Tip:
Highlight text to annotate it
X
so in this video we want to go over how to conduct
an independent sample t test
an independent sample t test is gonna take
one score that we've got one variable and see if it significantly different
by a dichotomous nominal variable
dichotomous means that we've got only to we'll see if it's got three or more
you're gonna wanna run an Anova instead
so in this case for a compare math test scores by gender
so
before we do that we want to check our assumptions of the independent sample
t test
uh... the first assumption is that the math test scores are normally
distributed
we're going to analyze nonparametric tests
one sample k-s test
and we want to check to see if our math test scores are normally distributed
okay and here the results of this one sample k-s test
the results are not significant
which means that
the scores are not
or are normally distributed if the results were significant
we would have
the not normally distributed score
so that assumption is out of the way next will actually go to the second assumption
which is the equality of variance
what that means is essentially we want
the variability
of our math test scores to be the same
for both of our categorical variables for a nominal variable there
so to do that
we'll actually just run the independent sample t test
so here we'll put our math test scores there
and then this case we're going to look by gender
and we've got
values of one for male and two for female
and we'll just go ahead and hit okay
here are the results
now right here is the the Levine's test and this is going to tell us that if
our uuh..
math test scores are
equal in variance by gender
so again like the k-s test we want this to be non-significant to meet the
assumption
if it was significant we can still run our t test we just have to interpret
something else
so since we have non-significance here we can actually interpret the t test like
normal
so here the results of the t test we've got a t of negative three-point seven
five two
and a p value less than point zero zero one so in this case are math test scores
are significantly different by gender
and if you look at the means
the males
have lower test scores than female so we can conclude
at the t test results were significant and males had significantly
lower scores than females
now if the Levine's test was significant
you'd instead interpret this bottom
row here
where equality of variance not assumed and this would be called the welsh's
estimate for the t test
it's interpreted just the same way just calculates the statistics slightly
different
in order to compensate for the non equality of variance