User blog comment:Homuhomu123/Estimation of Radar Modifiers on Artillery Spotting Chance/@comment-25534439-20141213144852/@comment-25779449-20141213222952

That's really helpful man. According to the formula, it sort of scared me to know that 256 trials would only gaurantee an error of 5%, which really doesn't pay off the work... I just hoped the errors ain't that far. But as I noticed when the true probability gets closer to 100%, results become much more accurate even for a smaller sample size. If it's acceptable, I'll try ships of which estimated CI/DA rates are far from 50%. And as for confidence level, is 68% good enough or you'd prefer 90%? I'm glad to hear your opinion