going to talk about biases in more depth in
a further lecture.
So now let’s switch gears a bit and talk
about the null hypothesis, which is one of
the foundational ideas for doing statistics,
what is the null hypothesis? It is a statement,
it's a statement that there is no relationship
between the variables that we’re trying
to test for. It is important that you understand
that the null hypothesis always says, there
is no relationship. The null allows us to
assess our statistical tests and tells us
whether or not to reject or fail to reject
our null hypothesis. I'm a bit of a stickler
for protocol and philosophy, so I say fail
to reject, some people say accept. For reasons
that I won't get into here, philosophically
we never actually accept a null hypothesis,
we either reject it or we fail to reject it.
Remember, the null hypothesis says, nothing
is going on, there is no relationship between
the things we are measuring. What I want to
do is to be able to reject that hypothesis.
If I reject that hypothesis,
I know something is going on, I know I've found
something interesting in my test. I can write
the null hypothesis mathematically like this,
H0 is notation for the null hypothesis, the
Greek letter µ is notation a mean, an average
in one group. So here I’m saying that the
mean measurements in two separate groups are
the same. Maybe I’m running a randomized
controlled trial, maybe I’m looking at the
mean responses in my treatment group versus
my placebo group. My null hypothesis says,
they're exactly the same. In other words,
I found no effect, nothing is going on, I'm
assuming that the thing I'm measuring has
no effect whatsoever. That's my null hypothesis.
I'm seeking to reject that null hypothesis.
In this example, I’ve shown the comparison
group to be a placebo group.
That’s a specific example.
More generally, calling it a control group would be more accurate.