I attempted to use melatonin on myself in a pseudo-study and wrote about it here: http://lesswrong.com/r/discussion/lw/ngj/open_thread_april_4_april_10_2016/d7s5

Following gwern's post about Melatonin, I did a self-trial of melatonin to see if it improved my sleep.

Objective: To see whether I (myself only) should take melatonin regularly or not. Therefore: Have a clinically significant decrease in sleep time or increase in daytime awakeness. Clinically significant=statistically different & meaningful (i.e. 1 minute difference is not worth the effort, for example)

Note that at baseline, I never have trouble falling asleep, and I never have trouble staying asleep. Melatonin is theoretically advantageous in inducing sleep, which is not an issue for me.

Method: I calculated an n required for a significant difference based on pvalue<0.10 but I don't remember the details. I think my required n was about 150-250 with half in each arm

Every night, before bed, I would arbitrarily unblindedly decide to take 1 mg melatonin or take nothing (SWANSON MELATONIN 1 MG 120 CAP from LH-Nutrition on amazon. This was unblinded and arbitrary to better reflect the "intention to treat" if I were to continue to take melatonin in the future. That is, if this study on myself indicated that I should take melatonin, then I want the study to reflect my reality instead of a generalized "lab" experiment to generalize to others. In the same vein, there was no controlling for biases or other factors as if things such as the placebo effect improved my sleep, then that was fine by me

In theory, I was to take melatonin vs no melatonin about 50%-50%, although in practice this was not the case

Results: I recorded sleep data from 2015/09/22 to 2016/01/26 before the study stopped due to apathy on my part and lack of perceived effect. On 3 nights I did not record data (because my sleeping was disturbed e.g. working a night shift). n=124 (n=41 melatonin, n=83)

I did not do any statistical tests because on graphing the results, on gross examination there was clearly no significant change in my daytime awakeness or number of hours slept (see graphs). On average, without melatonin I slept 7.68 hours (vs 7.67 with melatonin). The STD with melatonin was lower. The median sleep time was 7.68 without vs 7.60 with melatonin. When I slept <=6 hours, I tended not to have taken melatonin. This is likely an effect of the intention-to-treat (i.e. if I am sleeping very late so that I will lack sleep, I am too tired/lazy to take melatonin)

My wake-up energy was grossly unchanged regardless of if I took melatonin. My perceived awakeness throughout the day was lower on the days I took melatonin compared to the days I did not, (grossly looking at the graph, not formally evaluated).

I noticed no difference taking melatonin or not. No side-effects/perceived benefits.

Conclusion: Study was underpowered as study was abandoned with a low n. No blinding, no controlling for biases, "intention-to-treat" in the loosest meaning. No side effects or subjective difference noted on/off melatonin. No reduction in sleep time and no improved awakeness during the day on melatonin. Melatonin 1 mg did not have any significant effect on me (?doesn't work on me; ?wrong dose; ?bad drug manufacturer - I went with the cheapest amazon offer). Conclusion: I will cease taking melatonin as the effect is not significantly advantageous in myself.

Raw data and graphs below: http://s000.tinyupload.com/index.php?file_id=05048577364961130526