Sutk Nakasa wrote:
> How many replication is required
There is a simple answer: You have to repeat an experiment until you are
convinced that further repetition will not yield a different result. So
the actual number of course depends on things like signal/noise ratio,
but as a general rule, 3 repetitions are probably the minimum one can
get by with even under ideal circumstances.
There may be circumstances where that is not possible (valuable clinical
samples for example), but then the results must be clearly marked as
preliminary.
Most journals require statistical tests of validity, but in my
experience those confirm only what an experienced operator can see on
first glance. If results "look" decent, they'll pass statistical tests
with high significance.
