Entire Strength
Train hard. Think harder.

Strength Science

What A Control Group Controls In Training Research

Comparison groups in training studies rarely control for attention, expectation or total work, which limits what can be concluded from a difference between conditions.

A fit man performs push-ups using kettlebells indoors, showcasing strength and fitness.
A fit man performs push-ups using kettlebells indoors, showcasing strength and fitness. · Photo via Pexels
Health information notice. General information — not a substitute for professional advice. Read the full disclaimer.

Training studies compare groups, and what the comparison group did determines what the result means. In this field the comparison is frequently weaker than the conclusions drawn from it.

What a control group is for

A control condition exists so that change in the treated group can be separated from change that would have happened anyway.

Participants improve for reasons unrelated to the intervention: they learn the test, they are being observed, and untrained participants improve on almost anything.

Without a comparison, all of that is credited to whatever was being studied, which is why uncontrolled results overstate effects so consistently.

Untrained control groups exaggerate everything

Comparing an intervention against people doing nothing produces large differences, because doing nothing is an unusually low bar in a population that was already untrained.

Such a design can only establish that training beats not training, which is not usually the question being asked.

Comparing two training methods against each other is far more informative and produces much smaller differences, which is one reason those studies are harder to fund and to publish.

Equating work is difficult

To compare two methods fairly, the amount of work has to be matched, and there is no single way to do that.

Matching sets, matching total load lifted and matching time under tension each produce a different pairing, and a study can appear to favor either method depending on the choice.

This is why apparently contradictory results about set structures or rep ranges often reflect different equating decisions rather than genuine disagreement.

Blinding is largely unavailable

Participants know which training they are doing, and so do the people supervising them, so neither expectation nor differential encouragement can be removed.

Novelty compounds it, since the group receiving the new method knows it is new and typically trains with more enthusiasm.

Blinding the person assessing the outcome is achievable and is one of the few safeguards available, which makes its absence worth noticing in a paper.

The within-subject alternative

Some designs train one limb one way and the other limb differently, which controls for diet, sleep, genetics and motivation completely.

The limitation is cross-education, in which training one limb produces measurable change in the untrained one, contaminating the comparison.

It also confines the design to single-limb exercises, which excludes most of the movements the results are then applied to.

muscle damagehypertrophymechanismsresearch
Hiro Tanabe
Sports Science Writer, Entire Strength

Hiro has a background in exercise physiology and a habit of reading the methods section first. He is the reason this site rarely reports a single study as news.

More from Hiro →

Also by Hiro Tanabe

Strength Science

What we still do not know

A survey of the open questions in strength training, which is a more useful thing to read than another confident answer to a settled one.

Hiro Tanabe··3 min read

Programming

Volume: how many hard sets a muscle actually needs

The research on training volume is messier than the confident infographics suggest, but it does support a workable range and a clear principle about how to move within it.

Devon Abara··4 min read

Programming

Writing your own programme

At some point following someone else’s template stops fitting your schedule, your weak points and your recovery. Writing your own is more straightforward than it looks.

Devon Abara··3 min read