Previum’s sample report, as a Word file and on screen, on a demonstration manuscript: editorial decision, editor letter and two reviewers commenting in the margin.
This is the full Professional report: the manuscript back, with the editorial decision and the editor letter up front, numbered comments from two reviewers, and every objection pinned in the margin to the sentence it criticizes. The Basic report shows one finding from the reviewer, the one that decides, with its quote and its fix, plus the checks: references against eight databases, p values recomputed and figures that contradict each other.
This manuscript is a paper of mine that has already been published, with co-authors’ names removed and some details generalized so it cannot be identified. The data and results are the real ones.
Dear authors: I have read your manuscript alongside the reports of two reviewers, one from the substantive side and one methodologist. I agree with both that the cross-country comparison is of interest and the text reads well, but also that the interpretation runs ahead of what the design can support. The central objection is shared: the emotional difference between the two countries is attributed to their pandemic management without being able to rule out other differences between the samples, and the analyses do not control for sex despite a very unequal composition across countries. On top of that there are nine cross-country comparisons with no correction for multiple comparisons and a worry measure with no reliability evidence. My decision is to invite major revisions: the work is recoverable, but it requires revising the analyses and moderating the interpretation throughout the discussion.
The aim is framed as if pandemic management explained the emotional differences between countries ("to examine to what extent worries explain the differences"), but the cross-sectional, correlational design described afterwards does not support that explanatory reading.
Quoted sentence: «The aim of this study is to analyze emotional state, life satisfaction and pandemic-related worries in adolescents from two countries, and to examine to what extent those worries explain the emotional differences observed between them.»
Why it matters: An editor reads the aim and the method back to back: if the aim’s language promises more than the design can deliver, the manuscript loses credibility before reaching the results.
Fix: Reword the aim in descriptive, associational terms, and reserve the explanatory question for a design that compares pandemic management more directly (for example, with more than two countries or with objective severity indicators for the measures).
The conclusion not only interprets the difference between countries as caused by their pandemic management, it derives a public-policy recommendation from it. This is the longest leap in the manuscript.
Quoted sentence: «We conclude that each country’s socioeconomic situation explains the greater distress observed in one of the samples, and therefore recommend incorporating context-adapted psychological interventions into the response to future pandemics.»
Why it matters: Recommending a policy on the basis of a comparison between two samples uncontrolled for other variables is exactly the kind of overreach a reviewer flags in the first round.
Fix: Drop the policy recommendation and replace it with what does follow: the observed difference justifies a study with more countries and a design that can isolate the effect of pandemic management.
The discussion presents the difference between countries as if it confirmed the causal role of pandemic management, with no limitations section acknowledging alternative explanations and no qualification of the range of explained variance (7% to 30% depending on the variable).
Quoted sentence: «The differences found between the two countries are consistent with their different baseline socioeconomic situation and suggest that the pandemic’s impact on mental health was greater where the starting point was less favorable.»
Why it matters: A missing limitations section is the first thing a trained reviewer looks for, and it stands out here because the reported R² range itself already suggests the effect is not uniform.
Fix: Add an explicit limitations section acknowledging the uncontrolled variables between countries and qualify the heterogeneity of R² across variables.
The title calls this a "comparative study", but the abstract and discussion repeatedly use impact and effect language, which is not consistent with a cross-sectional design.
Quoted sentence: «Four linear regression models were fitted in each country with the three worries as predictors; they explain between 7% and 30% of the variance depending on the dependent variable, with neither sex nor age included as covariates.»
Fix: Unify the language throughout: if the design is cross-sectional, the title, abstract and discussion should speak of association, not impact.
The manuscript’s central comparison is between countries, yet emotional symptoms are measured with a different national adaptation in each, and participants are sorted into severity levels using their own adaptation’s cut-offs. No evidence of measurement invariance between the two subsamples is provided.
Quoted sentence: «Emotional symptoms were measured with a 21-item scale, using each country’s own national adaptation and that adaptation’s severity cut-offs to classify participants into five severity levels, which were then compared across countries.»
Why it matters: Without factorial invariance there is no way to tell whether the observed difference is one of emotional state or of how the instrument behaves in each country, and comparing severity categories defined with different norms turns a difference in norms into a clinical difference.
Fix: Test configural, metric and scalar invariance between the two subsamples before comparing means, reporting fit at each step. If scalar invariance does not hold, compare only what the level attained allows and drop the severity-category comparison or present it as descriptive.
The seven variables are contrasted between countries and, within each country, those same seven between girls and boys: twenty-one t-tests with no correction for multiple comparisons. At alpha = .05 with twenty-one contrasts, the probability of at least one false positive exceeds 65%.
Quoted sentence: «The two countries were compared on the seven emotional and worry variables assessed, and within each country girls and boys were also compared on those same seven, with no correction for multiple comparisons applied.»
Why it matters: Without correction a real finding cannot be told apart from an artifact of the number of tests, and the discussion rests precisely on that set of comparisons.
Fix: Apply a Holm-Bonferroni correction or FDR control (Benjamini-Hochberg) to the whole family of contrasts, report the corrected p alongside the original, and rewrite results and discussion around the comparisons that survive.
The regression models estimating the variance explained by worries include neither sex nor age as covariates, even though the sex composition differs greatly between the two countries (85.2% versus 63.2% girls) and R² ranges from .07 to .30 depending on the variable, with no discussion of why.
Quoted sentence: «Four linear regression models were fitted in each country with the three worries as predictors; they explain between 7% and 30% of the variance depending on the dependent variable, with neither sex nor age included as covariates.»
Why it matters: This is a textbook omitted-variable case: part of the variance attributed to worries may stem from the different sex composition of each sample, and the unexplained variability in R² suggests the model does not fit all variables equally well.
Fix: Fit hierarchical models with sex and age in the first step, report the R² increment contributed by worries alone, and discuss why the model performs better for some variables than others.
The sample is recruited by convenience through schools in each country, with no report of how many schools took part, how they were selected, or the response rate, which severely limits the population each subsample can be generalized to.
Quoted sentence: «A total of 1,078 adolescents aged 12 to 17 took part (M = 15.18; SD = 1.25), 530 from one country and 548 from the other. Girls make up 74.7% of the total, with a markedly unequal composition across subsamples: 85.2% in one and 63.2% in the other.»
Why it matters: Without that information it cannot be judged whether the differences between countries reflect adolescents in each country or only the schools that chose to participate.
Fix: Report the number of schools, the selection procedure and the response rate in each country, and explicitly discuss the population each subsample can be generalized to.
The two standard deviations for that subsample’s life satisfaction cannot both be true: 2.04 in the country table and 6.48 and 7.89 in the sex table, over the same participants and the same variable.
Quoted sentence: «In the table of between-country differences, the standard deviation of life satisfaction for one of the subsamples is 2.04, whereas in the table of sex differences that same subsample shows standard deviations of 6.48 and 7.89 on that same variable.»
Fix: Check the value in the country table: with the by-sex SDs reported, the whole-group SD has to be of that same order of magnitude. Fix the figure and check whether that row’s t and d were computed with the correct value.
In one subsample health worries predict MORE life satisfaction and LESS depression, the opposite of the other subsample and of the stated hypothesis, and the manuscript does not comment on that sign reversal.
Quoted sentence: «In one of the subsamples, health worries positively influence life satisfaction and negatively affect depression and stress, in the opposite direction to the other subsample and to what was theoretically expected.»
Fix: Check collinearity among the three worry types (VIF and inter-predictor correlations) and consider whether this is a suppressor effect. Report zero-order correlations alongside the betas, which is where the reversal shows.
The procedure does not specify how many schools took part in each country or whether ethics approval was the same in both, which matters in a study combining data from two different countries.
Quoted sentence: «The questionnaires were administered online between late May and early June 2020. Participants had previously collaborated with the research team or had expressed interest in doing so, and schools were contacted from there following snowball sampling.»
Fix: Specify the ethics committee(s) that approved the study in each country and whether the consent procedure was identical in both.
Generated with comments from two reviewers and one editor, all simulated, and fact-checked before delivery: no invented quotes or data.
Six published, openly licensed papers run through the Basic report exactly as they are: every critique pinned to its sentence, p values recomputed and references checked against eight databases.
Upload your paper or thesis (Word or PDF) and in a couple of minutes you get your Basic report. Your text is not stored.
Analyze my manuscript for free