For each enhancement, suggestion, or criticism that I find useful, I will tip between 20 and 50 RSC (up to 300 total).

Pre-Registration:
The Big Taping Truth Trial

Summary

Title

The Big Taping Truth Trial

Description

Does mouth taping during sleep improve sleep quality, snoring, and facial attractiveness compared to a placebo taping condition in the general population?

Hypotheses

(Sleep Quality) Participants will show improvements in sleep quality when mouth taping, as measured by fitness tracker metrics.
(Snoring) Participants will experience a reduction in snoring intensity, as measured by SnoreLab scores, when mouth taping.
(Attractiveness) There will be no significant difference in participants’ facial structure after a period of periodically using mouth tape during sleep, compared to when not using it.

Study Design

Study Type

Experiment (random assignment of treatments)

Blinding

None

Study Design Description

This is a repeated-measures crossover study, meaning that each participant will experience both of two conditions over multiple nights:

Intervention (“Vertical”): Mouth tape applied vertically to encourage closed lips during sleep.
Placebo (“Mustache”): Tape applied horizontally on the upper lip without sealing the mouth.

The total trial period will be 30 days for each participant, because this balances sufficient data collection with an amount of time that seems achievable to participants.

First, they will have an initial taping Adjustment Period where they do vertical every night to get used to taping. This period will be two weeks because it is a typical length of time to get used to mouth tape (James Nestor, Reddit). Then, each participant will proceed to the Randomized Phase, where they will receive an individualized randomized taping assignment for each of the 16 subsequent nights.

The crossover design controls for individual differences in baseline sleep patterns by allowing each participant to act as their own control.

Randomization

During the 16-day Randomized Phase, participants are assigned a different taping condition for each night. The assignment order is random, and the assignments will be balanced so that they have 8 nights in each condition. Since each participant is their own control, we want sufficient data in each condition.

Data Collection Procedures

Population: The study will recruit healthy adults aged 18+ who regularly track their sleep using an approved fitness tracker (Whoop, Oura, Apple Watch). The target population includes individuals interested in sleep improvement, biohacking, and health optimization, as these groups are more likely to adhere to the study protocol.

Recruitment Methods: Advertisements, partnerships with relevant influencers, online forums, DeSci communities, word-of-mouth

Payment:

Amazon gift card which covers the cost of recommended mouth tape ($7-16) for the first 160 participants
If they participate in tracking snoring, 6 months free access to SnoreLab premium
If they complete the whole study, entered in a raffle to win Apple Watch Series 10
If they complete the whole study, customized results report

Eligibility: Adults (18+) in good health, such that taping your mouth in the way we described would not be dangerous. We recommend participants who cannot breathe through their nose or have easily irritated skin don’t participate. Participants should use their own judgment and stop if they have issues.

Timeline: The study will be ongoing across the next 6 months, or less, or more (see next section for explanation).

Sample Size and Rationale

We aim to recruit as many people as possible, since there are no costs to scaling with all the tools being automated. The study can theoretically remain open indefinitely, but we will analyze the data once either 2x the minimum desired sample size or 6 months have passed, whichever comes first. If the minimum is not met after 6 months, we will keep it open until the minimum is met.

Phase	Condition	Data available for	Estimated % remain
Baseline	Pre-trial (probably no tape for most)	All participants	100
Adjustment Period	Vertical	Early dropouts & full completers	80
Randomized Phase	Vertical vs. Mustache	Only full completers	50

Given the likelihood of dropout, we will use all available data. Participants who do not complete the whole study will be used solely in the baseline (pre-trial) vs adjustment period (vertical taping) comparison. Those who do complete the full study will be used in the baseline vs. adjustment period test, and also in a vertical vs. mustache comparison from the randomized phase.

Minimum desired sample size:

In order to detect a small effect size, assuming conventional power analysis values (Cohen’s d = 0.2, 80% power, significance level 0.05, two-tailed paired t-test), we’d need 198 participants with complete data.
However, in our case using a lower significance level is justifiable because this is an exploratory study where Type II errors (false negatives) are of greater concern than Type I errors (false positives). Since mouth taping is a low-risk intervention, prioritizing sensitivity makes sense. Using a significance level of 0.10, our minimum is 156 participants.

Alternatively, if we take a Bayesian approach, traditional power analysis does not apply because Bayesian inference continuously updates belief based on accumulating data.

But let’s just say our minimum is 160 participants for the sake of this pre-registration and our stopping condition. So: We will analyze the data once either 320 participants have completed or 6 months have passed, whichever comes first. If we don’t have 160 after 6 months, we will keep it open until the minimum is met.

Variables

Manipulated variables:

Vertical vs Mustache

Outcome variables:

Sleep metrics measured by fitness trackers (Whoop, Oura, or Apple Watch)
Snoring metrics measured by SnoreLab
Facial structure changes measured by anonymous survey respondents comparing participants’ before and after photos. They will also view photos of non-participants’ before and after photos as a control group.

Metric Name	Abrv	Formula	Improvement
Primary Sleep Metrics
Sleep Efficiency	SE	Sleep.efficiency (Amount of total sleep time that is spent actually asleep)	Increase
Blood Oxygen Level	BOL		Increase
Respiratory Rate	RR	sleep.respiratory_rate	Decrease
Heart Rate	HR	sleep.average_hr	Decrease
Secondary Sleep Metrics
Total Sleep Time	TST	Duration of the sleep period, excluding awake time	Increase
Percent Light Sleep	N1/2	100 * Sleep.Light / Sleep.Total	Decrease
Percent REM Sleep	REM	100 * Sleep.REM / Sleep.Total	Increase
Percent Deep Sleep	N3	100 * Sleep.Deep / Sleep.Total	Increase
Sleep Latency	SL	Time to fall asleep	Decrease
Heart Rate Variability	HRV	average_hrv	Increase
Snoring Metrics
Time Snoring	TS	Given by SnoreLab	Decrease
Snoring Percentage	SP	Given by SnoreLab	Decrease
Snore Score	SS	Given by SnoreLab	Decrease
Facial Structure
Perceived Change Index - Jawline Structure	JA	Answer to Survey Question: On a scale of 1 (no change) to 7 (very noticeable change), how much change do you perceive in Jawline Structure?	Increase
Perceived Change Index - Cheekbone Structure	CB	Answer to Survey Question: On a scale of 1 (no change) to 7 (very noticeable change), how much change do you perceive in Cheekbone Structure?	Increase
Perceived Change Index - Facial Width	FW	Answer to Survey Question: On a scale of 1 (no change) to 7 (very noticeable change), how much change do you perceive in Facial Width?	Increase
Perceived Change Index - Chin Projection	CP	Answer to Survey Question: On a scale of 1 (no change) to 7 (very noticeable change), how much change do you perceive in Chin Projection?	Increase

Predictors/covariates:

Primary:
- Mouth breather index: If the total >= 2, they are a mouth breather.
  - Tongue position (+1 if floating or bottom)
  - Subjective opinion (+2 if mouth, +1 if unsure)
  - Snoring (+2 if a lot, +1 if sometimes)
Secondary:
- Fitness tracker type
- Age
- Whether they have allergies, and severity
- Sleep position

Analysis Plan

Statistical Models

This study will use both Bayesian and frequentist analysis. Although we prefer Bayesian analysis for its more sensical probability-based interpretations, the frequentist approach is more well-known and hence easier to communicate. Additionally, if both methods agree, the confidence in the results is strengthened.

For each continuous outcome metric, we will:

Determine if there is an effect.
1. (Frequentist) Compute the p-value and confidence interval for a two-tailed paired t-test.
2. (Bayesian) Compute the Bayes Factor and credible interval for a paired Bayesian t-test (using a default prior, since we lack prior data on the effect size).
Determine the size of the effect.
1. (Frequentist) Compute Cohen’s d.
2. (Bayesian) Compute the posterior distribution of the effect size.
Adjust for confounders.
1. The confounder of greatest interest is whether the person is likely a habitual mouth breather, indicated by our Mouth Breather Index. We will conduct stratified analyses and compare effect sizes for likely mouth breathers vs. non-mouth breathers.
2. We will also determine if any other variables are possible confounders.
  1. (Frequentist) Run an ANCOVA model with secondary predictors specified above.
  2. (Bayesian) Run a Bayesian hierarchical regression model with secondary predictors specified above.
Conduct separate analyses for baseline vs. adjustment period. This is especially important if there is high dropout before the randomized period (>30%).

Inference Criteria

We will not specify thresholds; rather we will provide full interpretations of each statistic. We emphasize Bayes Factor and credible interval as they are easier to understand, and we provide p-value and confidence intervals for familiarity.

The p-value measures the extremeness of the data, and will be interpreted as: “If it were true that mouth taping actually has no effect on this outcome, there would be only a p% chance of getting these results.”

The 95% confidence interval [a,b] gives a sense of the uncertainty, and will be interpreted as: “If the study were repeated many times, [a,b] would contain the true effect size in approximately 95% of those repetitions.

The Bayes Factor BF indicates the strength of the evidence, and will be interpreted as: “It is BF times more likely that mouth taping is beneficial rather than that it has no effect.”

The credible interval [c,d] tells us the range of probable effect sizes, and will be interpreted as: “Given our data and prior knowledge, there is a 90% probability that the actual effect of mouth taping on outcome is within [c,d].”

Data Manipulation

Missing data: Participants who do not complete the whole study will be used solely in the baseline (pre-trial) vs adjustment period (vertical taping) comparison. Those who do complete the full study will be used in the baseline vs. adjustment period test, and also in a vertical vs. mustache comparison from the randomized phase.

Data exclusion: Participants must have at least 5 nights in each treatment, as indicated by their daily adherence survey, in order to be included in either baseline vs. adjustment period or vertical vs. mustache tests. We base this cutoff on the research of Lau et al 2022, which suggests that 3-5 days are representative of a week of sleep timing data and 5-10 are representative of a month of sleep timing data.

Exploratory Analysis

We may consider exploring variations on the mouth breather index, since there is not enough evidence to say that our computed index is correctly indicative of being a mouth breather.

Question

Review the pre-registration for The Big Taping Truth Trial