Analysis of Ariel & Saville (1972): "Anabolic Steroids: The Physiological Effects of Placebos"
What the Paper Claims
The study's central claim is that strength gains observed in athletes taking what they believed to be an anabolic steroid (Dianabol) were actually driven by the placebo effect, not by the drug itself — because the pills contained no active ingredient.
Evaluating the Claim: What's Valid
The placebo effect on performance is real. The motivational and psychological component of believing you're on a performance-enhancing drug is well-documented and plausible. Athletes who believe they are taking steroids may train harder, push through more discomfort, and experience genuine neuromuscular benefits from intensified effort. The study's regression line comparison is a reasonable approach to detect rate-of-improvement changes.
The design is internally consistent. The key manipulation — telling athletes some would receive Dianabol, then giving all 6 remaining subjects placebo — was clever. Subjects were primed with positive information from the University Health Service and access to favorable research. Strength gains were indeed statistically significantly greater in the placebo period (P) vs. the pre-placebo training period (PP) for bench press, military press, squat, and total combined (all at p<.05 or p<.01 per Table 1 and Table 2).
Serious Methodological Limitations
| Problem | Detail |
|---|
| Tiny sample size | Only 6 subjects completed the study. This is far too small to draw broad conclusions. Statistical power is extremely low. |
| No true control group | There is no group that received actual Dianabol vs. placebo. The study only compares training-period progression vs. placebo-period progression within the same subjects. |
| Confounded training timeline | The placebo period (4 weeks) followed 7 weeks of pre-placebo training. Strength gains are non-linear — athletes often improve faster after establishing a base, so the steeper gains in the P period may simply reflect normal training progression rather than a placebo effect. |
| Unequal period lengths | PP = 7 weeks; P = 4 weeks. Comparing regression slopes across periods of different duration introduces bias. |
| No blinding verification | There is no confirmation that subjects actually believed they were taking steroids. The design assumes the manipulation worked without measuring belief. |
| Selection bias | 8 subjects were selected from 15 as "most improved," then 2 were excluded after screening — potentially removing outliers in ways that shaped results. |
| No pharmacological control | The paper does not address whether 4 weeks of Dianabol at 10 mg/day would even be expected to produce measurable strength gains in already-trained athletes. This is the actual comparator that's missing. |
What the Broader Evidence Says
According to Harrison's Principles (p. 11013) and sexual health guidelines (p. 23), androgenic-anabolic steroids do produce real increases in muscle mass and strength in the context of resistance training — this is well-established in the literature. Meta-analyses in older/hypogonadal men show modest effects; in healthy trained athletes, supraphysiological doses have consistently demonstrated genuine strength gains beyond placebo in properly controlled trials conducted after 1972.
The Ariel & Saville study was published before the modern consensus on AAS efficacy was established, and its design cannot actually test whether steroids work — it only tested whether belief in taking steroids affects performance.
Verdict
Partially true, but overstated. The finding that expectation/belief can enhance strength performance is credible and consistent with placebo literature. The psychological enhancement of knowing you're on a "performance drug" is real.
However, the title implies that anabolic steroid effects are merely placebo effects, which is false based on decades of subsequent controlled evidence. The study does not contain a steroid arm at all — it cannot prove steroids have no real effect. It only shows that psychological expectation alone can drive gains beyond normal training progression, which is a narrower and more defensible claim.
The study is historically interesting but methodologically too weak (n=6, no steroid comparator, confounded timeline, no blinding check) to support the broad conclusion implied by its title.