Alerts thresholds and trials factor documents needs more clarity
The userguide is a rather unclear in documenting how trial factors come in for high and low significance events ( https://emfollow.docs.ligo.org/userguide/analysis/index.html#alert-threshold ). The opening sentence on rate thresholds being "before considering the trials factor. " leaves unclear whether trial factors are meant to multiply or divide the rate. For the high significance the "1/month" is meant to be divided by the trials factor, thus implying a 1/(6months) per pipeline threshold. FOr the low significance events, the "trials factors" seems to be multiplying the quoted rate (2/day) to state what is the expected resulting rate out of the union of all pipelines.
Given that algorithmically (and physically, in running our searches) we can only control the per pipeline FAR and thus thresholds, I suggest our point of departure in discussing FAR thresholds is the per pipeline thresholds. We then consistently state that given the N pipelines that run separately, we estimate the global "OR" of alerts to have an rate of N times the per pipeline one.