Alright. So, why are statistical significance tests useful? So,
they do provide a formalized framework for comparing and
evaluating data. And they do enable us to evaluate
whether perceived effects in our data set reflect differences across
the whole population. They do not make a bad
result look good. Significant sets are useful because they provide
a formalized framework for comparing and evaluating data. Different
tests have different assumptions and rules that they incorporate, and
using a particular test ensures that everyone is on the
same page in so far as what we're assuming about
our data. Significance tests also enable us to evaluate whether
perceived effects in our data set reflect differences across the whole
population. As was the case with our company, where ten
out of ten people polled preferred the color blue. Sometimes an
affect that we seen in a small sample does not
reflect what might be true across the entire population. A statistical
significance test let's us formally determine whether or not this might
be the case. Unfortunately, a bad result is not going to
look any better or worse as a result of using a
statistical significance test. If our data's
bad, or there's really no difference
between our two samples. We're not going to be able to
undo that with a test. It is possible though that different tests
might give us different results. The really important thing and we'll
go into this a bit more is that you need to use
the right test in the right situations. Why don't we talk a
little bit about how we might actually run a statistical significance test.