When we say our clinical study was "double-blind, placebo-controlled, and cross-over," we are not listing buzzwords. Each of those terms describes a specific design choice that exists for one reason: to make sure the results reflect what actually happened, not what anyone hoped or expected would happen.
Most people have heard these phrases before. Fewer people know what they mean in practice, or why they matter. This article walks through the key elements of rigorous clinical trial design, explains each one in plain language, and shows how they applied to the Sonopeace study conducted in partnership with SleepScore Labs.
Randomization: Removing the Researcher's Hand
Randomization means participants are assigned to treatment groups by chance, not by choice. No one picks who gets the real treatment and who gets the placebo. A random process handles it.
Why does this matter? Because without randomization, conscious or unconscious selection bias can creep in. A researcher might, without realizing it, assign healthier participants to the treatment group or sicker ones to the control group. Randomization prevents this by distributing known factors (like age, gender, and baseline sleep quality) and unknown factors (like genetics or stress levels) evenly across groups.
In the Sonopeace study, 45 enrolled participants were randomly assigned to one of two protocols. Protocol A received the placebo device first, then switched to Sonopeace. Protocol B received Sonopeace first, then switched to placebo. The assignment was entirely random, with neither the participants nor the research team influencing who ended up in which group.
Double-Blind: Nobody Knows Who Gets What
In a double-blind study, neither the participants nor the researchers know who is receiving the active treatment and who is receiving the placebo. This is one of the most important safeguards in clinical research.
The reason is straightforward. If participants know they are getting the real treatment, they may feel better simply because they expect to. This is the placebo effect, and it is powerful. On the other side, if researchers know which participants are receiving the treatment, they may unconsciously interpret data more favorably for that group, or interact differently with those participants. Double-blinding eliminates both sources of bias simultaneously.
In our study, Sonopeace partnered with SleepScore Labs, a commercial sleep laboratory that managed all participant communication. No one at SleepScore Labs knew which participant was assigned to which protocol. Participants received two sealed devices by mail, one for each period, with no indication of which contained the active Sonopeace frequencies and which contained the placebo track. The allocation remained hidden until after all data had been collected and the study was complete.
Placebo-Controlled: Comparing Against Something Real
A placebo is not "nothing." A well-designed placebo mimics the experience of the active treatment as closely as possible, without delivering the component being tested. This is what allows researchers to isolate the specific effect of the intervention from the general effect of simply doing something new at bedtime.
Designing a convincing placebo for a sound-based device is genuinely difficult. You cannot use silence, because participants would immediately know they were in the control group. The placebo needs to feel like a plausible treatment without containing the active ingredient.
The Sonopeace study solved this with a carefully constructed audio masking approach. The placebo device played a continuous shruti drone, an Indian drone instrument that provides a rich, ambient soundscape. The active Sonopeace device played the same shruti drone overlaid with Sonopeace's proprietary frequency profile. To prevent participants from hearing the difference, the shruti volume was set high enough to mask the frequency layer. Both devices were calibrated to the same total volume before shipping.
This is what good placebo design looks like: participants in both groups had an identical nightly experience of placing a device under their pillow and falling asleep to sound. The only difference was whether that sound contained the active frequencies.
Cross-Over Design: Each Person as Their Own Control
In a standard parallel-group trial, one group gets the treatment and another group gets the placebo, and you compare between groups. The problem is that people are different. One person's baseline sleep may be fundamentally different from another's, and those individual differences can obscure treatment effects, especially in smaller studies.
A cross-over design addresses this by having every participant experience both conditions. Half the group starts with the treatment, then switches to the placebo. The other half does the reverse. Because each person serves as their own control, you can compare the treatment effect within the same individual, which dramatically reduces noise from person-to-person variability.
The Sonopeace study used a two-period cross-over. After a three-week baseline observation period, participants entered two consecutive two-week treatment periods. Protocol A participants used the placebo device first (Period 1), then crossed over to Sonopeace (Period 2). Protocol B participants used Sonopeace first, then crossed over to placebo. This meant every one of the 35 completers experienced both conditions, and their results could be compared against their own baseline.
Washout Periods and Carryover Effects
One challenge with cross-over designs is the carryover effect: the possibility that the first treatment's impact lingers into the second period. If someone uses Sonopeace in Period 1 and then switches to placebo in Period 2, any improvement still present during Period 2 might be a lasting benefit from Sonopeace rather than an effect of the placebo.
Researchers typically address this with a washout period, a gap between treatment periods where no intervention is given, allowing any residual effects to dissipate before the next phase begins. The appropriate length depends on the intervention: for drugs, it is often three to four times the elimination half-life.
In the Sonopeace study, the transition between periods was immediate rather than separated by a formal washout. The statistical analysis accounted for this by testing explicitly for carryover effects. What the data revealed was actually one of the study's more interesting findings: Sonopeace benefits did carry over. Protocol B participants (Sonopeace first) maintained superior outcomes even after switching to placebo, with PROMIS scores continuing to improve during the placebo period. This persistence pattern was itself informative, suggesting that the benefits of Sonopeace may be durable rather than dependent on continuous nightly use.
IRB Approval: Independent Ethical Oversight
Before any clinical study involving human participants can begin, it must be reviewed and approved by an Institutional Review Board (IRB). An IRB is an independent committee charged with protecting the rights, safety, and welfare of study participants. It evaluates whether the research is ethical, whether the methods are sound, whether risks are minimized, and whether participants have been given enough information to provide truly informed consent.
IRBs are regulated by the Office for Human Research Protections (OHRP) within the U.S. Department of Health and Human Services. Their approval is not a rubber stamp. They review the study protocol, consent forms, recruitment methods, and data handling procedures before granting clearance.
The Sonopeace study was approved by Sterling Institutional Review Board under protocol number #13207. All 45 enrolled participants signed electronic informed consent forms prior to enrollment.
Statistical Significance and P-Values
After a study is complete, the question becomes: are the observed differences real, or could they have occurred by chance? This is where statistical testing comes in.
A p-value answers a specific question: if the treatment had no real effect, how likely would we be to see results at least this extreme? The conventional threshold is p < 0.05, meaning there is less than a 5% probability that the observed result is due to chance alone. The lower the p-value, the stronger the evidence against the "no effect" explanation.
In the Sonopeace study, the primary outcome measure (PROMIS Sleep-Related Impairment) reached p = 0.020 on a two-tailed Welch's t-test. For context, two-tailed testing is more conservative than one-tailed because it accounts for the possibility of effects in either direction. Reaching significance on a two-tailed test with 35 participants reflects a genuinely robust finding.
Effect Size: How Much Does It Actually Matter?
Statistical significance tells you whether an effect probably exists. Effect size tells you how large it is. These are different questions, and both matter.
Cohen's d is the most widely used measure of effect size in clinical research. It expresses the difference between two groups in standard deviation units. The conventional benchmarks, established by statistician Jacob Cohen, are: 0.2 is a small effect, 0.5 is a medium effect, and 0.8 is a large effect. An effect can be statistically significant but trivially small (common in very large studies), or it can be meaningfully large but fail to reach significance (common in small studies). The ideal result is both significant and substantial.
The Sonopeace study reported a Cohen's d of 0.72 for the primary outcome, placing it in the medium-to-large range. This means the difference between Sonopeace and placebo was not just statistically detectable but clinically meaningful. For comparison, many pharmaceutical sleep aids produce effect sizes in the 0.3 to 0.5 range on similar instruments.
Validated Instruments: Measuring What Matters
How you measure outcomes is as important as the design itself. The Sonopeace study used both objective and subjective measurement tools, each chosen because it has been independently validated by the broader sleep research community.
On the objective side, the SleepScore Max is a contactless biomotion sensor that sits on the nightstand and tracks sleep stages, duration, and disruptions using non-contact radio frequency technology. Participants did not need to wear anything to bed, which eliminated the common problem of wearable devices disrupting the sleep they are supposed to measure.
On the subjective side, the study used four standardized clinical questionnaires. PROMIS Sleep-Related Impairment (the primary outcome) is maintained by the National Institutes of Health. The Pittsburgh Sleep Quality Index (PSQI) is the most widely used subjective sleep quality measure in clinical research. The Insomnia Severity Index (ISI) is the standard tool for assessing insomnia severity and treatment response. The Epworth Sleepiness Scale (ESS) measures daytime sleepiness, serving as a safety check to confirm that improved nighttime sleep was not achieved through sedation.
Using multiple validated instruments provides convergent evidence. When different tools measuring different aspects of sleep all point in the same direction, confidence in the findings increases substantially.
Why All of This Matters
Every element described above exists to answer the same fundamental question: did the treatment actually work, or is something else explaining the results?
Randomization prevents selection bias. Double-blinding prevents expectation bias. Placebo control isolates the treatment effect from the general effect of doing something new. Cross-over design controls for individual differences. IRB approval ensures ethical conduct. Statistical testing quantifies the probability of chance findings. Effect sizes quantify practical significance. Validated instruments ensure you are measuring what you think you are measuring.
No single safeguard is sufficient on its own. Together, they form a system of interlocking protections that give us confidence in the results. The Sonopeace study incorporated all of them, producing findings that showed statistically significant, clinically meaningful improvements in sleep quality across multiple measurement approaches.
We built this study the way we did because we believe you deserve to know whether something actually works before you trust it with your sleep. The full study details and results are available on our research page.