SPSS - Quick Primer
While there are various SPSS help
resources that will be made available to you (eg Steve Jost's, Prof Sabatini's,
etc), this file is intended as a very quick primer on the basic things you'll
need to do to get started using SPSS.
Accessing/copying files between
the local machine and the remote machine: Recall that you will want to somehow
copy all of the files from the book's CD onto the remote computer. Review the
steps discussed in lecture on how to do this. Vert briefly, it involves
changing the settings when you start Remote Desktop. Be sure that the 'Options'
button is selected so that you can see all of the tabs. Choose: Local Resources
tab >> under 'Local devices and resources' click on 'More' >> click
on the plus sign next to 'Drives' >> select the drive(s) you want to be
able to access from the remote machine. Note that these drives will show up only
when you navigate up to the folder/directory level that shows all of the drives
on your computer.
Opening files from your textbook in
SPSS: Your textbook uses files with the extension 'POR': However,
SPSS defaults to looking for files with an extension of 'sav'. What this
means is that after clicking on File >> Open, you will need to change the
'Files of Type' option away from 'sav' to 'por'. For this problem, and
any subsequent problem that asks you to draw a graph, be sure to paste your
graph into your Word document.
Random Tip: If you get tired of having to always
choose 'POR' file, you can try the following: Rename all of the files from
their POR extension to SAV. The easiest way to do this is from the command line
(for those of you wh know how to use it). As far as I can tell, this does
not harm the data.
PASTING CHARTS FROM SPSS INTO A DOCUMENT:
Sometimes
you will find that SPSS does not properly allow you to copy/paste. For example,
you may find that it only pastes half of the chart. Here is a method that will give you a nice
image of any chart you wish to paste:
Right
click the chart --- choose EXPORT --- Under ‘DOCUMENT type’
choose ‘Graphics Only’ --- Under ‘Graphics Type’ choose JPG (probably your best
bet) --- Choose a location and save.
(Remember
that this file is saved on the remote machine unless you specify your local
computer).
Scatterplot:
1.
From
the 'Data Editor' view, choose: Graphs >> Legacy Dialogs >>
Scatter/Dot
2.
Choose
Simple Scatter, and click 'Define'
3. Remember to put your explanatory variable on the x-axis and your response variable on the y-axis
4. Click OK
To generate a line:
1. In the 'Statistics Viewer' view, double click the graph
2. This will bring up the 'Chart Editor' view.
3. Look for the little icon of a scatterplot with a line through it. You'll know you're on the correct icon if, when you hover over it for a second, you see the label 'Add Fit Line to Total'
4. In the 'Fit Line' tab, choose 'Linear' (since you are trying to see if there may be a linear relationship), and click Close.
5. You can close the 'Chart Editor' view. You will now see a line in the 'Output' view.
It probably wouldn't hurt to go through this process 2-3 times to get
comfortable since you'll be doing it so much. I'd begin all the way at
the beginning. In other words, re-open a data file and go through the whole
thing again until you're comfortable opening files and creating a scatter
plot. Other graphs have their own sequence of commands, but are often
similar to scatterplots.
Correlation ('r'): Analyze
>> Correlate >> Bivariate. Choose your variables, and click
on OK. The Pearson Correlation is
your value for ‘r’.
Drawing a histogram: Graphs
>> Legacy Dialogs >> Histogram à select your field.
Plotting a curve over a histogram:
Click on the histogram to open the ‘Chart Editor’. Click the icon that
shows a curve (hovering will display the label ‘Show Distribution Curve’)
Normal Quantile Plot: Recall that
plotting two quantitative variables using this plot can give you an idea
whether the data follows a "Norma" distribution. (eg you ca do things such as
calculate z-scores, etc). To plot in SPSS: Analyze >> Desc
Stats >> P-P Plot. Simply choose your variable and click ‘OK’.
Boxplots: Graphs >>
Legacy >> Boxplots. Choose ‘Summaries of SEPARATE variables’
Finding descriptive statistics such as
mean, standard deviation etc:
Analyze >> Descriptive Statistics >> Descriptives (for things like mean, SD).
You can select which stats you want by
choosing ‘Options’. You can get even more information such as
Medians by: Analyze >> Descriptive
Statistics >> Explore.
Descriptives including quartile scores:
Analyze >> Descriptive Statistics >> Frequencies.
Select the
column. Click the ‘Statistics’ button
and choose all information you want to see (e.g. quartiles).
Listing z-scores
for all datapoints in a series:
Begin determining the descriptive statistics (as discussed earlier: Analyze >> Descriptive Statistics >>Descriptives). Select the column for which you wish to find the z-scores. Then check ‘Save standardized values as variables’ at the bottom. This will create an additional row in your SPSS table showing the Z score for every datapoint in the column you selected.
Regression Analysis:
Analyze >> Regression >> Linear.
Choose
your explanatory and response variables.
Note that SPSS still uses the labels ‘dependent’ and ‘independent’. Recall that we are avoiding these labels in
favor of explanatory (independent) and response (dependent).
If
you are also interested in the residuals: choose Save, then under Residuals,
select ‘Standardized’ and click Continue.
Then click OK.
You
will see a few tables. The ‘Model Summary’ table gives you r and r2. If you want to include the regression line,
you use the technique to generate the line as described above in the ‘Scatterplot’
example.
Look
for the table labeled ‘Coefficients’. In
the coefficients table, the information for your regression line is under the
column labeled ‘B’. The top one (183.08
in this example) is the y-intercept. The lower value (-0.184 in this example)
is the slope.
Coefficients |
||||||
Model |
Unstandardized Coefficients |
Standardized Coefficients |
t |
Sig. |
||
B |
Std. Error |
Beta |
||||
1 |
(Constant) |
183.080 |
53.550 |
|
3.419 |
.014 |
|
-.184 |
.366 |
-.201 |
-.503 |
.633 |
Residuals:
For
residual analysis, see ‘Regression Analysis’.
This step adds an additional column in the Data Editor view that shows
the residual for each datapoint. Then
proceed as though you were creating a scatterplot: Graph >> Legacy
>> Scatterplot, choose ‘Simple Scatter’ and click Define. In this case,
however, you place the Residual column on the y-axis (instead of the response
variable).
Incidentally,
you will notice that the residual line does not give you the 0-axis line that
we have seen in our examples. To get one, double-click the graph, and look for
the little icon that shows a straight horizontal line. (Hovering will display:
“Add a reference line to the y axis”). When the ‘Properties’ window appears,
you can type 0 under ‘Position’ assuming that is where you want the line. Click
‘Close’.
Stemplot:
Analyze >>
Descriptive Statistics >> Explore.
Choose the
variable you want to plot. Click on 'Plots' >> check Stem and Leaf. Click
Continue, click OK.
Pivot Tables:
Analyze >> Descriptive Statistics >> Frequencies.
Choose the two columns you want to create a pivot table out of. Be sure the ‘Display Frequencies Table’ checkbox at the bottom is selected.
Crosstab Tables:
Analyze >> Descriptive Statistics >> Crosstabs.
Choose the two columns you want to create a crosstab table out of.