Specimen

What a report looks like

Six parts, about 1,200 words, delivered as one document within seven days.

This is a composite. It is written in the format and to the depth of a real report, but it is not a critique of any actual manuscript. We do not publish criticism of identifiable work — the researchers whose papers we would have used are the same people who review for us. In a real report, Part 4 names specific papers with DOIs.

The manuscript

A cross-sectional analysis of urinary phthalate metabolites and thyroid function in ~1,400 adults, drawing on a single spot urine sample per participant. Target journal: a leading environmental health title.

1 · The one-line verdict

"Submitted as it stands, the most likely outcome is desk rejection — not because the analysis is wrong, but because a single spot urine cannot characterise the exposure window your thyroid outcome requires, and the manuscript never addresses that."

2 · Fatal problems

At most three. Each written as what is wrong → how a reviewer will phrase it → what you can actually do. A problem with no route out is a complaint, not a finding.

2.1 One spot urine, a chemical with a half-life measured in hours

Most phthalate metabolites clear within hours. A single sample captures the last meal, not habitual exposure. Your within-person variability is almost certainly large, and the resulting misclassification is non-differential — it biases toward the null. Two of your associations are borderline; a reviewer will say they are indistinguishable from noise introduced by the exposure measure itself.

What to do: if repeat samples exist in any subset, report an intraclass correlation coefficient and use it to bound the attenuation. If they do not, say so in the abstract, not only in the limitations — reviewers treat a limitation buried on page 19 as an attempt to slip it past them.

2.2 Causal language on a cross-sectional design

The abstract says exposure "led to" reduced free T4. Thyroid dysfunction changes behaviour, diet, and body composition, all of which change phthalate exposure. You cannot separate the directions here, and the paper does not try. Rewrite to associational language throughout — including the title.

2.3 Adjusting for BMI

BMI is plausibly on the causal pathway between these exposures and thyroid outcomes. Adjusting for it may be removing part of the effect you are trying to estimate. What to do: present a directed acyclic graph, state the assumed structure, and report both adjusted and unadjusted models. If you keep the adjustment, justify it against the DAG rather than "consistent with prior literature."

3 · The three things a reviewer will attack

Single-pollutant models in a mixtures field. Your metabolites are correlated; separate models for each will read as a decade out of date. Response: add a mixture analysis — BKMR, WQS, or quantile g-computation — as a secondary analysis. You do not have to lead with it, but its absence is now conspicuous.
Multiple comparisons. Nine metabolites against four thyroid measures is thirty-six tests. You report none of this. Response: state the number of tests, and either apply a correction or frame findings as hypothesis-generating. Choose one and defend it; saying nothing is what draws the comment.
Urine dilution. You use creatinine correction without discussion. Creatinine varies with muscle mass, age, and sex — all associated with your outcome. Response: report a specific-gravity or covariate-adjusted standardisation as a sensitivity analysis.

4 · Who you failed to cite

In a real report this section names specific papers, with DOIs, and explains for each why its absence will be noticed. Below is what the reviewer actually looks for.

Work that contradicts you. Two recent analyses of this exposure–outcome pair report null associations in larger samples. You cite neither. A reviewer will conclude either that you did not find them, or that you chose not to engage — both are damaging, and the second more so.
The methods literature you are implicitly using. If you adopt a mixture method in revision, cite its methodological source. Reviewers in this area are frequently the people who developed those methods.
A live controversy you have stepped around. Non-monotonic dose response remains actively contested for endocrine-active compounds. Your linear model may be defensible, but not addressing the question reads as unfamiliarity with the field rather than a considered choice.
Anything you cite that should not be cited. We check your reference list against retraction and expression-of-concern records. Authors are usually unaware, and a reviewer who spots it will question the whole manuscript.

This is the part authors tell us they did not expect, and the reason your reviewer has to be someone publishing in your subfield this year. It is not available from a language service, and it is not in a training corpus.

5 · Where to send it

Your target journal: reachable, but not in this state. The exposure-measurement problem is the kind of thing its editors screen for before review. Fix 2.1 and 2.2 and it becomes a reasonable submission.

If you would rather submit now: two solid field journals would consider this as-is, at meaningfully lower impact. That is a real trade — three months of additional work against a lower-tier placement — and it is your call, not ours. We have set out what each path costs.

6 · Statement of limits

Printed on every report, unaltered:

This report is pre-submission advice from an independent researcher active in the field. It is not peer review, carries no weight with any journal, and is not affiliated with one. It makes no promise or prediction about acceptance. It involves no writing, editing, submission, or handling of your data. The reviewer has declared no collaborative, competitive, or refereeing conflict with this manuscript. Responsibility for the manuscript and the decision to submit remains entirely the authors'.