How Studies Are Selected and Graded Here

The core rule

Nothing is cited that has not been retrieved and read.

Not a title from a search result. Not a claim repeated in a blog post. Not a study described in another study’s introduction. If the source has not been opened, it does not appear here.

This is a low bar that is routinely not cleared. A substantial fraction of what “everyone knows” about neck training traces back to a misremembering of something which, once located, does not say that.

Full text versus abstract

Every entry in this library states which was read. The distinction matters more than it sounds.

An abstract omits the sample size problems, the dropout rate, the exclusion criteria, and the limitations section — which is frequently where a paper’s most important sentence lives. A study whose abstract reports a significant result may, in its full text, disclose that the significance came from a per-protocol analysis of ten participants.

That is not hypothetical. It is precisely what happened in the twenty-week military aircrew trial cited throughout this site: the headline effect and the per-protocol effect differ by roughly a factor of three, and only the full text explains why.

Where only an abstract was accessible, the entry says abstract only and no numeric claim is made beyond what the abstract states.

Verification

Each study is checked for:

A working identifier. DOI or PMID, confirmed live rather than copied.

Retraction. A site whose entire proposition is accuracy cannot cite a withdrawn paper.

Correct attribution. Author list, journal, and year checked against the record, not against a secondary citation. Secondary citations propagate errors — a date that is wrong in one place tends to stay wrong everywhere downstream.

How evidence quality is reported

Where a systematic review has graded the evidence, that grade is used. The 2024 British Journal of Sports Medicine review rated resistance training’s effect on neck strength as high quality and its effect on head acceleration as very low to low quality. Those ratings appear on this site as the reviewers assigned them. We do not upgrade evidence the reviewers downgraded.

Single studies are labelled as single studies. One trial in ninety-five aviation cadets is informative and is not a body of evidence. Pages that rest on one study say so.

Effect sizes and confidence intervals are reported where available, rather than a bare “significant” or “not significant.” A confidence interval spanning zero is reported as such.

Journal tier is noted where it is low. One paper in the military literature appears in a journal whose editorial standards are questionable; it is used as a pointer to other work, not as authority, and the entry says so.

Study design, ranked as this library treats it

1. Systematic review with meta-analysis — the best available synthesis, and the level at which the concussion question is currently answered.
2. Randomised controlled trial — the strongest single-study design. Two exist in this literature at meaningful duration.
3. Controlled non-randomised trial — most of the aviation work.
4. Observational and cross-sectional — useful for prevalence and for generating hypotheses, unable to establish causation. Most claims about neck strength and injury rates come from here, and the pages that use them say so.
5. Cadaveric and biomechanical testing — the source of the injury tolerance data. Excellent for structural limits, silent about training.
6. Case reports and expert opinion — not cited as evidence.

What gets omitted

Claims that cannot be verified are left out — not softened, not attributed to “some researchers,” omitted.

This is why some pages here say less than competing pages elsewhere, and why several record that a widely repeated claim has no locatable source. Recording an absence is a finding, and it is frequently more useful than the claim would have been.

The clearest example on this site: no study has assessed whether neck resistance training changes concussion incidence. Not studies that found no effect — studies that have not been run. That absence is stated plainly on the head impact page because the commercial temptation to fill it with implication is exactly why it needs stating.

Known gaps in this library

Three documents that would materially improve it have not been obtained: the US Air Force’s Viper Pilot Neck Health and Conditioning Guide (DTIC ADA571450), NATO STO-TR-HFM-252 Aircrew Neck Pain Prevention and Management, and Alricsson and colleagues’ 2004 study of supervised training in Swedish Air Force fighter pilots (PMID 14736129).

They are listed rather than quietly skipped. A reference work should be explicit about the shape of its own holes.

Corrections

If a study is cited incorrectly here, or a limitation has been understated, or evidence has been missed that changes a conclusion — support@theneckflex.com. Corrections are made publicly and the page records that a change was made.

The full editorial policy, including how historical sources are handled and a documented case where this site published something wrong, is on the editorial standards page.