Service overview
Twenty review formats, from systematic and scoping reviews to meta-analyses and landscape reviews. This service is part of that division and is scoped individually around what you need to learn, decide or publish.
A meta-analysis statistically combines results from multiple studies to estimate an overall effect and examine how much results vary. When studies are sufficiently similar, it increases precision and can reveal patterns not visible in single studies. In engineering, meta-analyses are feasible where studies report comparable effect measures, such as strength improvement from an additive, efficiency gains or error reductions, with enough information to calculate variance.
Meta-analysis is easy to do badly. Pooling studies that are not comparable, ignoring heterogeneity or relying on a biased sample of studies yields precise but misleading answers. Manuscript support concentrates on clear statistical reporting, honest handling of heterogeneity and bias and clear plots, with the analysis itself performed by the authors or in the analysis service.
What does a meta-analysis manuscript need to report?
The eligibility criteria, search and selection process, the effect measure and how data were harmonised, the pooling model, heterogeneity statistics and how they were explored, sensitivity analyses and an assessment of possible publication bias.
What this service covers
Problems this service can address
- Reporting of pooling methods that is incomplete
- Heterogeneity discussed without exploration
- Forest and funnel plots that are hard to read
Methods and analytical approaches
- Statistical-methods reporting
- Forest, funnel and subgroup-plot presentation
- Heterogeneity and sensitivity reporting
- Publication-bias discussion
Applications
- Performance meta-analyses
- Effect-comparison papers
- Thesis meta-analytic chapters
Relevant engineering disciplines
Scope, inputs and deliverables
Scope of work
- Assessment of your material: data, methods, draft and target venue
- Structuring and drafting or editing strictly from your supplied results and methods
- Consistency and source checks on text, tables, figures and references
- Journal-formatted files and a submission checklist
- Feasibility check on: extraction dataset and analysis outputs
Typical inputs required from you
- Extraction dataset and analysis outputs
- Protocol or analysis plan
- Draft or notes
Deliverables
- Meta-analysis manuscript
- Plot and table review
- Reporting checklist support
Highly specialised requests are first checked for feasibility from this information. Please do not send confidential data through the public form; describe it first and agree secure transfer afterwards.
Project workflow and method
- 01Enquiry and requirement assessment
- 02Technical scoping and proposal
- 03Execution
- 04Quality review and delivery, with an agreed round of revisions
Read the full project workflow.
In depth: how the work is done
Effect measures and model choice
The manuscript should state the effect measure, such as standardised mean difference, ratio of means or correlation, and justify it. Fixed-effect and random-effects models answer different questions, and random effects are usually more appropriate when studies differ. The estimation method for between-study variance and the software used are reported.
Heterogeneity
Heterogeneity is reported with statistics such as Q, I-squared and tau-squared, and with prediction intervals, which show the range of effects in a new setting. Large heterogeneity requires exploration with subgroup analysis or meta-regression, and the interpretation must acknowledge that the pooled estimate may not represent any particular setting.
Forest plots
Forest plots show each study's estimate with confidence interval and weight, together with the pooled estimate. Clear labelling, ordering and scale help readers. We check that the plot matches the tables and that axes and directions of effect are clear.
Subgroup and sensitivity analyses
Subgroup analyses should be prespecified where possible and few in number, to avoid spurious findings. Sensitivity analyses test robustness by removing influential or low-quality studies and by changing assumptions. The manuscript reports which analyses were planned and which were exploratory.
Publication bias and small-study effects
Funnel plots and tests such as Egger's regression examine whether small studies show systematically different results. These tests have low power with few studies and can mislead when heterogeneity is high. The discussion considers plausible biases and uses cautious language.
Data extraction and effect calculation
The paper describes how means, standard deviations and sample sizes were extracted, how missing information was handled and how multiple outcomes from one study were treated to avoid double counting. Calculations should be reproducible, and data tables can be provided as supplements.
Interpreting and applying results
Interpretation considers practical significance as well as statistical significance, the quality and heterogeneity of the evidence and the applicability to engineering practice. Overgeneralisation from pooled estimates is avoided.
Checking the inputs before pooling
The reliability of a meta-analysis depends on the numbers extracted. Common errors include confusing standard errors with standard deviations, mixing units, using values from different time points and counting one specimen group twice. A second person verifying a sample of extracted values catches most of these. We help authors design an extraction audit and report its result, which reviewers increasingly expect.
Quality control and good practice
Quality-control approach
- Every statement, number and reference is checked against your supplied data and the original sources.
- No data, results, citations or author contributions are invented; gaps are flagged to you.
- Authorship and contribution decisions stay with you and your co-authors, and the use of AI tools is disclosed where journals require it.
- Journal guidelines and reporting checklists are followed; acceptance is never promised.
Common pitfalls this service helps avoid
- Pooling dissimilar studies. A pooled number may have no real-world meaning.
- Ignoring heterogeneity. A single estimate hides variation among studies.
- Many unplanned subgroup analyses. These generate false positives.
- Double counting. Using the same data twice biases results.
What good reporting looks like
Support for a meta-analysis manuscript is documented with a statistical-reporting checklist, a check that figures match tables, a data-extraction description and a bias discussion review. The analysis is carried out by the authors or an analysis service. We do not create data, alter effect sizes or promise acceptance.
An illustrative scenario
Hypothetical example written for explanation. It is not a client case, a result or a testimonial.
Imagine a meta-analysis of the effect of fibre addition on concrete flexural strength across 28 studies. The pooled increase is moderate, but I-squared is high, and a prediction interval includes no effect. Subgroup analysis by fibre type and dosage explains part of the heterogeneity. The manuscript reports these findings and warns that dosage ranges differ among studies and that few studies report long-term behaviour.
Questions to settle before you begin
- Are my studies comparable enough to pool?
- How large is heterogeneity, and what explains it?
- Which analyses were planned in advance?
- What evidence is there of publication bias?
- Does the pooled estimate apply to engineering practice?
Key terms
- Random-effects model
- A model allowing the true effect to vary across studies.
- I-squared
- The proportion of variability due to heterogeneity.
- Prediction interval
- The likely range of effect in a new study.
- Funnel plot
- A plot used to assess small-study effects.
- Meta-regression
- Regression explaining variation among study effects.
Working together on Review Article Writing projects
Review article engagements begin by choosing the review type that fits the question, because systematic, scoping, narrative and bibliometric reviews have different methods and reporting expectations. We ask for the topic, the target journal and any searches or papers you already have. The method is documented so that the review can be audited, and every source cited is checked against the original. We do not invent references or studies, and we state limitations such as database coverage. Acceptance depends on editors and reviewers and cannot be promised.
Frequently asked questions
Do you perform the pooled analysis too?
The analysis itself is the Engineering Meta-Analysis service. This service writes it up.
Can you perform the meta-analysis?
The statistical analysis can be supported through the research services. Manuscript support concentrates on reporting.
How many studies do I need?
There is no strict minimum, but few studies limit heterogeneity and bias assessment.
Which software is used?
Common choices include R packages and dedicated tools; the manuscript should name the software and version.
Can meta-analyses combine experimental engineering data?
Yes, when studies are comparable and report adequate statistics.
Does ENGRexpert guarantee acceptance in a journal?
No. Editors and reviewers decide. We support the clarity, rigour and fit of your manuscript and never promise an outcome.
How is this service scoped and priced?
After your enquiry we review what you need, ask for missing information and send a written scope and quotation. Price depends on complexity, deliverables and timeline. See the project workflow.