Where Should You Get Your Medical Thesis Analysed?

The expensive mistake: asking at the wrong time

Medical research needs statistical input at two points, and most people only ask at the second one. The first is the protocol and ethics stage: how many patients you need (a priori power analysis), what your primary endpoint will be, and in what format the data should be collected. The second is the analysis stage, once the data is in.

The problem is that skipping the first stage limits what the second can do. In an underpowered study, "no significant difference was found" may reflect insufficient power rather than an absence of effect — and that can only be reported honestly, not repaired afterwards. Variables missing from a chart review cannot be collected retroactively either. Half an hour before data collection is cheaper than weeks at the end of the thesis.

Where you can get the analysis done

Each of these works in some situations; which fits you depends on your study type and your calendar.

Eight questions to ask before you commit

These are specific to medical research, and how clearly they are answered predicts the comments you will get from reviewers:

• How will you analyse time-to-event data? You want to hear Kaplan-Meier curves, the log-rank test and Cox regression for risk factors; summarising follow-up data with a simple chi-square discards information.

• How will you choose the cut-off in my diagnostic test study? ROC curve, AUC and the Youden index for the sensitivity-specificity trade-off — plus an awareness that PPV and NPV depend on prevalence.

• Which variables go into the multivariable model? The answer should combine clinical rationale with a univariable screening criterion; "we put everything in" invites overfitting, so the events-per-variable rule is worth asking about.

• How will missing data be handled? Missingness is the rule in chart reviews; "we drop incomplete rows" is not a complete answer on its own — how much is missing and how it will be reported both matter.

• Which reporting guideline will you follow? STROBE for observational studies, CONSORT for trials, STARD for diagnostic accuracy — editors notice this immediately.

• Can you write the sample size justification for the ethics committee? They expect a paragraph explaining the assumed effect size and power, not just a number.

• Will the tables match my target journal's format? Medical journal conventions (n (%), mean ± SD or median [IQR], a p-value column) differ from thesis formatting.

• What happens if a reviewer asks for more analyses? The scope and cost of revisions should be agreed in writing; additional requests during peer review are entirely normal.

Red flags

These are not negotiating points — they are signals about the validity of your study:

• Any hint of "we'll make the p-value significant." Significance cannot be promised; promising it means tests will be repeated or group boundaries shifted until something appears. Even if a committee misses it, peer review often will not.

• A proposal to run dozens of subgroup analyses and report only the significant ones. Multiple comparisons without correction manufacture false positives.

• Being asked for identifiable patient data. Names, national ID or file numbers are not needed for analysis; de-identify before sending.

• Raw SPSS output with the interpretation left to you. What you defend in a thesis is the interpretation, not the output.

• A price quoted without asking about the study design. A chart review and a randomised trial are not the same effort.

• An offer to "write the thesis too." Analysis support has a clear boundary; beyond it lies an academic integrity problem.

What does it cost? (2026 market ranges)

Compiled from public service listings; these are not GetBayes prices. Your own quote comes in writing once the scope is clear.

What to send in your first message

Whoever you contact, sending these in one message gets you a realistic quote and a realistic timeline:

1. 01

   Study type and question

   Retrospective or prospective, case-control or cohort; your primary endpoint and the groups being compared.

2. 02

   De-identified data

   Your Excel or SPSS file with names, IDs and file numbers removed. Missing values and messy coding are fine.

3. 03

   Ethics approval and timeline

   The scope of your ethics approval and your submission or defence date; if you are still at protocol stage, say so and we start with sample size.

4. 04

   Target format

   Your thesis formatting guide or the journal you plan to submit to — tables are built accordingly.

The ethical line, and where we stand

Statistical support is routine in medicine — in many clinical studies a biostatistician is part of the team from the start. The line is clear: the data must be real and yours, and you must understand and be able to defend the methods. A good report therefore states not only the result but the justification for every methodological choice.

GetBayes is one of the options above: we analyse medical research from retrospective reviews to randomised trials, from survival analysis to diagnostic test evaluation, and report it in the table layout medical journals expect, with effect sizes and confidence intervals. The analysis itself usually takes 15 minutes; delivery is same-day, often within hours or even minutes. Revisions driven by an advisor, committee or reviewer are free, and we keep answering questions after delivery. If you are at the ethics application stage, we prepare the sample size calculation together with its justification text.