# Red Flags in Supplementary Materials: What to Look For Supplementary materials are the files a research paper keeps off the main pages. They hold raw data, extended methods, uncropped gels, full statistical output, video records, and long tables. Reviewers see them. Readers often skip them. That habit is expensive, because many of the warning signs of unreliable work live in these files rather than in the polished main text. The most common red flag is a mismatch between what a supplementary file contains and what its caption or the main text claims it contains. The first thing to check is simple. Open every supplementary file, read its caption in the manuscript, and ask whether the file actually shows what the caption says. This one check catches a large share of the problems, because most supplementary failures come from rushed assembly rather than from dishonesty. This guide walks through the supplementary materials red flags worth looking for, explains the mechanism behind each one, orders them from the most common cause to the rarest, and gives you a diagnostic sequence you can follow on any paper. It closes with fixes for authors and prevention practices for anyone who plans to publish. ## Why Supplementary Materials Matter When You Read Research Papers When you read research papers, the main text is an argument. Supplementary materials are the evidence room. Journals such as PLOS, Nature, and Science now expect authors to deposit data and to make methods detailed enough that a trained scientist could repeat the work. Funders, including the NIH, have data-sharing policies that push the same direction. The files are also where the pressure shows. Authors polish the abstract and the figures. The supplementary PDF, the raw dataset, and the analysis code often get assembled in the final days before submission. That is why most red flags in supplementary materials trace back to time pressure, and why the rarer causes are worth learning to recognize precisely because they hide behind the same symptoms. ## Start Here: The First Ten Minutes on Any Paper Follow this opening routine before you judge any result in the main text. 1. Download every supplementary file. Count them against the manuscript's own list. A paper that references Supplementary Table 7 but supplies only 5 tables has a problem before you open anything. 2. Open each file. Confirm the file opens, is not corrupted, and is not a duplicate of another file with a different name. 3. Compare each caption in the manuscript to the actual file content. Look for wrong panel labels, wrong sample numbers, and descriptions of experiments that are not in the file. 4. Check the data availability statement in the main text against reality. If it lists a repository accession, search that repository for the accession number. 5. Scan the extended methods for the key reagents and parameters: cell line sources, antibody catalog numbers and validation, primer sequences, [statistical tests](/blog/guides/statistical-tests-choosing-the-right-one-for-your-data), and the software with version numbers. If all five checks pass, the paper is in better shape than most. If any fail, the sections below tell you what the failure usually means and how far it should shake your confidence. ## What Supplementary Materials Should Contain Standards differ by field, but a well-assembled package for a typical life science paper includes these elements. - **Extended methods.** Reagent sources with catalog numbers, antibody validation data, primer sequences with efficiency data, instrument settings, and software versions. - **Raw or processed data.** Underlying values for every figure, usually as spreadsheets, or a deposit in a public repository such as GEO for expression data, SRA for sequencing reads, or PRIDE for proteomics. - **Uncropped or unprocessed images.** Full gel and blot images with molecular weight markers, and original microscopy fields before any cropping or adjustment. - **Statistical detail.** Exact tests, exact sample sizes, effect sizes, [confidence intervals](/blog/data-analysis/confidence-interval-formula-examples), and the analysis code or at least the software and version. - **Reporting checklists.** CONSORT for trials, STROBE for observational studies, ARRIVE for animal work, and MIQE elements for quantitative PCR experiments. - **Ethics documentation.** Protocol numbers, the approving body, and species or sample details for work with animals or human material. A missing item from this list is not automatically misconduct. Context matters, and the next sections give that context. ## Troubleshooting Table: Symptom, Likely Cause, How to Confirm, Fix | Symptom in the supplementary files | Likely cause | How to confirm | Fix (author) / Action (reader) | |--|--|--|--| | Caption describes panels that are not in the file | Rushed assembly | Read every caption against every panel, one by one | Regenerate the file from the current figure set / Email the corresponding author for the correct file | | Two supplementary figures contain what looks like the same image with different labels | Copy and paste error during assembly | Overlap the images visually, check field numbers and scale bars | Re-export from original data / Flag to the journal editor | | "Data available on request" with no repository deposit | Data never organized for sharing | Check the journal's policy, search repositories by author and dataset | Deposit the data with an accession number / Treat reproducibility as unverified until data arrive | | Raw data files are unreadable, proprietary, or lack a description | Data exported without a codebook | Open the files, look for column definitions and units | Add a plain-language codebook / Ask the authors what each column records | | Sample sizes in supplementary tables disagree with the main text | Numbers updated in one place but not the other | Trace one dataset through table, figure legend, and main text | Reconcile all numbers against the analysis dataset / Ask which number is correct | | Uncropped blots absent, only cropped bands shown | Cropping hid splice sites or irrelevant lanes, or worse | Look for marker lanes, consistent spacing, uniform backgrounds | Post full uncropped images with markers / Weigh the claim lower without them | | Statistical tests named without test statistics, p values without effect sizes | Analysis run casually, output not retained | Look for n, effect size, confidence interval, and software version | Retain full statistical output and publish it / Treat p values alone as weak support | | qPCR section lacks primer sequences and efficiency values | Methods written from memory after the fact | Check for primer lists, reference genes, and melt curve data | Add a primer table and efficiency data / Treat fold-change values as approximate | | Files are dated years before submission or version numbers conflict with the text | Old drafts recycled into the final package | Check file properties and internal dates | Rebuild the package from final materials / Note the discrepancy when contacting authors | | Supplementary methods describe a protocol the results could not have used | The paper was cut from a larger project and methods were not pruned | Compare protocol details to the reported sample sizes and conditions | Trim methods to what was actually done / Ask authors to reconcile the mismatch | | No ethics approval number anywhere in the package | Documentation was never requested or collected | Search the main text and supplements for the approval statement | Add the protocol number and approving body / Contact the institution if animal or human work is involved | | Every file present, internally consistent, and fully documented, but the claim still seems too strong | Rarely, deliberate concealment behind a clean facade | Request raw data directly, attempt to reproduce the analysis | Cooperate with reasonable data requests / Await independent replication | ## Red Flags From Most Common to Rarest, With the Mechanism Behind Each ![Chart that sorts a red flag into assembly error, data gap, or patterned absence, with the confirming step for each.](/article-images/illustrations/red-flags-in-supplementary-materials-red-flag-branches.webp) ### 1. Assembly Errors From Time Pressure The single most common cause of supplementary red flags is that the files are built last, under deadline, by tired people. The mechanism is mundane. The figure set changes during revision, someone edits a caption in the main text and not in the supplement, a table gets updated in one copy of a spreadsheet and not another, and the package is zipped at midnight before the deadline. Watch for these symptoms: - Captions that reference panels labeled 3a through 3e in a file whose panels run only 3a through 3d. - A supplementary table with an n that differs from the n in the main text for the same experiment. - A figure numbering scheme that restarts or skips numbers. - Two files that contain overlapping content under different names. None of these prove anything about the science. They do tell you something about process quality, and process quality correlates with data quality. A team that cannot keep a table consistent across two documents may also struggle with sample tracking. **How to fix it.** Authors should generate the supplementary package from the same source files as the manuscript, in the same revision pass, and should run a caption-to-panel audit as a named checklist item. Readers who find an assembly error should email the corresponding author. Most reply, most fix the file, and the reply itself tells you how a lab handles scrutiny. ### 2. Data Availability Gaps The second most common red flag is a data availability statement that promises data on request, or a repository accession number that leads nowhere. The mechanism here varies. Sometimes data were never deposited because no one on the team had done it before. Sometimes the data sit on a former student's laptop. Occasionally the data cannot support the published claims, and "on request" is a shield. A repository is a permanent public archive. GEO holds [gene expression](/blog/guides/gene-expression) and [functional genomics](/blog/guides/functional-genomics) data, SRA holds sequencing reads, PRIDE holds proteomics results, and many journals require deposits in one of these at submission. A real accession number resolves to real files you can download. How to confirm: - Paste the accession number directly into the repository search box. - Search the repository by author name and organism if no accession is given. - Check whether the journal's author guidelines list data deposition as required or as encouraged. A required-policy journal with "on request" statements is a stronger signal than the same phrase in a journal with no policy. **How to fix it.** Authors should deposit data during analysis, not after acceptance, because deposition forces a format check while the data are still fresh. Readers can and should request data. A polite email with a specific question, such as which replicate was excluded from Figure 2b, gets better results than a broad demand for everything. ### 3. Methods Too Thin to Reproduce Reproducibility depends on details, and the supplementary methods section is where those details belong. The mechanism behind thin methods is usually honest. Authors write for colleagues in their own subfield, assume shared knowledge, and compress. The result is a protocol that insiders can reconstruct and outsiders cannot. Red flags inside extended methods: - Antibodies named without catalog numbers, vendor, species raised in, or validation evidence. The Focal Adhesion Analysis Server documentation and similar community resources emphasize that antibody validation details belong in the record, not in a footnote. - Cell lines described without a source or a note on authentication. Cross-contamination and misidentification are established, widespread problems in cell culture, which is why standard cell culture texts recommend authentication at key points such as line receipt and before publication. - Quantitative PCR results without primer sequences, reference gene justification, or efficiency values. The MIQE guidelines, short for Minimum Information for Publication of Quantitative Real-Time PCR Experiments, set out exactly these requirements, and a qPCR supplement that ignores them is describing a measurement you cannot audit. - Software named without a version. A statistic computed in R version 3 and R version 4 can differ, and pipeline tools change output between versions. **How to fix it.** Authors should write methods against a checklist derived from the relevant reporting standard for their field, and should keep a running protocol document during the project rather than reconstructing it later. Readers can check whether the relevant standard exists and whether the paper claims to follow it. ### 4. Statistical Red Flags Statistics cause a distinct class of problems, and the supplements are usually the only place you can see them. The mechanism is often statistical literacy rather than intent. Many practicing biologists learned statistics informally, and the errors that follow are predictable. Look for these patterns: - Test names without the test statistic itself. A statement like "a t test showed a significant difference, p less than 0.05" tells you almost nothing. The supplement should show the t value, degrees of freedom, effect size, and confidence interval. - Sample sizes that shrink silently. Supplementary Table 2 reports n of 12, the figure reports n of 9, and no exclusion criteria appear anywhere. - Every comparison in the paper sits just under the significance threshold. One p value of 0.049 is common. Twelve of them, with no corrections for multiple comparisons, suggests a flexible analysis path. - No statement of how normality was assessed, which test was chosen, or whether paired and unpaired data were handled correctly. - Analysis code absent. When code exists and runs, it is the strongest form of statistical transparency available. **How to fix it.** Authors should archive the full statistical output and the analysis script alongside the data. Readers should not treat a lone p value as the result. Effect size and confidence interval carry more information about the finding than a threshold crossing does. ### 5. Image Red Flags Images deserve their own category because gel, blot, and microscopy images can be manipulated, and the uncropped originals in the supplement are the main defense. The mechanism ranges from innocent beauty work, such as contrast stretching, to deliberate fabrication. Innocent causes are far more common. Red flags: - Cropped bands with no molecular weight marker and no lane context. - Splice points where two gel images were joined. Look for discontinuities in the background texture and for duplicated bands at different exposures. - Microscopy fields where a scale bar is missing or where two "different" conditions show identical morphology in overlapping regions. - Captions that do not state how many biological replicates an image represents. A single representative blot from three replicates is acceptable only if the number of replicates is stated. **How to fix it.** Authors should post uncropped images with markers for every blot cited in a figure, a practice several major journals now require outright. Readers should compare the uncropped image against the published crop and note anything hidden at the edges. **Safety note.** Supplementary archives occasionally contain executable files or macros. Open supplementary files in a sandboxed or updated environment, and treat unexpected executable attachments in a manuscript package as suspicious rather than as science. ### 6. Version, Dating, and File Integrity Problems The package itself can carry evidence of carelessness. The mechanism is that supplementary files accumulate across drafts, and nobody prunes them. - File modification dates that predate the final experiments described inside them. - Two versions of the same table with different numbers, both left in the package. - File names like "final_v3_REAL_use_this.xlsx", which reveal how the package was assembled. - Extremely large files that are just screenshots of spreadsheets, rather than actual data files a reader could analyze. - PDFs that are scans of text, so the tables cannot be copied or parsed. **How to fix it.** Authors should build the package from a clean folder at submission time and use a consistent naming scheme such as FigureS1, TableS2, MethodsDoc. Readers who find version confusion can usually get the correct file from the authors within a few days. ### 7. Ethics and Reporting-Checklist Gaps Work involving humans or animals carries obligations that surface in the supplement as protocol numbers and checklists. The mechanism behind missing ethics documentation is usually administrative. The approval exists, but nobody was asked for the number, and the journal did not require it. Red flags: - No animal protocol number or approving body anywhere in the paper or the supplement. - Human sample work with no statement on consent or IRB review. IRB stands for Institutional Review Board, the committee that approves human subjects research at US institutions. - A reporting checklist uploaded but filled in with page numbers that do not exist in the manuscript, which indicates the checklist was completed as a formality. - [ARRIVE guidelines](/knowledge/veterinary-medicine/veterinary-research-methods/arrive-guidelines-animal-research-implementation-compliance) for [animal research](/blog/guides/animal-research), or CONSORT for clinical trials, ignored in the methods despite the journal requesting them. **How to fix it.** Authors should collect the ethics statement at project setup, not at submission. Readers who find no ethics statement for animal or human work should raise it with the journal editor, who is the correct channel. ### 8. Deliberate Concealment (The Rarest Cause) The rarest cause behind red flags in supplementary materials is deliberate concealment. Its mechanism is strategic. An author with results that do not fully support a claim provides supplements that look complete but lack the specific items that would allow detection, such as raw data, uncropped images, or the exact analysis script. The tell is that concealment hides behind the symptoms of ordinary sloppiness. That is why a single missing item means little, while a patterned absence matters. Every missing element is the kind that would enable verification, and every present element is the kind that merely impresses. A package with beautiful rendered figures and no underlying data fits that pattern. A package with a messy, working analysis script does not. Confirmation requires asking for raw data or attempting to reproduce the analysis. Independent replication remains the final check, and it is the only check that resolves the question fully. ## A Diagnostic Sequence You Can Follow ![Decision path that starts with whether every supplement file is present and ends with a recommendation to weigh the pattern of gaps.](/article-images/illustrations/red-flags-in-supplementary-materials-supplement-triage.webp) Work through this order on any paper, most informative checks first. 1. **Inventory.** List every supplement the manuscript references. Note missing files immediately. 2. **Open and match.** Confirm each file opens and matches its caption. 3. **Trace one number.** Pick one key result. Follow its sample size from figure legend to supplementary table to methods description. If the same n appears consistently in all three, process quality is high. 4. **Check the data deposit.** Resolve the accession number or note its absence. 5. **Audit the statistics.** Find the full test output for the headline result. Effect sizes, confidence intervals, and code present or absent. 6. **Compare images.** Look at uncropped versions of any image-critical result. 7. **Check compliance items.** Ethics statement, reporting checklist, authentication and validation statements. 8. **Weigh the pattern.** One inconsistency is noise. A pattern of verification-blocking absences is a signal. This sequence takes under an hour for a typical paper and becomes faster with practice. It is also excellent training if you are learning how to read scientific papers critically, because it forces you into the parts of the paper that arguments cannot smooth over. ## How to Fix Red Flags in Supplementary Materials Fixes differ depending on whether you are the author, a reviewer, or a reader. **For authors.** Build the supplement continuously during the project. Deposit data when the analysis stabilizes. Write methods against a reporting checklist. Keep raw statistical output as a habit, the way you keep a [lab notebook](/knowledge/molecular-biology/lab-notebook). Before submission, run the eight-step diagnostic sequence above on your own package. Every red flag you find in your own files is one a reviewer will find later, or worse, one a reader will find after publication. **For reviewers.** Journals increasingly expect reviewers to check the supplement. Open every file. Verify that data availability claims resolve. Compare at least one number across main text and supplement. Raise inconsistencies in your review, in neutral language, because assembly errors are usually innocent and authors fix them readily. **For readers.** Email the corresponding author with one specific, answerable question. Cite the paper in proportion to the strength of its documentation. If you teach or run a lab, use flawed supplements as training material, because reading the evidence room is a skill and skills need examples. ## Prevention Practices That Actually Work - **One source of truth.** Generate figures, tables, and captions from the same files, and regenerate everything after any revision. - **Deposit during, not after.** Data deposition while the project is active forces format decisions early. - **Keep a protocol document.** A living methods file with reagent lots, catalog numbers, and software versions removes the reconstruction problem entirely. - **Archive the analysis.** Code, environment, and raw statistical output, stored next to the data. - **Run a caption audit.** A named person checks every caption against every file before submission. - **Use reporting checklists as working documents.** CONSORT, STROBE, ARRIVE, and MIQE checklists filled in during writing catch gaps while they are still cheap to fix. ## Common Mistakes and Limitations - **Treating any red flag as proof of misconduct.** Assembly errors dominate, and the honest base rate is high. A missing catalog number means the claim is unauditable, not that it is false. - **Skipping the supplement because it is long.** The supplement is often the only place where the analysis is visible. Skim it strategically rather than skipping it. - **Judging image quality by appearance.** A sharp, pretty blot can conceal a splice. Look for markers and context, not polish. - **Assuming "on request" data will arrive.** Requests fail often enough that you should plan your reading of the paper as though the data may never come. - **Contacting authors aggressively.** Hostile emails get ignored. Specific questions get answered. - **Individual cases need judgment.** These patterns describe process quality, not the truth of any single result. For any decision that depends on one paper, read the field broadly and consult colleagues with domain expertise. ## Frequently Asked Questions ### What are the most common red flags in supplementary materials? Caption-to-file mismatches top the list, followed by sample sizes that differ between the main text and the supplements, and data availability statements that point to nothing concrete. All three usually come from rushed assembly rather than dishonesty. ### How do I read a research paper's supplementary files efficiently? Open every file, match each caption to its content, and trace one key number from the figure legend to the supplement. This takes under an hour and tells you more about reliability than rereading the abstract ever will. ### What should I do if supplementary data are missing entirely? Email the corresponding author with a specific, narrow request, and check whether the journal's policy requires deposition. Until the data arrive, treat the relevant claims as unverified rather than as false. ### Are cropped gels always a red flag? Cropping is standard practice and acceptable when the uncropped image is provided in the supplement with molecular weight markers. Cropping with no uncropped original available is the red flag, because you cannot check what was removed. ### How can I tell sloppy work from deliberate concealment? Sloppiness produces random errors, such as typos, mismatched numbers, and leftover files. Concealment produces a patterned absence of exactly the items that enable verification, paired with polished presentation. One inconsistency is noise, a pattern is a signal. ### What tools help me verify a data deposit mentioned in a paper? Go directly to the repository named in the availability statement and search the accession number. Repositories such as GEO, SRA, and PRIDE are public, so a valid accession resolves to downloadable files in minutes. ### Do review papers need supplementary materials? Sometimes. Reviews may carry supplementary tables of the literature screened, full search strings, and exclusion criteria. A systematic review without a searchable supplement is hard to update or audit, which is a legitimate concern when you read one. ### How do I raise a supplementary materials problem with a journal? Write to the editor handling the paper, describe the specific inconsistency, and attach the file in question. Editors act on concrete, verifiable observations, and they are the correct channel for ethics and image concerns, not the authors.