I have some trouble with Error in VPHOT.
I have read and reread Caroline Hurtless institute’s excellent document ‘AAVSO guide to photometric uncertainty’.
When several Comps are possible, we have n values of magnitudes calculated for the variable.
The document explains that we use the standard deviation (=sigma) of the n values to calculate the error** (= sigma).
What we are calculating is indeed an average of these values, and we want to estimate the error associated with this average.
However, the higher n is, the greater the precision, thus reducing the dispersion of the mean estimate due to random factors and, consequently, the error. So why don’t we use the standard deviation of the mean =sigma/sqrt(n)?
And since we are talking about error, why don’t we use
+/- 1.96 sigma/sqrt(n)
as the error, which gives the range of variation of the estimated mean in 95% of cases?
Thanks a lot!
** I consider the SNR to be high for simplicity. Therefore, 1/SNR->0