Confounders, baselines, and product identity

1. Data ≠ proof

“It worked for me” can be a labeled observation. It is not proof of a general effect. Logs without confounders become ads. This lesson teaches what to name so a personal story stays data.

2. Baseline

A baseline is what was measured or noted before the change you want to interpret. Without a baseline window, “I felt better” has no contrast. Duration bands from Lesson 4 still apply: days / weeks / months — not a calendar plan.

3. Confounder catalog

Major confounders in personal peptide discourse include:

• Concurrent changes — training, sleep, food, other medicines, illness course, attention.
• Expectation / open-label optimism — believing a change was made can shape what gets noticed.
• Regression to the mean — extreme days often move toward a person’s average even with no real intervention effect.
• Selective memory and selective logging — recording only “good” days.
• Measurement change — looking harder and finding more.

These can mimic an effect. Naming them does not make a story worthless. It keeps the story a report.

4. Regression to the mean

When a variable is extreme on first measurement, the next measurement tends to be closer to that person’s average even with no real intervention effect. Barnett et al. (2005) explain the phenomenon in epidemiology (PMID 15333621). Morton and Torgerson (2003) discuss how untreated extremes can be misread as treatment success in health-care decisions (PMID 12750214).

Lesson use: a bad week that improves after you started watching it is not automatically proof of a compound effect.

5. Expectation, placebo, and nocebo — without invented percentages

A Cochrane review of placebo interventions across clinical conditions (Hróbjartsson and Gøtzsche, 2010) found little clinical effect on binary outcomes and modest average effects on continuous patient-reported outcomes in the trials examined (PMID 20091554). That is not a universal partition of every personal anecdote, and this lesson invents no “most of the effect is placebo” percentage.

Open-label placebo research shows expectation and ritual can still shape reported outcomes even when people know they received placebo (Kaptchuk et al., 2010; PMID 21203519). That is literacy about uncontrolled optimism — not a claim that “peptides are placebos.”

Nocebo names nonspecific adverse effects shaped by expectation and labeling (Barsky et al., 2002; PMID 11829702). Logging “side effects” without naming expectation as a possible confounder can inflate a causal story.

6. Formal n-of-1 vs casual self-tracking

Formal n-of-1 methods use planned periods, predefined outcomes, and often blinding or crossover to reduce confounds (Guyatt et al., 1986, PMID 2936958; CENT 2015, PMID 25976398). Casual self-tracking can generate rich personal series but typically lacks those controls (Swan, 2013; PMID 27442063). Course 0 teaches labeling. It does not teach you to run a formal n-of-1 drug trial.

7. Confirmation bias in logs

Confirmation bias is seeking or interpreting evidence in ways partial to existing beliefs (Nickerson, 1998; DOI 10.1037/1089-2680.2.2.175; no PMID). In personal logs, analogs include recording only good days, reframing misses, or reading a paper’s molecule name onto a vial. Required uncertainty and confounder fields are the site’s response.

8. Product identity checklist (questions only)

• 1. Is the material in the paper the same object as the product being discussed?
• 2. Is this an FDA-approved finished drug product, a compounded preparation, a similarly named analog, or a research-use-only–labeled research material?
• 3. What would disconfirm the identity claim?
• 4. Am I letting a name match do the work that product identity still has to do?

No sourcing. No certificates of analysis theater that names shops. Status questions from Lesson 2 stay in Status.

9. Worked log

Stub: “Felt better over three weeks. Also changed training and sleep.”

Weak version (reject as proof): “It worked. The compound fixed it.”

Stronger labeled version: “Duration band: weeks. Baseline: low energy weeks before. Noticed: more steady mornings some days. Confounders: training block changed; sleep schedule changed. Uncertainty: I cannot separate those. Identity: compound as labeled on the box; I do not claim it matches a paper’s material. Source: firsthand. Data, not proof.”

10. Tie to Experience drawer

On live pages, Reported experiences and member discussion hold labeled reports after moderation. Readers and moderators look for this grammar. An empty Experience drawer is still not a safety rating.

11. Close

Close: Lesson 6 assembles a structured report and a clinician question list, and states when personal logging is the wrong primary tool. This lesson has one job: keep “it worked for me” as data.

CTA after the lesson: Open the library · Continue Course 0 · Guidelines