How to read running research: question first, conclusion later

Ask who was studied

When a headline says “runners should…”, find the population: elite athletes, recreational runners, university students, or people with an injury or chronic condition? How large was the sample? How similar were the sex, age, training level and surface to you? A small study of 30 male rear-foot strikers cannot be generalised to every runner.

Look at the design. A randomised controlled trial is better for causal questions, although training studies are hard to blind. Observational studies can show association but cannot prove causation. A cross-sectional study captures one moment and says nothing about sequence. Systematic reviews and meta-analyses pool studies but are still limited by the quality of those studies.

Look at how it was measured

Pace, heart rate, perceived effort, pain, injury and performance all need definitions. Is “injury” self-reported or clinically diagnosed? Is “easy running” defined by heart rate, lactate threshold or feel? Is “performance” a laboratory time trial or a race? Different measurements cannot be compared directly.

Look at effect size and uncertainty, not only “significant”. A statistically significant result may be a few seconds in practice; a wide confidence interval means the estimate is unstable. Check whether there was a comparison condition, randomisation, and funding or conflicts of interest.

From one study to the body of evidence

A single study is one piece of a puzzle, not the final answer. A systematic review of running injuries and training parameters found conflicting evidence on weekly distance, frequency, intensity and recent change, and the authors warned against one “optimal” progression for everyone. A meta-analysis of tapering found that taper generally improves endurance performance, but the exact volume and duration still vary by event and study design.

The certainty differs: the first is closer to “we do not have a universal formula”, the second to “the overall direction has support”. Reading either as “science proves this method works for everyone” goes beyond the data.

Using research in practice

Treat research as a way to revise hypotheses, not as a search for absolute commands. For a new method, ask: what problem does it solve, for whom, at what cost and risk, and can I test it on a small scale and record the response? If a study supports only a small effect, or conflicts with several others, it is not a reason to overturn stable training.

Headlines, social posts and second-hand summaries often omit limitations. Try to find the abstract, design and stated limitations; if you cannot, keep the uncertainty.

Scope and limits

This page is not a statistics course, an individual training prescription or a quality appraisal of a specific paper. Research findings do not replace clinical diagnosis, coaching judgement and personal testing; consult a qualified professional for injury or health concerns.

Sources and scope

The conflicting evidence on running injury and training parameters follows a Journal of Athletic Training systematic review; the tapering example follows a PLOS ONE systematic review and meta-analysis; the multi-factor performance framework follows a PubMed physiology review. This is a reading framework, not a methodological audit of any one paper. Checked 2026-10-28.