I had written a reply contradicting the opening post and deleted it. Here is what survived. Sample size in n-of-1: you are the sample. Repeated measurements (weekly weighings, daily mood scores) increase the power to detect a real effect even though n=1. The disagreement above is smaller than it looks once the terms…
On the subject in the title: An ABAB design with a data table and honest limitations — a second dataset Working notes rather than a conclusion.
Posting a small dataset on ABAB design. It is mine, it is uncontrolled, and the method is stated so it can be discounted appropriately.
What I would like is not agreement but a second dataset collected by someone with no stake in mine. If one exists I would rather read it than argue for this one.
I had written a reply contradicting the opening post and deleted it. Here is what survived.
Sample size in n-of-1: you are the sample. Repeated measurements (weekly weighings, daily mood scores) increase the power to detect a real effect even though n=1.
The disagreement above is smaller than it looks once the terms are fixed.
On ABAB design, I would rather understate and be corrected upward than overstate and be quoted. That is a house style here and it is a good one.
Post #2 is right about the mechanism and I think understates the practical bit.
Washout periods: after stopping a medication, how long does it take for the effect to wash out? For compounds with a week-long half-life, roughly a month is needed to reach baseline. Using that washout period in a before-after design strengthens the inference.
I am confident about the direction and much less about the magnitude.
One caution on ABAB design: everything above assumes the underlying documentation is what it claims to be. That assumption is doing real work and is rarely stated.
I have been on both sides of the ABAB design argument in this category within eighteen months, which should tell you how strong the evidence for either side is.
Worth separating two things that post #5 runs together.
It does not tell you about anybody else, which is why aggregating these accounts does not produce evidence of the kind people want it to.
The number is defensible. The precision I gave it is not.
Adding a small correction to the ABAB design summary above rather than a disagreement with it. The substance holds; one of the figures is out by a factor that matters.
Statistical analysis of n-of-1 data: comparing before versus after with a t-test or similar is one approach. Plotting the data visually is another. Both are valid.
Answering the question post #7 raises rather than the one it answers.
Reading this ABAB design thread as someone who came in with a fixed view: the third and seventh replies moved me and the confident ones did not.
Measure the same thing the same way at the same time of day. Most of the noise in personal data is measurement protocol rather than biology.
Reading it again, the caveat matters more than the finding.
Blinding yourself is harder than it sounds and is not impossible. Somebody else preparing labelled containers is the usual approach and it requires a second person you trust.
I have left out the parts I could not verify.
Reporting the whole series rather than the interesting segment is the discipline that makes personal data worth reading. Selective reporting is the default without effort.
Where I part company with post #15, and it is a narrow parting.
The version of ABAB design that I was taught turned out to be a teaching simplification. Useful, and not true in the way I had assumed it was.
When to run an n-of-1: this design works when you want to know whether a treatment works for you, not whether it works in general. For that purpose, it is efficient.
Not disagreeing with anyone above, just adding the bit I keep having to look up.
Post #20 is right about the mechanism and I think understates the practical bit.
Withdrawal and reintroduction is the strongest design available to an individual, and it only works for effects that reverse on a timescale you can observe.
Flagging that the sources on this are thinner than the confidence in the thread suggests.
Expectation effects in an unblinded self-experiment are large and are not a character flaw. Knowing what you expect to find is a reason to design against it.
Genuinely open to being wrong about this one.
Since ABAB design keeps coming up, it should probably be a maintained page rather than a recurring thread. I am happy to draft it if someone with more direct experience will review it.
On post #22 — agreed on the reasoning, with one qualification.
The most useful reply I ever got about ABAB design was a request to state my units. It sounds like pedantry and it has saved me twice.
Designing a personal experiment that could actually change your mind: that is the standard for an n-of-1 design. An experiment designed so that any result confirms what you already believed has not changed anything.
Adding the measurement that post #24 says would settle it.
Reporting the whole series rather than the interesting segment is the discipline that makes personal data worth reading. Selective reporting is the default without effort.
Filing this under things that are true until someone shows me otherwise.
Worth stating the null on ABAB design before we explain it: the observation may be nothing. That possibility deserves a sentence and usually does not get one.
My position on ABAB design is current rather than settled. I have revised it once already and I expect to again, so treat it accordingly.