Tip:
Highlight text to annotate it
X
in this video goes over how to conduct a one-way anova
which is just to see if there is
differences
and continuous variable
by
nominal grouping variable
typically for a one-way anova you're gonna want that grouping variable to have three or
more groups
it only has two groups you can just run and independent sample t test instead
so before we run our Anova we're gonna wanna make sure
our assumptions are met
so we go to analyze nonparametric tests
and on the one sample K-S test
will look at math test scores make sure they're normally distributed
and the results of the scores are
not significant
non-significance means that the are normally distributed
if you did have significance here
you're gonna wanna run a nonparametric test instead of the end of the Anova
so next you wanna run
equality of variance
and that's actually
going to be run by running the Anova
so we go to analyze general linear model
and univariate
put our dependent variable there
and then this field be our
fixed factor which is our independent grouping variable
now we want to make sure we check for equality of variance before we actually interpret the
Anova
we go to options
and we click homogeneity tests and this will run the levine's test for us
now also want descriptive statistics just to see
you know how the different
ethnicity groups have
scored for
our dependent variable
and if the Anova was significant we're gonna wanna know where those differences
are
the Anova itself is just gonna tell us that the groups are different it's not
gonna tell us
which groups are different from each other
we'll put Ethnicity over here for display means for
and click compare main effects
and typically you're gonna wanna adjust this to a Bonferonni
and what that does is every time we do a pairwise comparison there's always that
small chance of making a type one error which means a false positive result
the
changing this to a bonferonni correction means that
it's a little bit harder to find significance here
but we have less of a chance of making that type one error
so we'll go ahead and hit
continue
and then from there we can go hand hit okay
now first we wanna look at our levine's test
from the significance over here here we see that that
the p value is not significant
and like the
K-S test we want nonsignificance
if it was significant
uh... again you wanna run
a nonparametric test instead
since we do have non-significance we can
go over to the f test
and that the P value for Ethnicity was significant
so that means that the each of the ethnicity groups were significantly
different from each other
now it doesn't tells which groups are significantly different from the other
groups but tells us that there are differences there
to find out the differences will want to go down here to the pairwise comparisons
and this will compare each of the groups
so from this we can see that white
versus black
there was no difference
p values point zero eight two
if we compare white versus hispanic
p value is significant
and the mean difference is negative so that means white
had less than hispanic
we look at black versus hispanic again we do have significance here
without a star there
at the significance of less than point zero zero one
and if we look at the Hispanic group
like the others that we found the above
white is less than it black is also less than it
and if we go back up to descriptive statistics
that's what we find
so from the results we can say that
there was significant differences found in ethnicities and through pairwise comparisons
the Hispanic group
had significantly higher means than the white and black