Thirteen rugby players, thirteen sedentary adults, one cross-sectional snapshot — and a stack of causal verbs. Reading the methods, slowly.
The Paper · Machado et al. (2024)
Lower limb balance, ankle dorsiflexion, orofacial tissue pressure, and occlusal force of rugby players
Sports Medicine and Health Science, 6: 173–178
01
Claims & Bias
Where the language of the paper quietly outpaces what the data can say.
1
Causal verbs, correlational design.
The authors write that rugby "reduces," "causes," "leads to," and "promotes" physical changes — verbs that a cross-sectional study cannot support. Self-selection, a 6-year age gap, and body-composition differences offer simpler explanations. The language claims what the design cannot prove.
In plain termsComparing two groups at a single point in time can show they differ, but not that one activity caused the difference — the study's wording treats correlation as cause.
“The language claims what the design cannot prove.”
Point 01 · Claims & Bias
2
"Clinically observed" smuggles in a null result.
When the p-value failed, the authors declared the effect "clinically observed" instead. A genuine clinical-significance argument requires pre-specified effect sizes and thresholds — not a quiet swap of vocabulary after the statistics disappoint.
In plain termsIf a result isn't statistically meaningful, you can't relabel it as "clinically meaningful" after the fact and carry on as though it worked.
3
Significant findings get the spotlight; null findings get the basement.
Near-significant results are discussed at length and framed causally. Null findings, including bite force, barely register. The narrative reflects the authors' hypothesis, not the full dataset.
In plain termsResults that support the authors' idea get paragraphs of discussion; results that don't get a sentence or none at all.
4
Mechanisms invented in the discussion.
The discussion invokes sleep apnea, serotonin pathways, and compensatory muscle mechanisms — none of which were measured. These untested ideas carry the confidence of findings, blurring the line between evidence and speculation.
In plain termsThe paper explains its results by naming biological processes it never actually measured, and presents those guesses with the weight of data.
5
A four-link causal chain with zero links measured.
The authors propose: rugby → muscle imbalance → shoulder changes → orofacial changes. They measure only the first and last step, skipping every intermediate link. The premise connecting ankle mobility to tongue pressure is claimed, not demonstrated.
In plain termsThe story has four steps connecting rugby to changes in the mouth, but only the first and last were actually checked — the middle is assumed.
“Ankle mobility to tongue pressure is claimed, not demonstrated.”
Point 05 · Claims & Bias
02
Statistics & Methodology
Where the numbers underneath the narrative start to give way.
1
Underpowered, then rescued with invalid math.
With n=13 per group, most comparisons could not detect real effects. The authors ran post hoc power analysis — a method that simply restates the p-value — on the one variable that gave a flattering result. The less convenient outcomes went unexamined.
In plain termsWith only 13 people per group, the study is too small to reliably find real differences, and the statistical rescue the authors use doesn't actually fix that.
2
~30 tests, zero corrections.
The paper runs roughly 30–40 comparisons without Bonferroni, FDR, or any other correction. At α=0.05, chance alone predicts ~1.5 false positives. Most "significant" findings sit near p=0.05 and dissolve under any adjustment.
In plain termsRun dozens of statistical tests and a couple will look meaningful by pure chance — the paper doesn't account for that, and its positive results sit right at the borderline.
“Most significant findings dissolve under any adjustment.”
Point 02 · Statistics & Methodology
3
The data statement contradicts the paper.
The authors declare "no datasets were generated or analyzed," then report data from 26 participants on four outcome measures. Nine authors, peer reviewers, and the editor all missed the contradiction. The statement blocks replication and signals absent quality control.
In plain termsThe paper formally states that no data were collected, while its own tables show measurements from 26 people — a contradiction no one in the review chain caught.
4
Snapshot, not story.
A one-time comparison of two groups cannot show that rugby caused anything. The observed differences could equally reflect who chooses rugby, body composition, training history, or unmeasured confounders. The causal framing outruns the design.
In plain termsA single measurement of two different groups can describe how they differ today, but not what made them that way.
5
Two covariates cannot bridge a lifestyle gap.
The authors adjust for age and weight but leave BMI, training history, activity level, injury history, and dental status uncontrolled. Athletes and sedentary adults differ on every dimension that matters. Two variables cannot close the gap.
In plain termsAdjusting only for age and weight leaves out most of the ways athletes and non-athletes actually differ, so the comparison can't isolate rugby's effect.