Next: Error Estimate from
Up: 5: Goodness of
Previous: Typical Confidence Level
- Unfortunately, one often gets
plots that look like
- with no clear ``flat'' piece.
- This maybe because your ``signal'' is very small.
- It could be because you got your errors wrong (!), e.g.,
maybe the true
are all larger than ones used in calculating
.
- In this case, it makes sense to scale
until we
get a nice flat distribution for confidence level C near 1 and
analyze as above.
Geoffrey Fox, Northeast Parallel Architectures Center at Syracuse University, gcf@npac.syr.edu