Tip:
Highlight text to annotate it
X
So, when you are choosing a statistical test you need to figure out whether your independent
variable is "T" for two, "F" for few or whether it's "R" for relationship for example. You
also have to think about your dependent variable because the nature of your dependent variable
also maters. An example: If you're running an ANOVA or an "F" test you know
that part of what's being assessed is deviations or differences and that the distances between
different numbers are equal, they're meaningful, right? And the same is also true for a "T"
test. Its differences between means and its on a number line where the distances are meaningful
and equal between the scores. That only works if your dependent variable
is measured as a scale variable. It has to either be interval or ratio. So I'm just going
to ask you, think about it. What if you have apples, oranges, peaches and pears as your
dependent variable. How are you going to put those on a number line, compute a mean and
compute your deviations and figure out mean squares? You're not, because you can't do
an "F" test or an ANOVA on nominal data. It has to be at least interval or ratio. It has
to have those equal intervals in between items. So, for the "DV" it has to be a scale variable
if you want to do an "F" test, or a "T" test or compute an "R" because equal distances
between items matter in these computations. In contrast there may be studies where your
dependent variable is nominal, apples, oranges, peaches, pears. That's fine, the only thing
is you can't run these kinds of tests. You would need to run, for example, you could
run a couple kinds of chi square tests, which we will cover later. So don't worry right
now about what chi squared is but just know your "DV" has to be a scale variable if you
are going to run "F", "T" or "R" on it.