shows parent and subcode structure; a focused code map expands one code and its connected evidence; a two-case map compares code patterns between selected documents; and a co-occurrence map shows codes linked by the selected co-document association measure. Choose the map whose relationship matches the analytical question.

What to do

1. Select the map type, root code or cases, threshold, and displayed memo or frequency information.
2. Move nodes only to improve readability; layout position does not create an analytical relationship.
3. Open linked evidence and explain what nodes and lines represent before reporting the map.

This is a code-linked comparison. Rows represent the selected main code and its subcodes, while columns represent groups from one metadata field, such as gender, site, class, or participant type. The main-code row aggregates the complete selected hierarchy. Use it when the analytical question concerns how coding appears across one set of document or case groups.

What to do

1. Choose a metadata field whose groups are meaningful for the research question.
2. Choose the code family and decide whether its subcodes should be displayed.
3. Select Document presence, Within-group evidence share, or Coded evidence count and report that measure accurately.
4. Check each group denominator and open supporting evidence before interpreting a contrast.

This is also code-linked, but every column is an intersection of two metadata groups. For example, selecting Gender and Age can produce Female · Age 6, Female · Age 7, Male · Age 6, and Male · Age 7. Rows remain the selected code hierarchy. This view is useful when a one-variable comparison would hide an analytically important subgroup combination.

What to do

1. Select two different metadata variables and a relevant code family.
2. Read the document n below every combined group and the cell count shown with each percentage before comparing groups.
3. Use filters only when an additional scope restriction is required; record the active filters in reporting.
4. Inspect coded evidence within the intersecting groups, including low-frequency and contradictory cases.

This is a sample-composition table and does not use codes or coded evidence. It crosses two document metadata variables to show how documents or cases are distributed across their categories. For example, a Gender by Age cross-tab describes the number or percentage of documents in each Gender–Age combination.

What to do

1. Choose the row and column metadata variables.
2. Choose n only, n (% of total), n (% of row), or n (% of column). Percentage of total is the default for sample description.
3. Check the row totals, column totals, grand N, and any documents excluded because paired metadata are missing.
4. Use Code Patterns by Two Metadata Variables when the question concerns codes rather than sample composition.

Document presence is the percentage of documents in a group containing the code and is often the clearest choice for comparing differently sized groups. Within-group evidence share is the code’s share of all coded segments in that group and describes coding composition. Coded evidence count is the raw number of coded segments and is sensitive to group size, document length, and coding opportunity.

What to do

1. Use Document presence when the primary question is how widely a code occurs across cases.
2. Use Within-group evidence share when comparing the composition of coding inside groups.
3. Use Coded evidence count when raw volume itself is relevant and group denominators are reported.
4. Do not switch between measures without updating the caption and interpretation.

The network summarizes codes that occur next to one another in the encoded sequence of a source. It is useful when source order carries meaning, such as conversation turns, chronological fieldnotes, narratives, or ordered transcript passages. It can suggest transitions to inspect but does not prove a temporal or causal process.

What to do

1. Confirm that document or segment order is methodologically meaningful.
2. Inspect the passages that create a prominent transition.
3. Report the adjacency rule and avoid calling the network temporal when order is arbitrary.

Clouds are exploratory displays of code labels or terms in selected coded evidence. Size reflects the configured frequency measure. Use controls for scope, code, metadata, exclusions, minimum length, shape, and style to inspect vocabulary or code prominence, and use a term to locate or provisionally autocode occurrences only after checking context.

What to do

1. Remove stop words, transcription artifacts, and analytically unhelpful terms.
2. Check whether the cloud represents code labels or source terms and which evidence is included.
3. Return to concordance-style evidence before interpreting a prominent word.

The curve shows how the number of distinct observed codes accumulates as documents are added in the displayed order. It is useful for examining code discovery and whether later documents continue to introduce codes. Interpretation depends strongly on document order, sample design, coding maturity, and unequal source length.

What to do

1. Use a defensible sampling, chronological, collection, or randomized order and record it.
2. Inspect which documents introduce late codes and whether those codes are substantively new.
3. Describe the output as code accumulation unless the study supports a formal saturation claim.

The plot contrasts a code’s calculated prominence with how broadly it is dispersed across documents. It helps distinguish codes that are frequent and widespread from codes that are intense in a small number of sources. Use it to select patterns for closer examination, not to rank themes automatically.

What to do

1. Read the displayed formula and units before comparing points.
2. Inspect evidence for high-prominence, high-dispersion, and concentrated codes.
3. Consider document length, case count, and coding opportunities when explaining position.

The network places cases or documents and codes in two different node sets, with links showing coded presence or strength between them. It is useful for seeing shared and case-specific themes, selecting contrasting cases, and examining how a theme spans the corpus without implying that cases are directly connected to one another.

What to do

1. Confirm whether a node represents a participant, case, document, or code.
2. Use thresholds to reduce visual clutter without hiding important rare cases.
3. Open linked evidence for both typical and distinctive case-theme connections.

This view compares co-occurrence profiles across metadata-defined document groups. It is useful for examining whether candidate code relationships appear similarly or differently across sites, participant groups, waves, or other case attributes. Interpret the profiles descriptively and verify that each group has enough comparable documents and coding opportunities.

What to do

1. Choose a metadata field that represents a meaningful analytical comparison.
2. Check group sizes, missing values, and the same co-occurrence definition across facets.
3. Inspect evidence within each group before describing a difference.

The plot shows how coded themes vary across an ordered series of documents, stages, waves, or cases. It is appropriate when the order has a defensible meaning, such as time, intervention stage, interview sequence, or an explicitly chosen analytical progression. Use it to identify candidate increases, decreases, and shifts for contextual review.

What to do

1. State what the horizontal order represents and why it is meaningful.
2. Check whether changing source length or coding opportunity explains an apparent movement.
3. Review the evidence around turning points before describing a trajectory.

The profile identifies documents whose coding distribution differs from the corpus pattern. Use it to locate candidate outliers and then read those cases closely for contradictory evidence, unusual circumstances, data-quality problems, or genuinely different experiences. A statistical coding-profile outlier is not automatically a methodological negative case.

What to do

1. Open every flagged case and compare it with the proposed finding.
2. Distinguish substantive contradiction from unusual length, missing questions, or inconsistent coding.
3. Record whether the case refines, limits, challenges, or does not affect the interpretation.

The network connects researcher memos with the codes or analytical subjects to which they are attached. It is useful for tracing concept development, identifying under-memoed codes, comparing competing interpretations, and making reflexive decisions visible while preparing findings.

What to do

1. Use meaningful memo titles and attach notes to the correct project, document, code, question, or evidence item.
2. Review clusters for repeated ideas, unresolved tensions, and changes in interpretation.
3. Return to participant evidence before turning a memo-based concept into a finding.

Results Builder is an editable workspace for assembling reportable findings from selected tables, figures, evidence, and analytic writing. It preserves the analytical state of added outputs so you can return to the source, explain the basis of a claim, and prepare a coherent results draft without disconnecting interpretations from evidence.

What to do

1. Add a defensible table or figure from Analysis.
2. Choose evidence that illustrates variation, not only the clearest supporting quotation.
3. Write or generate an editable summary and interpretation.
4. Review numbering, captions, notes, and evidence before downloading DOCX.

Use Add Figure or Add Table beside the relevant Analysis output. The current variant, filters, thresholds, metadata scope, and underlying analytical data are captured with the item. Open Results to inspect the saved output and add its title, caption, note, interpretation, and selected evidence.

What to do

1. Set the intended filters and visual controls before adding the output.
2. Choose the figure or table form that communicates the pattern accurately.
3. Open Results and verify that the captured output matches Analysis.
4. Remove obsolete duplicates when the analysis is revised.

Start from the complete evidence pool associated with the relevant codes, documents, query, or output. Keep quotations that are clear, contextually faithful, and collectively show the range of the finding. Include contrasting, negative, infrequent, or ambiguous cases when they qualify the interpretation. Remove redundant quotations rather than selecting only evidence that confirms the preferred claim.

What to do

1. Review every candidate with document and metadata context visible.
2. Retain enough surrounding text to prevent misleading quotation.
3. Balance typical evidence with meaningful variation or contradiction.
4. Check anonymization and permissions before publication.

Check every generated statement against the selected evidence and the analytical measure represented by the table or figure. Remove unsupported generalizations, causal language, invented participant characteristics, and claims about groups excluded by filters. Preserve uncertainty and negative cases where the evidence requires them.

What to do

1. Verify factual statements, counts, labels, and quotations.
2. Separate description of the output from interpretation of its meaning.
3. Make the unit of analysis and active scope explicit.
4. Edit the text into your own disciplinary and methodological voice.

Results Builder keeps a sequence of captured tables, figures, and finding sections. Reorder items to match the argument, edit titles and captions so they describe the actual analytical scope, add interpretation and methodological notes, review selected evidence, and remove obsolete or duplicate items.

What to do

1. Organize sections around findings or research questions rather than the order in which charts were created.
2. Keep descriptive results separate from interpretation where this improves clarity.
3. Check numbering and cross-references after moving or deleting items.

When supported, Results Builder captures the selected output together with its underlying rows, visual type, active filters, metadata scope, thresholds, selected codes, and other controls needed to explain what the item represents. This preserves the version used in the draft even if you later explore a different Analysis view.

What to do

1. Set the intended scope and controls before choosing Add Figure or Add Table.
2. Verify the captured result immediately in Results Builder.
3. Add a replacement when later analytical changes materially alter the result.

Refresh from Analysis captures the current rows, filters, metadata scope, and visual state again. Your existing title, caption, interpretation, and notes are kept rather than deleted, but they must be reviewed because the underlying output or evidence may have changed.

What to do

1. Use Refresh from Analysis only after intentionally changing the analytical view.
2. Read the preserved writing against the updated figure or table.
3. Regenerate or edit a title, caption, or finding when the revised output changes its meaning.

The selected Analysis view did not yet contain a usable table or a renderable figure when it was added. Results Builder blocks AI writing in this situation so it cannot create text detached from analytical data.

What to do

1. Return to Analysis and make sure the relevant output contains data.
2. Run any required query, calculation, or reliability comparison first.
3. Use Add Figure or Add Table again, or choose Refresh from Analysis for the existing item.

Large analytical outputs are reviewed in evidence batches. The progress card shows completed parts based on saved work, not an estimated timer. After all evidence parts are reviewed, the Assistant combines them into the requested title, caption, or finding.

What to do

1. Keep the Analysis window open while the current request is running.
2. If a run pauses or fails, use Retry rather than creating a duplicate item.
3. Review the saved draft and AI coverage details after completion.

The DOCX download turns the current Results Builder sequence into an editable report draft containing supported titles, narrative text, tables, figures, captions, notes, and selected evidence. It is designed for continued writing and formatting in a word processor, not as an automatically finished manuscript.

What to do

1. Review quotations, identifiers, captions, and figure readability before downloading.
2. Open the DOCX and apply the required journal, thesis, or institutional style.
3. Check anonymization, citations, accessibility, and disclosure of AI assistance before sharing.

When you add a figure or table, Results Builder automatically assembles the eligible coded evidence connected to that output’s codes, documents, filters, and analytical scope. This gives you a complete starting pool rather than asking you to find quotations one by one. Results Builder AI reads the output data and only the quotations that remain included in that pool; excluded quotations are not sent to the AI for that item and are not included in its DOCX evidence section.

What to do

1. Open the Evidence Pool before generating a finding and inspect the document, code, and metadata labels.
2. Keep evidence that is faithful to its context and collectively represents the pattern, variation, uncertainty, or contradiction.
3. Exclude duplicates or passages that do not genuinely support the proposed result, then generate or edit the writing.
4. Re-open the pool after refreshing an item because the eligible evidence may have changed.

Search and the document, code, metadata, and Show controls help you inspect a subset of the Evidence Pool. They do not by themselves remove evidence from AI or export. Include visible and Exclude visible change the inclusion status of the quotations currently shown after those filters are applied. Role and Researcher note let you record why a quotation is supporting, contradictory, negative, uncommon, uncertain, or otherwise analytically important.

What to do

1. Use filters to inspect one code, document, or subgroup at a time.
2. Use Include visible or Exclude visible only after confirming what the current filters show.
3. Mark a meaningful contradiction or boundary case with a role or researcher note rather than silently removing it.
4. Return to Show: All before deciding that the pool is balanced.

Each Results Builder item has editable reporting fields. The title identifies what the figure or table shows; the caption explains its scope or reading; findings provide the analytic narrative; and notes record methodological qualifications or publication details. Regenerate asks the Assistant to draft the selected field from the saved output and the currently included evidence. You can edit any field before or after generation.

What to do

1. Use a specific, descriptive title rather than a sentence beginning with “Figure shows” or “Table shows.”
2. Use the caption to state the unit, scope, measure, or visual reading needed by a reader.
3. Use Findings / Interpretation to distinguish what the output describes from what you infer from it.
4. Add a note when a small subgroup, missing data, non-exclusive codes, or another qualification affects interpretation.

Auto-number keeps figure and table labels in sequence when you add, remove, or reorder items. Turn it off only when you need to manage numbering manually. Include figure data tables adds editable supporting tables for figures that have structured rows when exporting DOCX. It is off by default because many reports need the figure but not a repeated data table; enable it when a journal, supervisor, appendix, or reproducibility plan requires the underlying values.

What to do

1. Keep Auto-number on while the draft is changing.
2. Check labels and cross-references after manual numbering or a major reorder.
3. Enable figure data tables only when they improve transparency or are required for the intended report.

Use native JSON when you need the most complete Byleron QDA backup, including product-specific state. Use REFI-QDA Project Exchange QDPX when moving standards-compatible sources, codes, codings, variables, cases, and ordinary memos to or from another qualitative analysis application.

What to do

1. Create regular native JSON backups while the project is active.
2. Use QDPX before moving to another QDA application.
3. Read the compatibility report before relying on an exchange file.
4. Test-import important exports into a separate project or destination application.

A warning explains that an item has no exact equivalent in REFI-QDA or in the destination application. The export can still be valid. Read whether the item was represented through a standard alternative, retained as a note, exported as an external source, or omitted. Decide whether that difference affects the purpose of the transfer.

What to do

1. Review every warning before sharing or archiving the QDPX file.
2. Check the named documents, memos, media, or workspace features in the destination application.
3. Keep the native project backup and original source files.

Large audio, video, image, or PDF files may be referenced rather than embedded, and a source cannot be embedded when the original upload is no longer available. The compatibility report identifies files that require their original media. Keep the QDPX file and its external media together when moving or archiving the project.

What to do

1. Review embedded and external media counts after export.
2. Copy required original files with the exchange package.
3. Open several media-linked evidence items after import.

Keep the original source files in a secure research archive, create regular native project exports, and retain milestone QDPX exports when interoperability matters. Store backups in an approved encrypted location separate from the live website. Test recovery before a deadline or migration.

What to do

1. Back up after major imports, codebook revisions, coding rounds, and results milestones.
2. Use clear dates and project-version names.
3. Protect backups according to consent, ethics, and institutional data-management requirements.
4. Verify that at least one backup can be imported successfully.

Use + File to add new research material to the active project. Use Import project to restore a Byleron QDA JSON backup or create a project from a compatible QDPX exchange file. Import project can bring a codebook, codings, memos, metadata, and project structure; it should not be used merely to add one ordinary source.

What to do

1. Confirm the active project before using + File.
2. Use Import project when the uploaded file represents an entire project package.
3. Review the import report and inspect several sources and coding assignments afterward.

Choose Import project, select the native JSON backup or QDPX package, review the detected project information and compatibility messages, and allow the import to create a project. After import, verify titles, code hierarchy, evidence locations, memos, variables or metadata, and media availability before continuing analysis.

What to do

1. Keep the original package and source media unchanged until verification is complete.
2. Import into a new project rather than overwriting an active study.
3. Open representative text, PDF, spreadsheet, image, audio, and video evidence where applicable.

Choose Export project and select native JSON for the most complete Byleron QDA backup or QDPX for exchange with another compatible qualitative analysis application. Read the compatibility report, download the package, and store it with any required original or external media in an approved research archive.

What to do

1. Use dated native exports at important project milestones.
2. Use QDPX when interoperability is the goal and verify it in the destination application.
3. Keep original source files and never treat a single web-hosted copy as the only backup.

Select Collaborate beside a project to choose between two approaches. Work together invites a co-researcher into one live project and codebook for discussion, reflexive practice, and consensus; reliability statistics do not apply to this mode. Send to coder creates an isolated coding copy for independent agreement calculation and adjudication. Choose the approach that matches the methodological claim you intend to make.

What to do

1. Open the project you want to collaborate on and select Collaborate.
2. Choose Work together for shared, discussion-based coding on one live project.
3. Choose Send to coder when you need documented, independently produced coding to compare or adjudicate.
4. Do not switch approaches mid-study without noting the change in your method section.

Work together invites a co-researcher by email into the same shared project. Both the owner and invited co-researcher need a plan that includes collaboration. Once accepted, they can work with the documents, codebook, evidence, and memos. Changes are saved to one authoritative project and become visible when collaborators refresh or reload project data; this is not simultaneous character-by-character editing. Only the project owner can rename, delete, or share the project, or revoke access.

What to do

1. Select Collaborate, choose Work together, and enter the co-researcher’s email.
2. Send the invitation. The dialog always provides a secure link you can copy, even when the website mail service rejects the email.
3. The co-researcher opens the link, signs in or registers with the invited address, chooses an eligible plan if needed, and returns to the link to join.
4. Review active and pending collaborators in the same dialog. You can copy a pending link again, cancel it, or revoke active access.
5. Use the project activity log to see who made which change.

Byleron QDA hands invitation email to the mail service configured for your WordPress website. An accepted message is not proof of inbox delivery: the provider may delay it, place it in spam, or reject it because the site has no authenticated transactional email or SMTP configuration. A mail-service error does not cancel the invitation: the dialog keeps a copyable secure link as the reliable fallback.

What to do

1. Confirm the invited address and ask the recipient to check spam or junk folders.
2. Copy the secure link from the Work together dialog and send it through a trusted channel.
3. If the website reports that email was rejected, ask the site administrator to configure transactional email or SMTP and verify the sending domain.
4. If an internal-error reference is shown, give that reference to the site administrator so it can be matched with the PHP error log.
5. Cancel an unused pending invitation before creating a replacement. Sending again to the same address invalidates the earlier link.

Your plan controls how many active or pending collaborators and independent coding rounds a project can have. A pending invitation reserves a place until it is accepted, cancelled, replaced, or expires. Both participants must retain an eligible plan while working in a shared project. Existing research data is not deleted when access is paused by a plan change.

What to do

1. Open Work together to review active and pending invitations.
2. Cancel an invitation that will not be used, or revoke an active collaborator who no longer needs access.
3. Use the account usage panel to review the limits of the current plan, and upgrade if a larger team is required.

Select Collaborate, choose Send to coder, and enter the coder’s email. Both the owner and invited coder need a plan that includes independent reliability coding. The coder receives an emailed link; opening it copies the current documents and a locked codebook into their own account (they sign in or register first if needed), while a frozen snapshot of your own coding is kept for comparison. Coders should apply codes independently until the comparison stage.

What to do

1. Finalize and document the codebook version before sending the round.
2. Select Collaborate → Send to coder and enter the coder’s email address.
3. Let the coder apply codes freely; the codebook itself stays locked so definitions cannot drift mid-round.
4. The coder selects Send Back to Owner when finished; you are notified and can run intercoder reliability immediately.

Open disagreements, read the source context, and discuss whether the difference came from evidence boundaries, code definitions, hierarchy, interpretation, or simple omission. Record the decision and its reasoning. Update the codebook when a disagreement reveals ambiguity that could affect later coding.

What to do

1. Review disagreement patterns before individual cases.
2. Resolve concept-definition problems before boundary details.
3. Document retained disagreement when consensus would hide meaningful interpretive difference.
4. Apply agreed changes consistently to the remaining corpus.

The round timeline helps the project owner see when a project was sent to a coder, returned, reviewed, or completed. Use it as an orientation and audit aid alongside coder instructions, codebook versions, and adjudication memos.

The project owner has full control: renaming, deleting, sharing the project, and revoking access. In Work together, an invited collaborator can work with documents, the codebook, evidence, and memos alongside the owner, but cannot rename, delete, or share the project, or revoke another collaborator. In Send to coder, the coder works independently on their own copy with a locked codebook; they cannot change code definitions or affect the owner’s original project until the round is sent back.

Yes. The project activity log records changes such as edits to files, codes, codings, and memos, including which account made the change and when. Use it alongside the collaboration round timeline to understand how a shared or independently coded project developed.

Yes. Interface language controls menus, buttons, messages, and help content. Project/source language describes the research material. AI output language controls generated summaries, code explanations, and results writing. You can therefore use a Turkish interface, analyze Turkish sources, and request English AI writing.

What to do

1. Choose the interface language in your account language settings when available.
2. Set the project/source language in Project Guidance.
3. Leave AI output language as “Same as project language” or choose another language.

The Assistant uses the scope shown for the chosen task, such as selected text, one document, selected documents, answers to one interview question, the active codebook, relevant memos, or project guidance. Review the “data sent” or scope description before starting. Ordinary manual coding, reading, and Analysis do not require an AI request.

What to do

1. Choose the smallest scope that can answer the question.
2. Remove direct identifiers when they are not necessary for analysis.
3. Follow consent, ethics approval, institutional policy, and the site privacy notice.
4. Do not submit material to an AI service when your data agreement prohibits it.

Deletion removes the selected item and may also remove relationships that depend on it. Deleting a file removes its associated coding and memos; deleting an entire project removes the project’s workspace data. Deleting a code does not delete source files, but its coding assignments may be removed or transferred only when you deliberately merge it.

What to do

1. Export a backup before deleting material that may be needed later.
2. Read the confirmation dialog and check the exact item count.
3. Use Merge instead of Delete when evidence should move to another code.

Use a current desktop version of Chrome, Edge, Firefox, or Safari with JavaScript and cookies enabled. Very large PDFs, media preparation, and complex visualizations benefit from sufficient memory and a stable connection. If controls stop responding, finish or save the current dialog where possible, refresh once, and reopen the project.

What to do

1. Check whether the same action works in a private window without browser extensions.
2. Clear only the site cache before clearing broader browser data.
3. Disable aggressive content blockers for the workspace domain.
4. Record the browser version, action, and exact error when reporting a repeatable problem.

Use the website account and membership pages to view your current plan, change membership, update billing details, review invoices, edit your profile, or change your password. Workspace feature availability and usage allowances will be shown in account settings when membership controls are enabled.

What to do

1. Open the User or Account page from the site navigation.
2. Use the membership actions shown for the active plan.
3. Contact support when payment succeeded but access did not update after signing out and back in.

Explain what you were trying to do, what you expected, what happened instead, and whether the problem repeats. Include the visible error message, file type and approximate size, current browser, plugin version, and the workspace area involved. A screenshot is useful when it does not reveal confidential research material.

What to do

1. Search this Help Center using the exact error or task name.
2. Try the relevant recovery steps once.
3. Use Contact support from Help and describe the problem without including confidential research material.
4. Attach research material only when support explicitly requests it and an approved secure method is available.

The workspace is built for desktop use, where coding, document review, and analysis benefit from a larger screen and precise text or region selection. It is not optimized for phone or small-tablet screens; some coding and analysis tools may be difficult to use accurately on a touch interface.

Eligible paid plans include a 7-day free trial. A payment card is required to start it, but the card is not charged during the trial. Unless you cancel before the seventh day, the selected plan begins and its normal recurring charge is made.

What to do

1. Choose a paid plan and review its price and billing terms at checkout before confirming.
2. Use the membership account page to check the trial end date.
3. Cancel before the trial ends if you do not want the plan to start or your card to be charged.

The current hosting location, subprocessors, transport encryption, retention, and backup arrangements are described in the site’s Privacy Policy and service documentation. Contact support when your institution requires written infrastructure details for a data-protection review.

AI-assisted features send the scope you choose (see “What research material is used when I run an AI task?”) to OpenAI’s API to generate a response. This processing is governed by OpenAI’s standard API terms, which are separate from the consumer ChatGPT terms. Review OpenAI’s current API terms and data usage policy directly, since providers can update them.

You retain ownership of the sources, codes, memos, and analysis outputs you create in your projects, subject to the site’s Terms of Service. The workspace does not claim ownership of your research content.

Cancellation stops the next recurring charge; your current plan, access, and unused allowance remain available until the current billing date. A downgrade works the same way: you continue using the current plan for the rest of its paid billing period, then the lower plan starts at the next billing date.

What to do

1. Use the membership account page to cancel or select a lower plan.
2. Check your billing date and continue using your current plan until that date.
3. Before a lower plan begins, compare your projects and files with its limits and export any backup you need.

Refunds are available for requests made within 7 days of payment. Submit a Refund request through Contact support so the request is routed correctly. Do not send payment-card details in the message.

What to do

1. Open Contact support and choose Refund request.
2. Include the account email, payment date, order or invoice reference if available, and a short reason for the request.
3. Submit the request within 7 days of the relevant payment.

Institution access is arranged directly with us rather than through a fixed self-service plan. We can discuss the number of seats, expected use, invoicing requirements, timing, and any other needs specific to your institution.

What to do

1. Open Contact support and choose Account or membership.
2. State the institution name, expected number of researchers or seats, and your preferred start date.
3. Include invoicing or procurement requirements so we can prepare the appropriate proposal.

Open Plans & Pricing, select the plan that suits your work, and review the amount and billing terms at checkout before confirming. Your membership account shows the plan currently active on your account.

What to do

1. Compare the project, file, AI, transcription, Results Builder, exchange, and collaboration allowances before choosing a plan.
2. Complete checkout using the same account that owns your projects.
3. If a completed purchase is not reflected in the membership account after signing out and back in, contact support.

Plans differ in project and file limits, and in whether AI assistance, Results Builder, project import, REFI-QDA exchange, and collaboration are available. Exact allowances for your account are also shown on the membership page.

What to do

1. Free: 1 project, up to 10 files per project, JSON export only; no AI assistance, Results Builder, import, REFI-QDA exchange, or collaboration.
2. Student: up to 10 projects, 100 files per project, AI assistance, Results Builder with DOCX export, import, REFI-QDA exchange, up to 3 collaborators or coder rounds, and 1 hour of transcription per month.
3. Academic: up to 100 projects, 500 files per project, the same feature set as Student, up to 5 collaborators or coder rounds, and 5 hours of transcription per month.
4. Institution: custom limits arranged directly with us — contact us for a proposal matched to your institution’s size and needs.

The workspace provides technical and organizational features that can support a compliant workflow, including account-level access control, project-level sharing permissions, and an audit history. Hosting and processing details must be checked against the current Privacy Policy. We do not make a GDPR or other regulatory compliance certification on your behalf.

What to do

1. Confirm with your institution’s data protection officer or ethics board whether this workspace is suitable for your specific study.
2. Document the technical and organizational features you rely on for your own compliance review.
3. Apply your own de-identification and access controls where your review requires them.

No. There is no built-in tool that automatically anonymizes or pseudonymizes research material. If your research requires removing or replacing direct identifiers, do this before uploading files, using your own de-identification process.

What to do

1. Remove or replace names, contact details, and other direct identifiers before import.
2. Keep the identifier key — the mapping from real names to codes — outside the workspace and secured separately.
3. Follow your consent form, ethics approval, and institutional data protection guidance for the required de-identification standard.

Before uploading, confirm you have the right to use the material, that consent and ethics approval cover digital analysis, and that any required de-identification is already applied — the workspace does not anonymize files for you. Also check that the file is readable and not corrupted, protected, or an unsupported format.

What to do

1. Verify consent, ethics approval, and any data-sharing agreement cover this use.
2. Remove or replace direct identifiers your review requires you to remove.
3. Confirm the file opens correctly and matches one of the supported types.
4. Keep an original, unmodified copy of the source material outside the workspace.