SPSS for newbies: Quickly create dummy variables for logistic regression Video Lecture | SPSS: For Beginners - Data & Analytics

hi everyone Phil from Statistics mentor
calm
continuing with the videos on logistic
regression I'm going to show you here
how we can easily create dummy variables
in logistic regression okay for this
example we've got the titanic data so we
have as a dependent variable survived
which it's about 1 code in one V
survived passenger survived zero
otherwise passenger the class of travel
of the passenger P class is one of three
classes 1 2 & 3 so we're going to kind
of regress survived on P class you can
see if the class of the passenger that
helps to predict whether the passenger
survived now we can see that P class is
not quantitative it is qualitative
you're in what you travel we're serving
only one of three classes so we need to
since it's got three categories 1 2 & 3
the rule is to avoid the time available
trap we create two dummies with one as a
reference and we also include the
intercept now how do I create the two
dummies for this we could do it the long
way doing it manually or we can use a
neat feature in this PSW SPSS which I
wish was available for standard
progression okay so what we do is you
take over into this sub dependent
variable box DV which is survived and
the co-vary class which we know is
categorical now look up here we've got
to create a dummy for p class because
it's code
one two and three at the moment we need
to we need to create two dummies go to
categorical call where is it take that
into the categorical covariance now look
under here unchanged contrast these
things now are highlighted contrast we
want it to be an indicator function so
leave that alone we need a reference
category the reference category can
choose the first or the first meaning of
this
think that ice coded with the lowest
value or the last which is a thin coded
with the highest value so by default
it's going on last or we can leave it
there which is same as saying since
they're three classes we leave out class
three click OK and run right so what we
see is that in the variables in equation
we've got a constant and we've got those
three categories but one of them being
the third class has been left out which
is why there is a empty space in the
coefficient there but the other two
class 1 & 2 they are in there alright so
we've got a coefficient on P class 11.6
and B Class two point seven nine eight
I'll show you that we could do this that
this indeed gets you the same result
I've earlier created the dummy variables
manually d1 for first class d2 for
second class third class is the
reference now if I run this logistic
without using the automatic
dummy creating function for these two
into covariance so these are say okay
we're checking that the coefficients are
the same alright so here you go constant
the d1 which is the first class 1.63
what we had before d2 point seven nine
eight so indeed that I saved us a lot of
service a bit of work there by using the
category function as I said I wish that
this category function was also
available in standard regression all
right that's nice short video