Disclaimers and Disclosures

Medical / Clinical Disclaimer:

The content on this site is not intended to diagnose or treat any medical or mental health condition and should not be constituted as medical advice.

By participating in any of the programs or training recommendations on this site, you agree that you have previously and concurrently discussed these practices and recommendations with a licensed and credentialed medical provider.

Activities, practices, or interventions you may find therapeutic are not a replacement for clinical mental health counseling.

Data Availability:

Annotated Data is available on: Google Drive

Raw Data is available on: GitLab

Affiliates:

I receive a small sales commission from the following companies:

Research Limitations:

Double Blind Impossibility: It is impossible to do a truly double blind study on on yourself (N=1) because you’re both the observer, the interventionist, and the observed. Thus, there will always be confirmation bias. However, I have a vested interest in finding correct answers and useful interventions because I want my health and fitness to legitimately improve as well.

Human Error: Sometimes data gets entered incorrectly, or I forget to check or measure one thing on a particular day and that data is missing. That’s why several data sets may exist for the same “study” and get parsed for the published result. Additionally, anomalies that happen on a less-than-monthly basis (my standard interval of observation), such as a true off-the-wagon cheat meal while a friend’s visiting from out of town, or a cold or flue, may greatly skew the data. Such days are often omitted from the data set.

Variance: For the non-statistically inclined, a certain amount of randomness or variance has to be introduced into a study in order to account fro “chance” or random interactions. For example, some reactions are dose dependent (i.e. effect size), so you need a wide range of intervention dosage and sufficient days of application for each does to avoid confounding variables (co-variance). By contrast, changing too many things at once is prone to false positives; you ascribe erroneous relationships between variables when a co-variant may be responsible. Thus, you need a more robust (larger) data set, a longer observation period, or a follow up / repeatable intervention as well.

Statistics Briefing:

Pearson’s R (r) is used to describe a “correlation”, or “relationship” of two variables on a scale of +1 to -1; where 0 means “no relationship.” Correlation is not causation, and it is impossible to identify intersectionality in the relationship with (r) alone. The closer R gets to +1 / -1 the stronger the relationship is. Positive (+) means that as one variable increases or decreases, so does the other variable. A negative (-) relationship means that as one variable increases or decreases, the other variable moves the opposite direction (ref.).

+/-1 = “perfect” one-for-one relationship.
+/- 0.8 = “very strong” relationship.
+/- 0.6 = “moderate” relationship.
+/- 0.3 = “fair” relationship.
+/- 0.2 (or smaller) = “weak” relationship.

Statistical Significance (p) indicates the confidence or probability that a given relationship (r) has occurred by random:

p = 0.05: 95% confidence, the standard for “statistical significance.”
p = 0.01: 99% confidence that results did not occur at random.
p = 0.001: 99.9% confidence that results did not occur at random.
p = 0.0001: 99.99% confidence that results did not occur at random.

ANCOVA is a test of co-variance to indicate whether or not two variables operate independent of each other on a given result.

Effect Size measures the magnitude of effect. This tells us the “strength of a relationship.” Sometimes this is expressed in terms of (r), but also Cohen’s (d). This is because two things -- like ice cream sales and shark bites -- can be strongly correlated (i.e. both happen during warmer weather), but have no practical relevance to each other (ref.).

d = 0.2: small effect size
d = 0.5: medium effect size
d = 0.8: large effect size
d = 1.3: very large effect size

Odds Ratio is sometimes used to express relative risk -- also expressed as (RR) -- between two interventions; typically an experimental and a standard one. It’s written in terms of percentages, for example (ref):

OR 1.5 (small) = 50% increased odds above normal or control group conditions.
OR 2 (medium) = experimental intervention is 2x as likely as control to produce a result.
OR 4 (large) = experimental intervention is 4x as likely to produce a given result.