SPSS 8.0 step-by-step evaluation guide

Mouse clicks

Description

I'm a market researcher who wants to try to find out key information about the auto industry. I could just as easily be an analyst for a government or academic organization.

Show ODBC Wizard on database

- click on File menu

- highlight the "Database Capture" item on the File menu

- select New Query. (from the Database Capture box)

- select MS Access

- click Next

- select Cars.mdb

- click OK

Drag Manufacturer and Model Year from Available Tables to Retrieve Fields in this Order.

Click Next

Click Next (until you get to step 6)

Click Cancel

- click on File menu

- highlight the "Database Capture" item on the File menu

- select Run Query. (from the Database Capture box)

- select SelectMo.spq

- enter beginning model year 1971

- click OK

enter last model year 1979

click Cancel

select File, Open

- select Cars5.sav

- click OK

click on variables weight, horsepower and engine displacement in data editor

First, I need to be able to get data from whereever it resides. I start with getting access to the data from my relational database. This data could be stored in any ODBC compliant database such as Oracle, Sybase, or Access. I use the ODBC Wizard to build a data query. The Wizard prompts me step by ste 20420x2320u p to retrieve my data. I just specify the type of database I have, such as Oracle or Access, and answer the questions. We call this "SQL-free hand held query." The wizard displays the database tables, so I can easily select which data elements I want. Then, I get a visual representation of the chosen tables with links. Now, I can specify which subset of this data I want. I just select the field and give screening criteria. For example, I want information of the automobile manufacturer as well as information about specific models. It's easy to do, but often an MIS person would setup this query once, and save it for me. Step 3 lets us specify relationships, such as relating ID in one table to ID in another table. In step 4, I can limit the cases that my query will retrieve, and in step 5, I can define my variables if they aren't already defined. As you can see in step 6 of the wizard, SPSS handles all the syntax for me which I could edit if I wanted to.

I can also run queries which were saved before that I can use to select more specific data. Here's what a saved query with prompt looks like. All of the SQL code to access the database and extract the data I need is hidden from me. I just answer which years I want the data for and it is downloaded to my PC.

Since the data we want is already in an SPSS data file, let's go ahead and just access the data by opening the file. As you can see, I have information like vehicle weight, horsepower and engine displacement.

- select Statistics, Summarize, Layered Reports.

- move to Summary Variable(s): box
Time to Accelerate
Engine Displacement
Horsepower
Miles per gallon
Vehicle Weight

- move to Grouping Variable(s): box
Number of Cylinders
Country of Origin
Model Year

- click on the Statistics. button

- remove "Sum" and "Percent of Total Sum"

- click on Continue

- click on OK

Easily create multi-dimensional pivot tables

I want to get a quick overview of my data, and the new Layered Reports feature in SPSS improves upon SPSS pivot tables by letting me drill down for more information. I want to see how cars compare by looking at their acceleration, displacement, horsepower, mileage and weight.

I can also look at cars by the number of cylinders, country of origin and the model year, and I can select only the statistics I want to show.

The layered report is a multidimensional report cube that quickly and simply lets me put all of my data into a single layered report that I can slice and dice in any way I want.

Point to both sides of the Viewer, output pane and contents pane.

- double click on the Layered Reports table to activate it (If the pivoting trays and formatting tool bar do not appear when you activate the pivot table, go to the Pivot menu and select Pivoting Trays; and from the View menu, select Toolbar.)

- click on Results Coach icon (far right) on the Formatting toolbar

- page through the help

- click on the drop-down boxes for "Country of Origin" and select different values, ending with "American"

- select Pivot, Bookmarks

- type in "American Cars" and click on Add

The layered report and all charts are created in the Viewer window, which is like the Windows Explorer, which helps me organize and quickly see all the layered reports, charts, notes, logos, video and sound files that I want to use in my report.

I'll just click on the pivot table so I can work with my results. The pivoting trays make it easy to change the variables that I can see, and the formatting toolbar lets me change the table, or launch another key feature, the Results Coach.

The Results Coach gives me clear examples of the kind of table I just created, in this case, a Layered Report. As I page through, I get complete help to interpret my table.

Now let's dig into the table. As you can see, we're looking at the total information, but what if we want to see information on just American cars? We can use the layered reports drop down boxes to get information on American cars.

I can bookmark key information so I can zoom in to find it later, or let my report recipients find information quickly, so I'll create a bookmark called American Cars.

- move the "Country of Origin" icon in the Pivoting Trays from the Layer dimension to the Column dimension

- move the "Statistics" icon in the Pivoting Trays from the Column dimension to the Layer dimension

- click on the drop-down box for "Statistics" and select "Mean"

- select Pivot, Bookmarks

- type in "Country Comparison" and click on Add

- go to the different bookmarks

SPSS recently won a Smithsonian award for its pivot tables because they help uncover key information. We can easily pivot our table to compare American cars versus Japanese or European cars, and we can select the statistic we want, in this case the mean.

We see that American cars have quicker acceleration, larger engines, more horsepower, poorer gas mileage, and they weigh more than their European and Japanese counterparts.

This is interesting, so we can add another bookmark called Country Comparison.

Now I can easily switch between table views by choosing the bookmarks from the toolbar.

Graphs, Interactive, Scatterplot.

Y=Horsepower

X=Engine Displacement

From Options tab, set the size of each axis to 3.00 inches.

Click OK

- double-click on graph to activate

- click on "Assign Graph Variables" icon (icon that is red, blue and yellow)

- click on 3D graph button (icon with 3 arrows)

- Z=Time to Acceleration

- rotate chart with rotation controls

Visually explore data like never before

Now that I'm familiar with my data, I can start visually exploring the data to uncover key findings with the amazing new SPSS 8.0 graphics system. A scatterplot is a chart that's very helpful for exploring data. I suspect there's a relationship between Horsepower and Engine displacement, so I just drag those variables over to create my chart.

As we create this chart, you'll see this is dramatically better than the way other products create charts, including SPSS 7.5. I can make the chart 3D with one click and just drag and drop time to accelerate over to change our chart.

As I suspected, the scatterplot shows a relationship between horsepower, engine displacement and acceleration, and I can see this relationship with SPSS' very intuitive rotation controls. I can get an interactive, three dimensional view of my data.

Graphs, Interactive, Scatterplot.

Y=Horsepower

X=Engine Displacement

Z=Time to Acceleration

From Fit tab, choose Smoother

From Options tab, choose Chart Look "Marina" and set the size of each axis to 3.00"

Click OK

- double-click on graph to activate

- rotate chart with rotation controls

Create truly interactive and fast surface plots

SPSS 8.0 also offers the most powerful statistical-based smoother available. It works with many data conditions, and types, including non-parametric and categorical data. As we rotate the smoothed plot, we can get a crystal clear picture of the relationship between horsepower, engine displacement and acceleration.

- Double click on the first scatterplot you created to activate it. (it should simply be a plot of circles with no smoothing)

- click on "Assign Graph Variables" button

style: Country of Origin
(remove style)

size: Miles per gallon
(remove size)

color: Vehicle Weight

- change graph to 2-D (icon with two arrows on it)

- Panel Variables: Country of Origin

See a lot of information simply

Now that we know that there's a relationship between engine size and horsepower we want to know more about where these cars come from, how good their mileage is, and how heavy they are.

As I drag over variables, I can put even more information in the chart and watch the charts update on the fly.

I can drag and drop the Country of origin variable to style to see where the cars came from. In this case, I see that American cars, which are represented by the circles, account for most of the fast cars with large, powerful engines.

I can drag and drop the MPG variable to size to see which cars got good gas mileage. In this case, I see that cars with good gas mileage, which are represented by the large circles, have small engines, low horsepower and poor acceleration.

I can drag and drop the Weight variable to color to see which cars were the heaviest. In this case, I see that heavy cars, which are represented by the blue and purple circles, have high horsepower, large engines, and quick acceleration.

I think the most interesting way to focus on this analysis is by looking at where the cars came from, so I can split the chart into several slices by paneling the charts based on Country of Origin. With these powerful features, I can easily see that heavy cars with big, powerful engines are mostly American.

- Graphs, Interactive, Pie, Plotted.

- Y=Horsepower Groups

- X=Number of Cylinders

- Slice By: Country of Origin

Point to blue pies

Point to red pies

Point to at 2^nd column of pies

Show more information with new chart types

SPSS includes many new chart types including a plotted pie which shows me even more interesting results from categorized data, which is really useful for summarizing data. We'll select plotted Pie and look at Horsepower Groups vs. Number of Cylinders vs. Country of Origin.

When I look at the blue pie charts, it shows me that 5 cylinder cars come from Europe, and by looking at red pie charts, I can see that 8 cylinder cars come from America.

Looking more closely, I can see that that 4 cylinder cars come from all of the countries, but there aren't too many American 4 cylinder cars with a lot of horsepower.

The beauty of all this is that I can keep visually exploring my data. When I look at these pies, I can see that American cars with small engines tend to have less horsepower than their European and Japanese cars.

- Graphs, Interactive, Bar.

- Y=Horsepower

- X=Model Year

- show "Bars Represent." drop-down box TAB?

( - double-click on graph to activate

- double-click on number inside of bar

- under "Bar Labels" uncheck "Values"

- click on Apply )

- click on "Assign Graph Variables" button

- click on 3D graph button

- Z=Country of Origin

- click on the "Summary Function" tab

- select "Median" and click on Apply

- select "Variances" and click on Apply

- select "Mean" and click on Apply

- click on OK

- drag & drop "Engine Displacement" onto Y-axis

- drag & drop "Number of Cylinders" onto X-axis

Discover information that you wouldn't be able to see easily.

All this exploration has let me see a lot of information, but I suspect there's more information when I look at the data over time. I can drag Horsepower over and drag model year. Now I see that the average horsepower has dropped over the years.

(If necessary, remove the "bar labels")

I know that the country of origin is important, so if I drag Country of Origin into the chart, I can see that the that the engines in American cars decrease dramatically, but Japanese and European cars changed very little.

Now I can see how variance shows where the biggest changes occurred by selecting variance instead of mean as the summary function. I can see that horsepower in American cars in the early and mid seventies was much higher, American cars become more alike by the early eighties. This is probably due to the energy crisis of the early 70s and the demise of the powerful muscle cars.

Since my graphs update on the fly, we can quickly look at how engine size has changed over the same time. We can just swap horsepower with displacement.

Earlier, we had seen that engines in American cars have more cylinders than European and Japanese cars. Now, we'll take a closer look at this relationship by swapping model year with number of cylinders. Here, we see that not only do American cars have larger engines, but that American four and six cylinder cars have larger engines than their European and Japanese four and six cylinder counterparts..

- close the "Assign Graph Variables" window

- show fills, textures, and lines

- change the symbol of a data point

- change the color of a data point

- move the title (only slightly)

Get a great professional look

Now I want to present my key findings and make my charts look just right. SPSS 8.0 comes with an incredible number of features to make charts look great!

Each chart has special toolbars that let me work with the chart and edit it like I never could before. I now have amazing control over almost any element of the chart as I change fills, textures and line types, and even directly edit specific parts of the chart like the axes, labels and the plot points. I can even make notes directly on the chart if I want to.

I can also move around chart elements like labels and legends just by grabbing them and moving them anywhere I want.

- double click on a bar

- select Bar Width

- move 2^nd bar slider to 65%

- click on the "Lighting" button and play with the settings.

- Choose the lower right hand light source

- Right click on Japanese bars and choose "select this origin"

- select properties

- click on Bar Fill Style drop-down box

- select the Japanese flag

- click on OK

(Note, you may have to go to the More styles dialog and browse for us.bmp, europe.bmp and japan.bmp. Also, before you select a texture, be sure that the texture color filter is set to "no color")

- rotate the graph so that you can easily select the European bars and repeat the above process

- repeat for the American bars

- double click on a bar

- under "Bar Shape" select "Circular Base"

- click on O.K.

- slowly rotate the graph

I can also easily change the width of the bars. I'll make the bars a little less wide using this slider control so they have less of a rectangular look.

Now, comes the part where I make the chart shine, literally! I can go to the lighting effects and make a chart that looks really dramatic by changing the direction the light shines, and the brightness of the light!

One of the most dramatic new features are the new texture fills. I can use the ones provided with SPSS 8.0, or I can use my own.

For example, I can show the American cars with an American Flag, the European cars with a European Flag, and the Japanese cars with a Japanese flag.

I want this chart to stand out, so I'll make the shape of the bars circular.

Now I have a beautiful chart that's ready for presentation!

(Use the following instructions until the SPSS Smart Viewer 8.0 for Windows is available. Note, when there is no live data in the Data Editor, the Viewer window works just like the SPSS Smart Viewer 8.0 for Windows will when it ships in late 1997.)

- Save file as report.spo

- Close SPSS 8.0 for Windows

- Start SPSS for Windows

- Choose File/Open

- Select report.spo

- Activate layered report,

- Access bookmark for Country comparison from Formatting Toolbar

- Double click on the first scatterplot you created to activate it. (You may need to scroll to the top of the report to see it, or you can use the tree list in the left pane)

- Switch any variables to watch chart update on the fly.

Make reports come alive with SPSS Smart Viewer

SPSS Smart Viewer is a great way to electronically share reports and empower other people to interact with SPSS' award winning pivot tables and unique interactive charts.

I can save my report and add notes, titles, video, sound etc. and the layered reports and charts that I created will stay live, so I can send the report to someone like a marketing manager who has SPSS Smart Viewer, or another SPSS 8.0 user. When people receive my report, they can pivot and change the live charts to make their own discoveries just like I did, even if they don't have SPSS installed.

NOTE: One copy of SPSS Smart Viewer is provided to the SPSS user, for their own use, with each copy of SPSS 8.0 for Windows. Any other report recipient who isn't an SPSS 8.0 for Windows user must purchase a license for SPSS Smart Viewer. Flexible volume discounts are available.