SPSS for newbies: Logistic regression Valentine's Day example Video Lecture | SPSS: For Beginners - Data & Analytics

hi everyone here's an interesting
question for those of you doing stats at
college just how do you work out chances
of getting a boyfriend or girlfriend I'm
going to show you how you could do this
using something called logistic
regression now here I've got data from a
survey I've done I've got a column here
called gender relationships yes or no
age previous number of relationships the
respondent has been in and whether the
respondent socializes before fitting the
model we note for any anomalies we have
one here someone has been in 99 previous
relationships well that is a bit out
character of the other respondents in
that similar age group so we remove that
we call an outline now we fit the model
do analyze regression binary logistic
will put the variables in the relevant
boxes and then press ok now i'm
interested in the output in the
variables in the equation what i want to
do is take these two columns to write
down asian these be here's a coefficient
of these hits with variables okay here
this first equation I take those numbers
have taken from the output and it's just
the usual equation where these variables
gender age previous number of
relationship socials they are going to
be different for each respondent
different for you and I the idea is that
you plug in the values for a particular
respondent to get a value which I've
called is EDD plug that is EDD into the
second equation and that will give you
the probability of being in a
relationship for that person let's
demonstrate female student at College in
London 20 years old no previous
relationships
goes out on the weekend we put these
values in to the model we get the value
minus naught point naught 77 plug that
into the second equation to get the
probability is forty-eight percent
chance of that such as student being in
a relationship based on my survey data
for a similar male we're going to get
the value of is said is minus 2.5 01 the
second equation where I've got a typo
here we plug my 2.5 01 instead we'll get
not five percent but about seven and a
half so much smaller compared to the
female now some remarks are in order
there so first yeah you can use this
equation and just plug in any values for
gender well gender age previous with
number of relationships socials and then
use the second equation to work out your
own probability remark here is that i've
done this quick and dirty I've not
thought very carefully about really
about the factors that might affect the
chances of being in a relationship so
you might think of a more sophisticated
model think more about the factors that
do affect the chances and include that
in your survey another thing I haven't
really done is I haven't tested my model
formally I have not for example tested
they cut the significance of each
coefficient as you would do in a
standard analysis or I've really tried
to show you is that you can use this
logistic model to reach to reach a
conclusion and hoping that you'll go
away and do a bit of exploration so have
fun all for now