ANOVA in StatCrunch (2).mp4 Video Lecture | Statcrunch Tutorials for Beginners - Data & Analytics

the second example that's given in the
notes details a condition that results
in abnormally low platelet platelet
counts that often occurs during NEC
forgive me for skipping over most of
those words but I can almost promise you
that I would have butchered their
pronunciation so we're gonna stick with
the acronyms the goal of this study was
to compare the mean platelet count
across four different categorizations of
NEC status and you can see in the setup
of the problem what those four groups
are in the example given in the notes we
didn't go through and actually check the
assumptions for this data and in this
video we will go through those
assumptions the first thing that we
would like to notice is this data is
structured differently than the data in
the previous example in the example that
dealt with the force to break the femur
we had one column that held our response
in this case our response is the log10
platelet count but now we have four
columns of those values one column for
each of our tour groups that were
interested in this data structure is
probably more familiar to you than the
previous structure was so if we start
with the assumption of normality we
would like to create histograms for
these four groups in this case I don't
have to worry about a group by variable
but has now I can individually select
groups 0 through 3 if you would like
this graph to come or to appear in the
same structure that the previous one did
we can click Next through to the end and
in this case since we're comparing four
groups I would like to have four columns
to this page of graphs in the resulting
output we have four separate histograms
the first for group zero the last for
group three and you'll notice that group
0 and group 1 appear to be somewhat
negatively skewed group 2 and group 3
while not perfectly symmetric do you
appear to be more bell-shaped so as with
the last example we might want to
proceed with caution in terms of the
normality assumption again we can also
calculate summary statistics for this
data to assess the variability and this
is similar to the approach that you took
before now though we want to select all
four columns that contain our data and
in the summary statistics that appear
you'll notice that if we start to
compare standard deviations twice the
smallest standard deviation of 0.3 7 is
much larger than our largest standard
deviation of 0.46 therefore we could
assume that the underlying variability
for these four classifications of any C
status concerning their platelet count
is approximately equal so now we're at
the point where we could go through and
do an analysis of variance of this data
to assess whether there's any difference
in the mean platelet count so if we come
to our ANOVA menu again we're going to
consider a one-way ANOVA because we have
one factor that we're interested in and
that factor is our classification status
the analysis that we choose to do in
this case is going to be slightly
different though because now we're not
going to compare values in a single
column we want to compare our four
columns of data group 0 through group 3
so I'm going to select each one of these
column and also going to select the
Tukey HSD again that will give us
multiple comparisons and then we can
select calculate
again the output that appears is exactly
similar to the output that we saw in the
previous example you'll notice that the
top half gives us our analysis of
variance results again the component
that we are interested in is the F
statistic and p-value if you look at the
bottom of this output you'll notice that
we now have more confidence intervals
now the reason we have more intervals
for this particular example is because
our factor of any C classification had
more levels instead of only comparing
three groups we were comparing four in
this case and if we looked at every
possible pair of groupings that left us
with six possible pairs now on this
example if you look at your confidence
intervals you'll notice that the
majority of them have a negative lower
limit and a positive upper limit so for
all of those groups we can't come to the
decision that those are statistically
different there's only one case where we
have an upper and lower limit with the
same sign and that occurs when we
compared group three to group zero so in
this case we notice that group three
minus group zero results in a negative
lower limit and a negative upper limit
so in that case we're able to conclude
that group zero has a greater mean
platelet count then group three had no
other significant differences were found
now in the practice problems that are
posted for this chapter material you'll
see data that's in both formats where we
have one column containing our response
one column containing our vector and
you'll also see the case like the
platelet count where we have individual
columns for each level of our factor
containing their response
be careful how you analyze that data
within the ANOVA menu you have to select
the appropriate format please let me
know if you have any questions on these
practice problems or on the procedure
analysis of aliens