Tip:
Highlight text to annotate it
X
Well, friends and colleagues, it's time now for us to use
SPSS to conduct multiple linear regression.
Now, I want to remind you, of course you know that SPSS is a
very powerful piece of statistical software.
The student version of SPSS is very much worth the cost.
Of course mine's an Apple version, there's IBM versions.
It's pretty cool.
The power of this software is going to especially be evident
as we progress towards more demanding analyses.
Multiple linear regression will require that we test the
assumptions and conduct the analyses for our problem.
Don't panic.
This will be fun.
We love SPSS, and we're going to have a blast doing multiple
linear regression.
Well, my friends, here we are, back with SPSS.
And you may recognize the data set that we have here.
This is the very same data set that I use
to do factor analysis.
We have four Texas public school districts for 2011.
We have the percent disciplinary placements, the
percent of students that are African American, percent
Hispanic, percent white.
The percent that are economically disadvantaged,
and the percent that are limited English proficiency.
the percent of students that are at risk, and the percent
of special ed students.
Now, what I propose to do in this particular data set is
that I want to make the percent of disciplinary
placements the dependent variable, and I want to use
the reminder of these as the independent variable.
So in order to do a multiple linear regression the first
thing I do is go up to analyze.
Now, remember how easy it would have been to jump on
general linear models?
Well, we need to go down to regression,
and let's go to linear.
Now, you might say to yourself, but wait.
That's where we went with simple linear regression.
Well, it's where we're going to go with multiple linear
regression.
We will click on it and up comes this
beautiful screen here.
Now, our dependent variable, again, is the percent
disciplinary placement, so we will place it there.
And the remainder of our variables will be our
independent variables.
Now we will do this as a stepwise analysis, and let's
go up and choose the statistics that we want.
Now, of course we want our descriptives, we want our part
and partial correlations, and our collinearity diagnostics.
This will assure us that none of the variables that we use
are, in fact, just another form of the same thing.
OK, the plots.
I think we might do some plots.
And our Y, let's see, we want our Z residual to be our Y,
and our Z predictive to be our X, and we will want a
histogram and a normal probability plot.
I think we're ready to go.
We can do it stepwise, or we can just do enter it, it
doesn't make any difference.
And let's see what we get.
Here it runs.
Now let me move this over to get it in the screen.
Man, I mean, it's a kicking.
Now, if you were out there doing this on your own, I
assume that it would be pretty rugged.
We now have a regression model.
I'm going to click right here, and I'm going to put By Dr.
Dawg, just to show you how that you might
enter in the amounts.
And I'm going to do that, and there we go.
So this one is done by Dr. Dawg.
We have the descriptives.
Notice each of these, we have the mean, the standard
deviation of the number.
Obviously, we had 1,230 school districts in the sample.
We have our correlations.
We can go right down through here and we can see how all of
our variables correlate with each other.
Have a lot of other stuff that we don't
need, the model summary.
Our R is 0.12.
Oh, not a real strong correlation,
a little bit weak.
But that's all right.
We have our beta coefficients.
My gracious, look at this.
We have the standardized beta coefficients.
And who knows, we might somewhere down here have a
constant, if we wanted to look at it.
Now, this is pretty cool.
It takes the ones that are the most important and then it
actually built us four variables
that were very powerful.
We'll come back and look at this in just a minute.
We have our collinearity diagnostics.
We don't have any collinearity.
We'll check that.
We have a histogram, a residual plot, a plot.
Man, this is really cool stuff.
So what you've seen is that you can go in with SPSS, and
with a click of a few buttons you have the
diagnostics that you need.
Now we'll go next to a video that interprets the
diagnostics for us.
Hang on.
Here we go.
Well, we got that SPSS analysis conducted.
All we have to do now is learn to read it.
It shouldn't be too bad.
Again, I want to thank you very much for your support.
May the odds be ever in your favor.
Live long and prosper.
Peace and long life.
You're coming along in this.
Just keep plugging.
You're going to get it done.
Have a great one.
This is the Dawg, signing off.