ANALYSIS SERVICE
Meta-Analysis Services
for Evidence Synthesis
Our meta-analysis services cover the statistical work behind a systematic review: effect size calculation, model selection, heterogeneity, sensitivity analysis, publication bias assessment, and transparent reporting, including a clear answer when pooling is not appropriate.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
Meta-analysis is the statistical combination of effect estimates from studies that ask a sufficiently similar question, quantifying an overall effect, its uncertainty, and how much the studies differ. Evidence synthesis is broader: it includes judging whether studies can be pooled at all, and synthesizing them without a pooled estimate when they cannot. Systematic review statistical analysis covers both, from effect measures and model choice to heterogeneity and sensitivity analysis.
WHEN TO ENGAGE
When Does a Systematic Review Need a Statistician?
Researchers typically seek meta-analysis statistical support once eligible studies are defined and decisions on effect measures, pooling, and bias need statistical judgment. The situations below are the most common.
01 · When Your Included Studies Are Ready to Synthesize Studies have been screened and data extracted, and you need to decide how effects are measured, combined, and presented, or whether pooling is justified at all.
02 · When Studies Report Different Outcomes or Metrics Means, standardized differences, risk ratios, odds ratios, hazard ratios, or proportions may need conversion so that the pooled effect remains interpretable and defensible.
03 · When Heterogeneity Complicates the Pooled Estimate Clinical, methodological, or statistical variation between studies may call for a random-effects model, a prediction interval, moderator analysis, or a decision not to pool.
04 · When Publication Bias Is a Concern Funnel plot asymmetry, selective reporting, or very few studies call for an assessment that is careful about what it can and cannot show, rather than a routine test.
05 · When the Review Needs a Specialized Synthesis Network, diagnostic test accuracy, individual participant data, or prevalence questions require models that differ from a standard pairwise meta-analysis.
06 · When a Reviewer or Editor Questions the Synthesis When a meta-analysis for publication draws comments on model choice, dependent effect sizes, sensitivity analyses, or PRISMA reporting, the analysis may need to be revised or justified in writing.
DELIVERABLES
What We Deliver From Your Extracted Data
We turn the extracted study data into a defined set of synthesis outputs, with the effect measures, models, sensitivity analyses, and reporting format agreed before work begins.
Effect Sizes & Data Preparation Study-level effect sizes and variances are calculated or converted from reported statistics, with multi-arm trials and other unit-of-analysis issues documented.
Pooled Estimates & Model Choice The pooling model is selected and justified, and the pooled effect is reported with its confidence interval and, for random effects, a prediction interval.
Heterogeneity & Moderator Analysis Between-study variance, Q, and I-squared are reported, with subgroup analysis or meta-regression interpreted cautiously when few studies are available.
Publication Bias & Sensitivity Analyses Small-study effects are examined where enough studies exist, alongside leave-one-out, risk-of-bias, and alternative-model sensitivity analyses.
Forest Plots, Tables & Figures Forest plots, funnel plots, summary tables, and evidence figures are prepared in publication-ready form for the manuscript, protocol, or thesis chapter.
Reproducible Code & Methods Reporting Analysis scripts and a methods and results description aligned with PRISMA are provided, so the synthesis can be traced and rerun if studies are added.
ANALYSIS & METHODS
Statistical Methods for Evidence Synthesis
The synthesis approach follows from the review question, the outcome metrics, the study designs, and how comparable the included studies are. In some reviews that means advanced pooling models, and in others a structured synthesis without meta-analysis.
01
Systematic Review & Synthesis Planning
We align the synthesis plan with the review protocol, eligibility criteria, outcomes, and extracted data before any model is fitted.
Review Protocol
Data Extraction
Risk of Bias
PRISMA
02
Pairwise Meta-Analysis
We pool comparable effect sizes with common-effect or random-effects models, chosen for the question being asked rather than by default.
Random-Effects Models
Common-Effect Models
Forest Plots
Prediction Intervals
Publication Bias
03
Heterogeneity & Moderators
We examine why effects vary between studies, using study-level moderators and subgroups while respecting how few studies often support them.
Meta-Regression
Subgroup Analysis
Between-Study Variance
Sensitivity Analysis
04
Network Comparisons of Interventions
We compare several interventions by combining direct and indirect evidence, and assess whether transitivity and consistency are tenable.
Indirect Comparisons
Consistency Assessment
Treatment Ranking
Network Plots
05
Diagnostic Accuracy Synthesis
We pool sensitivity and specificity jointly with bivariate or hierarchical models that respect their correlation and the threshold effect.
Bivariate Model
HSROC
SROC Curves
Pooled Sensitivity
Pooled Specificity
06
Participant-Level Synthesis
We synthesize participant-level data with one-stage or two-stage models, allowing consistent adjustment and treatment-covariate interactions.
One-Stage Models
Two-Stage Models
Participant-Level Data
Interaction Effects
07
Prevalence & Single-Group Proportions
We pool single-group proportions with transformations or mixed models suited to the data, including proportions close to zero or one.
Pooled Prevalence
Logit Transformation
Mixed Models
Prevalence Forest Plots
08
Synthesis Without Pooling
When studies are too diverse to combine, we structure the synthesis transparently, with explicit grouping, standardized metrics, and effect direction methods.
SWiM Guideline
Effect Direction Plots
Structured Grouping
Narrative Synthesis
Explore the full family of evidence synthesis methods, from standard pairwise pooling to network, diagnostic, and participant-level approaches.
SOFTWARE
Software for Meta-Analysis & Evidence Synthesis
Meta-analysis can be run in several environments. We use the platform that suits the model required, the reproducibility expected, and the output format your review or journal needs.
CASE EXAMPLE
How We Work in Practice
A clinical research example showing how study design, sample size, and statistical planning are developed before IRB submission.
ILLUSTRATIVE EXAMPLE · Clinical Research · UNITED STATES
A clinical research team was developing a prospective observational study and needed the protocol, sample size justification, and Statistical Analysis Plan finalized before IRB submission.
The study would measure change in a validated patient-reported outcome between baseline and 12 weeks. The protocol identified the general research aim, but the primary endpoint, required sample, missing-data approach, and planned statistical analysis had not yet been specified.
We defined the primary outcome as the within-patient change in score and used G*Power to calculate the required sample. Assuming a standardized paired effect of 0.40, 80% power, and α = .05, the analysis required 52 completed participants. The recruitment target was increased to 62 to allow for approximately 15% incomplete follow-up.
The Statistical Analysis Plan specified the primary paired analysis, an alternative if distributional assumptions were not met, secondary outcomes, variable coding, missing-data handling, and planned sensitivity analyses. The protocol structure was also reviewed against relevant STROBE reporting considerations.
Outcome: The research team received a 62-participant recruitment target, documented sample size justification, defined outcomes and variables, and a Statistical Analysis Plan ready for protocol finalization and IRB submission.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
A Defined Process for Your Research
We assess your review protocol, extracted data, and synthesis requirements before any work begins, so the scope, deliverables, timeline, and fixed project fee are clear from the outset.
01
Share Your Project
Send your review question or protocol, the extracted study data, any risk-of-bias assessments, and existing analysis files. We will let you know if anything else is needed.
02
Receive Your Assessment
Your statistically trained Research Coordinator reviews the materials and confirms whether pooling is appropriate, the planned analyses, deliverables, timeline, and fixed project price.
03
Work Completed
Once you confirm the scope, an in-house statistician carries out the synthesis, completes the agreed sensitivity analyses, and delivers the code, tables, figures, and methods text.
RESEARCHER VOICE
What Researchers Say
“We needed statistical input before finalizing the study protocol, particularly around the sample size and planned analysis. OnlineSPSS reviewed the design, worked through the assumptions with us, and gave us a clear statistical rationale to incorporate into the protocol. The revised methodology was accepted without further statistical queries.”
FREQUENTLY ASKED QUESTIONS
Questions About Meta-Analysis & Evidence Synthesis
Answers to the questions researchers most often ask before a synthesis begins.
What happens when I request an assessment, and how long does it take?
We respond to assessment requests the same day. You can share your review question, the stage of the review, and any extracted data or documents through the assessment form. We then confirm the scope, deliverables, timeline, and one fixed project fee by email, with no commitment to proceed.
Is meta-analysis always appropriate for a systematic review?
No. Pooling is justified only when the studies address a sufficiently similar question, measure comparable outcomes, and are not so diverse in design or population that an average would mislead. When these conditions are not met, saying so is part of an evidence synthesis statistician’s job, and we recommend a structured synthesis without meta-analysis, such as the SWiM approach, instead.
Should I use a fixed-effect or a random-effects model?
The choice rests on the question, not on a heterogeneity test. A common-effect (fixed-effect) model assumes one true effect shared by all studies, so its inference is limited to the included studies, while random-effects models allow the true effect to vary and support wider inference. With few studies, between-study variance is estimated imprecisely, so the estimator and confidence interval method are chosen deliberately and reported with a prediction interval.
Can publication bias be assessed with only a few studies?
Rarely with confidence. Tests for funnel plot asymmetry have low power with fewer than about ten studies, and asymmetry can arise from causes other than publication bias, including heterogeneity. We report what the data can support, use sensitivity analyses where appropriate, and describe the limits of the assessment plainly.
Can you work from data I have already extracted?
Yes. The synthesis usually begins from extracted study-level data, together with the protocol, risk-of-bias assessments, and any existing code. Which stages of the review the work covers is agreed and confirmed during the assessment.
What happens if reviewers question the synthesis after delivery?
Comments on model choice, effect size handling, dependence between estimates, or bias assessment are reviewed against the agreed scope, and the analysis or its justification is updated where appropriate. If new analyses or additional studies fall outside the original scope, we confirm this with you before proceeding.
How do you protect my research data and confidentiality?
Research data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
How does statistical consulting fit with academic authorship?
Our role is to provide the agreed methodological, analytical, and statistical reporting support. The research question, substantive interpretation, conclusions, and overall responsibility for the review remain with you or your research team. OnlineSPSS does not claim any intellectual contribution, and authorship remains yours.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Your Evidence. Pooled Where Justified.
Reported Transparently.
Share your review protocol and extracted data, and we will assess the synthesis required, the expected deliverables, timeline, and fixed project fee before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014