Well, as they say: The issue with statistics is that you can never be sure :)
As I am not aware of any existing software to analyze causalities in such an irregular log, I started writing my own this year.
It is an interesting challenge. The approach I currently take is a visual one: I draw a chart on which every notch on the x-axis is a 24-hour interval. And two lines on it. One is the line where I did not do something (say take Vitamin D3) and the other line is where I did do it.
The y-axis is the probability of an effect (say having a headache) in that time interval.
The center of the graph is the 24-hour interval after the treatment. Then the 24-48 hour interval, then the 48-72 hour interval etc. And left to the center is the 24 hour interval before the treatment. Then the 24-48 hour interval etc.
So it would be a "perfect" graph if the two lines look very much the same before the treatment and very much different after the treatment.
Some treatment/effect pairs look promising, but so far, I don't have a result clear enough that I feel confident enough to publish it.
As I am not aware of any existing software to analyze causalities in such an irregular log, I started writing my own this year.
It is an interesting challenge. The approach I currently take is a visual one: I draw a chart on which every notch on the x-axis is a 24-hour interval. And two lines on it. One is the line where I did not do something (say take Vitamin D3) and the other line is where I did do it.
The y-axis is the probability of an effect (say having a headache) in that time interval.
The center of the graph is the 24-hour interval after the treatment. Then the 24-48 hour interval, then the 48-72 hour interval etc. And left to the center is the 24 hour interval before the treatment. Then the 24-48 hour interval etc.
So it would be a "perfect" graph if the two lines look very much the same before the treatment and very much different after the treatment.
Some treatment/effect pairs look promising, but so far, I don't have a result clear enough that I feel confident enough to publish it.