SPSS for newbies: Normal Q-Q plot Video Lecture | SPSS: For Beginners - Data & Analytics

hey hi there so you want to do a QQ plot
that's pretty easy to do I wanna show
you how okay now what you do is you go
to analyze then you go to descriptive
stats for some reason they put the QQ
plot under graphs as you'd expect but
they put under analyze the script of
stats QQ plots there's a logic for the I
guess because it's one of the first
things you might do part your
descriptive stats so click on that box
appears and now all you have to do very
easy is you take over one or more
variables into this variable box
depending on whether you want a QQ plot
for it so I've got this variable called
sim how that only for now and then
that's pretty much it the whole load of
options but really you know if you're
running a QQ plot chances are you just
want to kind of see if your data
normally distributed so just note that
by default test distribution normal just
leave that of interest if we look at
underneath you can actually if you're
interested test to see if your data
follows any other distributions
chi-square exponential T distribution
etc but usually you want normal normal
okay
then leave everything else and we then
just click OK all right take some time
to chug along to get the results but
finally we've got something a model
description ignore that there's nothing
there to see we basically go straight
down to the graph here you go
normal QQ plot of my variable called sin
and you can see what it looks like ok
the idea is that if your variable is
normally distributed these dots which
each of these dot stands for well that
stands for corresponds to a value in
your observations in other words a data
point so if you add up all these circles
which stand to the data points you'll
come to total number of observations you
have for that variable what we're not
interested in we're not interested in
what the x axis and y axis are not only
interested in the theory of how you
compute the QQ plot which you know 9 out
10 of us and not we just want to see if
our data assembles a normal distribution
and so it seems to lie on the line
doesn't it most of the times if at the
ends there's a slight departure there's
a slight departure we wouldn't expect
even for a normal distribution we would
not expect every single dot to lie on
that line just as with a histogram we
wouldn't expect our curve to be
perfectly perfectly symmetrical and
that's because what we have is sample
data and this variation from sample to
sample
but more importantly that there's
natural variation in the sample ok well
that's the main plot but you'll see
underneath it there's another plot
pretty much this is not part of the QQ
plot it's called a d-- trended what's
giving you is d trended normal QQ plot
just it you can just ignore that all you
need to do if you're going to present a
QQ plot this is it ok for those of you
that are interested all this all these
dots are doing is it's looking at the
deviation of each of these dots from
that line so you can see that's why most
of them are centered around zero except
5 ends there which are far a few of
these dots of 5 and 0 because as we see
from graph here it ends there the depart
so you know there's a difference between
the observations and the line the actual
values obtained from here deviation from
the normal is the value on the line of a
line minus the value of the observation
so this second
here is that value on the line - this
value which is higher so you're going to
get a negative value and that's why it
translates down to negative value is the
second point my back just slightly
higher than - not 0.6 ok well that's
pretty much it really as I said that you
know one other thing is how what when it
comes to interpretation of the QQ plot
that's something we need help with
because I know we are more familiar with
Norma histograms and with QQ plots so
what I've done is I've done this right
I've actually presented this in another
video
look this is fantastic this is very
helpful because you can see that if your
distribution is symmetrical then the
corresponding QQ plot pretty much
everything lies on that line so I've
done like the major cases here I've done
the histogram corresponding to the QQ
plot or should I say QQ plot curves - a
histogram which we don't have to
interpret something which is negatively
skewed like this look that's what called
negative skew this is what the
corresponding normal QQ plot will look
like this deviation from the line and
you know since I made the previous video
I've had to look more at this and helps
try to think how how can we kind of
recognize this plot here without the
histogram we'll look at it what you can
see is that if you imagine this line as
the base the x-axis here what's going on
is that you've got major departure
towards the right hand - look this is
like lifting off it's same as the
histogram as you're going moving along
here you lift off see so just think
about it like that
likewise a positive skew that's the hump
there and then it goes down so it's like
up and then down here so it's the
deviation is at the beginning and then
it goes down just like this and down
down okay finally fat tails well this is
less if you do social sciences less
important just means that these are
tails around it's fatter than normal
distribution what you'll see is most of
the points lie on that line in that in
the center of the data points but
towards the end towards the tails you're
going to see a departure from the line
okay so that's pretty much it
don't to load your with anymore kind of
irrelevant information really all right
except I am tempted yeah just for the
eagle-eyed of you I just noticed the
theoretical quantiles going from minus 4
to plus 4 for a symmetrical case using
exactly the same data you just notice
that in a SPSS version it goes from
minus 6 to plus six doesn't it okay well
this is for the observed value this is
for theoretical quantiles so yeah I said
don't be worried about the numbers on
there just worry about the shape okay so
I'll leave you with the most useful
thing which is this okay I hope that's
been helpful