SPSS for newbies: Chow test for structural change part 1 Video Lecture | SPSS: For Beginners - Data & Analytics

hi guys Phil from Statistics mentor calm
so you'd like to run the child cast in
SPSS great well you've come to the right
place but it's one of the more unusual
requests I have because if you're
wanting to use a child test that usually
means that you're studying Econometrica
in which case you're better off using
one of the packages more suitable to
econometrics such as Stata eviews or
even our SPSS is not exactly a great
package for doing econometrics but let
me show you how to do the child test
anyway first of all why would you want
to do the child test in case you're
watching the wrong video you'll want to
do a child test if in layman's terms you
want to see whether a regression model
is the same regression model is
appropriate to explain the relationship
between the dependent variable and the
self explanatory variables between two
groups those two groups could be some
very qualitative things for example
gender is the in our example is the
relationship between earnings and tenure
the time spent in a company the same for
males and females or it could be
something to do with time related issue
such as is a relationship between x and
y's the same pre-war post-war or pre
policy post government policy and so on
note the difference between the child
test and the T test for comparing the
means between two groups if you simply
want to compare that the dependent
variable is whether they're the same the
mean value of the dependent variable is
the same between two groups that's just
the normal t-test we're not doing a
t-test here because we're trying to see
whether the a regression model
explaining the Y from the set of X's is
the same for two groups
okay now then moving on if you still
want to do this since it's a test we
need to state the null and the
alternative the null hypothesis is
undoing Damons terms first is that the
relationship between the dependent
variable and the independent variables
or the explanatory variables are the
same between the two groups another way
of saying that is to say that the
coefficients of the regression model is
the same across both groups more
technical way of saying that is that the
intercepts and the partial slope
parameters are the same between the two
groups and yet another way of saying
that even more esoteric is to say that
there is no structural change the
alternative hypothesis is there is
structural change
so that's stated null hypothesis no
structural change or model is the same
for both groups in this case males and
females first alternative that is a
structural change in my example here I'm
going to regress earnings which is going
to be my dependent variable on tenure at
the time spent in a company I wanna see
the relationship between earnings and
tenure is the same for males and for
females so I want to do a child test
second step after the stating the known
alternative is compute the test
statistic it's an F statistic now SPSS
doesn't spit out the F value for you for
the child test so you're going to have
to do it by hand and here's the formula
we need to run the procedure is we have
to run three regressions one or the
entire data so ignoring that there's a
difference ignoring the groups and then
which and then we get from that the
regression group residual sum of squares
typically you'll see them books they'll
put P subscript P which and is pulled
because you pulled the data across both
groups in other words you use all the
data that's regression one regression
and then you have to run two more
regressions and those two regressions
are running it on subsets you run the
same regression model but on the subset
so one or Group one one of group two
so here I have to run one for the male
group one for the female group and
record the residual sum squares for
those groups add them together n stands
for the number ops of observations and P
stands for the number of parameters in
your model and then mean part of the
model let's demonstrate this now in SPSS
so let's run the let's obtain first the
RSS residual sum of squares for the
whole lot of the data so I'm simply
going to regress earnings continue go to
analyze regression linear you put the
earnings independent box tenure in the
independent box and just click OK right
then well you've got a head down is to
the a know of a table a look for the sum
squares and the residual sum of squares
that's the figure okay we have to record
that again recorded that now I go back
and I'm not gonna have to run two more
regressions two more regressions
one for each subset well to do that go
to data go to split file and then you
want to say organize output by groups or
you can do compare groups it's fine
click on the dummy variable female which
splits the group into male and females
okay and then rerun the regression
analyze Russian yeah okay
right so we're going to have to lots of
regressions now one for each group so
first of all dummy variable female
female just just ignore that that's just
a label to say female dummy right we're
looking at the male group male group RSS
is there