<<CONTINUED FROM THE DOCUMENT CALLED “STATVIEW 4.0 DEMO DOC 1”>>
USE A TEMPLATE
• From the Analyze menu, choose Descriptive Statistics.
If Descriptive Statistics does not appear in the menu, choose Additional Templates instead. In the resulting dialog box, you can add any template to the Analyze menu by clicking in the check box to the right of the template’s name and clicking Change Menu. You can use the Descriptive Statistics template directly from this dialog box by selecting it and clicking Use Template.
• After you select a template, the Assign Variables dialog box appears.
• Click on the variable Horsepower in the list on the right. Drag to the “Variable(s)” slot on the left until the gray rectangle outline of the Horsepower variable appears in the slot.
• Release the mouse. The horsepower variable is placed in the template slot.
• Double-click on the variable Weight. It is also added to the slot, directly below Horsepower. Notice that the slot grows to contain multiple variables.
• Click the OK button and the results of the template appear in the view.
You have just created your first analysis using StatView. Notice how little time it took and how easy it is to use a template to generate results.
SAVING RESULTS
You can save your work at any time by saving the view. Remember that when you save a view in StatView, you are not simply saving the text or pictures of the output — you are saving all aspects of the work you have done. When you reopen the document, your work is as you left it, ready for you to pick up where you left off.
• Choose Save from the File menu. The directory dialog box appears.
• Name the view “Car Analysis” in the text box. Place the file in the Sample Data folder, and click Save.
• Close “Car Analysis” by choosing Close from the File menu.
• To re-open your work, choose Open from the File menu.
• Locate the file “Car Analysis” and click Open. The “Opening a View” dialog box appears.
The most common use of this dialog box is to leave the default settings and click OK. The default settings tell StatView to use the original variables, from the original dataset(s) and place the output in a new view. Using these defaults will open a view with everything as you left it when the view was saved.
• Click OK.
The view reappears exactly as you left it. In fact, the program did not even have to recalculate your results. By default, all results are saved with a view. If your data has not changed, you can re-open a view and get back to your analysis immediately.
Note, this demonstration version does not allow you to save datasets. As a result, this exercise of saving a view and applying to the original dataset will only work with the sample StatView 4.0 data shipped with this demonstration version. Using the full retail version you will be able to create views and save them with any dataset you create.
Before continuing on to the next example, close both Car Analysis and Car Data by choosing Close from the File menu. Note that you will want to close the view (Car Analysis) before closing the dataset. Do not save any changes to these documents.
EXERCISE 2: CREATE A DATASET AND ANALYSIS FROM SCRATCH
In this exercise you create a dataset that contains two variables: a continuous variable and a nominal variable. You then use the analysis and variable browsers (rather than a template) to analyze your data by performing an unpaired t-test. Once the analysis is complete, you will save your work as a template for future use.
As you complete this exercise, you learn that the steps you take to generate results are the same ones you follow to create a template. With this knowledge you can decide how you would like to use StatView to analyze your data. You can use the analysis and variable browsers exclusively or templates exclusively, or a mixture of both. The choice is yours. We recommend that you create custom templates to take full advantage of StatView and meet your specific needs. As this exercise shows, StatView makes this easy to do.
THE ATTRIBUTE PANE
• Choose New from the File menu. An empty dataset appears on the screen.
The rows above the body of the dataset contain attribute and summary information about the variable in the column below. Together, these rows are called the attribute pane, and you can show or hide as many of the rows as you desire. In a new dataset, the first five rows of the attribute pane are visible. These contain pop-up menus for each of the five attributes of a variable: its data type (string, real, integer, date/time, etc.), its source (whether it was user entered or created using a formula), its class (nominal, continuous, or informative), and the number of decimal places displayed (StatView carries 18 significant digits through all calculations and analyses; this attribute applies only to the way real numbers appear in the column).
CREATE A DATASET
First you will create a column (variable) containing continuous real data. The attribute pane shows the Type and Class for each variable. The default Type and Class for input columns are Real and Continuous. These defaults are appropriate for this variable so you do not need to make any changes before you enter data into the column. An empty cell appears below the attribute pane. This is called the input cell.
• Click the mouse in the input cell to select it, and enter the number 3.2. Press Return.
When you enter a value in the input cell, a new input cell appears directly beneath it and a new input column appears to the right. In this way the dataset grows to include as many rows and columns as are necessary to accommodate your data.
• Enter the numbers 4, 5.8, 6, 12, 8.5, 5.5 and 10.3 in the column. Move down the column using the Return key (or the Enter key if you are using a numeric keypad to enter data).
Now create the column containing the nominal grouping variable for the analysis.
• Use the Type pop-up menu in the attribute pane of the Input Column (the column to the right of Column 1) to change the variable’s Type to String. To do so, click on Real and when the pop-up menu appears, drag to String. Notice that the Class automatically changes to Nominal.
• Click in the top cell of the input column. Enter the values of the grouping variable as follows: Low, Low, High, Low, High, High, Low, High
Next you will name the two columns you have created and save the dataset.
• Select the name “Column 1” and type “Variable 1”. Use the Tab key to move to “Column 2” and type “Variable 2”. Click the mouse in another cell to enter the new variable names.
• Click on the variable attribute pane control between the two scroll bars on the right and drag down to expose the remaining rows in the attribute pane. You can examine descriptive information about the two columns you created.
• Double-click on the variable attribute pane control to hide all but the first five rows of the attribute pane. (Double-clicking a second time closes the entire attribute pane.)
CREATE YOUR OWN ANALYSIS USING THE ANALYSIS AND VARIABLE BROWSERS
Now you will use the variable and analysis browsers to analyze the data and create a template.
• Choose New View from the Analyze menu. An empty view appears on the screen.
• In the analysis browser, select Unpaired Comparisons and click the Create Analysis button above the list. A dialog box appears, allowing you to set the parameters for this analysis.
• The default dialog box settings, an unpaired t-test with an hypothesized difference of 0 between group means, are appropriate, so click OK. Empty placeholder tables will appear in the view with the variable requirements for the analysis noted beneath them.
• If the variable browser is hidden, show it by clicking the show/hide variable browser button at the upper right of the view.
• In the variable browser, select Variable 1 and click the Add button. An X appears to the right of it, indicating that it is assigned to the analysis.
• Select Variable 2 and click the Add button. An G appears to the right of it, showing that it is assigned as a grouping variable in the analysis.
• An unpaired t-test table and a group information table appear in the view.
You have just seen how easy it is to use the analysis and variable browser to perform analyses in StatView.
STATVIEW IS COMPLETELY INTERACTIVE
One of the unique features of StatView is that it is completely interactive. Any changes you make to the dataset, including the use of criteria (see Exercise 5), cause your results to recalculate automatically so the information in your results is always up-to-date. You do not have to start an analysis over from the beginning if you make a mistake or change a data value.
• Choose Untitled Dataset #1 from the Window ( or ∑ ) menu. Change the value in row 3 from 5.8 to 7.8. Press Return or Enter. The rotating yin-yang symbol indicates that the results are recalculating. In fact, whenever this symbol appears, you can cancel the current program operation by pressing Command-period.
• Choose Untitled View #1 from the Window menu to see the effect of this change on the results.
If you want to make many changes to your dataset, you may want to turn automatic recalculation off, by clicking in the Recalculate check box (or directly on the word “Recalculate”) in the upper left corner of the view.
Recheck the box after you have completed your changes, and results in the view will recalculate.
SAVE YOUR ANALYSIS AS A TEMPLATE
In a few steps, you created a template that performs an unpaired t-test on any set of data. You will now add this template to the Analyze menu for easy access.
• Choose Save from the File menu.
• Find and open the StatView Templates folder in the directory list. Name this file “Exercise Template” and click Save. (NOTE: you will see an alert recommending that you save the dataset before saving the view. Because you cannot save datasets with this demonstration version, simply click Continue and proceed with the following steps.)
• From the Analyze menu, choose Additional Templates. Select Exercise Template in the scrolling list and click in the checkbox to its right.
• Click the Change Menu button. The template you created is now available for easy use through the Analyze menu.
• Close the Exercise Template and the Untitled Dataset #1. Do not save any changes that you have made.
EXERCISE 3: WORK WITH TEMPLATES AND ACTION OBJECTS
One common use of templates is for a supervisor or statistical consultant to set up a template tailored to a particular task. The work is done only once. After that, anyone can use the template, simply by specifying a new dataset and variables. You can also create a template to speed the production of reports or journal articles that must be meet standard requirements for figure size and appearance, font size and type, etc.
In this exercise, you will use the template you created in the last exercise and the analysis and variable browsers to generate tables and graphs. Finally, you will customize a graph.
APPLY A TEMPLATE TO NEW VARIABLES
First, make sure that all datasets and views from previous examples are closed.
• From the Analyze menu, choose Exercise Template. The directory dialog box appears for you to locate the dataset you wish to analyze.
• Locate Lipid Data in the Sample Data folder and click Open.
The Assign Variables dialog box appears for you to assign variables to the template.
Note that the order of template slots is not fixed. Either Variable 1 may appear on top (as pictured above) or Variable 2 may appear on top. Make sure that as you continue the exercise below you assign the variables to the correct slot.
HINTS
On-line help, in the form of a Hints window, is available for almost every feature of the program. Hints explain everything from the function of dialog box buttons to the buttons in the analysis and variable browsers. The Hints window is a valuable tool to help you understand how to use StatView. You will use Hints here to familiarize yourself with the use of the Assign Variables dialog box.
• Choose Hints from the Window menu. The Hints window appears on the screen. It is a floating window that floats above all other StatView windows, including the browsers and dialog boxes.
• Click in the slot labeled “Variable 1” in the list on the left of the Assign Variables dialog box. The Hints window describes how the variable you drag into this slot will be used in the view generated by this template. Select “Variable 2” to read a hint about its function in the template. Close the Hints window to get it out of the way. You may choose it from the Window menu any time to learn about the program.
You will now assign variables to the slots in the template. These variables contain data on the percent of ideal body weight of students before and after three years of medical school.
• Drag the variable “% ideal body wt.” into the Variable 1 slot.
• Use the scrolling variable list to locate “% ideal body weight-3 yr”. Drag it into the Variable 1 slot as well. (You can also double-click on a variable to assign it to a slot. In addition, holding down the Command key and double-clicking allows you to control which slot variables go into.)
Since you assigned both variables to the same slot, they will be used identically in the analysis. When a template contains several variable slots, each slot represents one variable in the original view. As you assign a variable to a slot, it is used in all the places the original variable appeared. If you assign two variables to one slot, then, everywhere the original variable was used there will now be two.
• Select the Variable 2 slot by clicking on it. Double-click on Gender in the variables list to assign it to the Variable 2 slot (the grouping variable in the analysis). Click OK.
A new view opens containing the results of the template. As you can see, four tables are generated, one pair for each continuous variable assigned.
ACTION OBJECTS AND ADDITIONAL OUTPUT
Now you will be introduced to another central feature of StatView: action objects. All output generated by the program, whether tables or graphs, are “action objects,” because they retain information about the variables and analysis parameters that define them, and contribute this information to subsequent analyses. When you click the Create Analysis button or assign a variable with the variable browser, StatView uses the attributes of the selected action object to determine what to do next, so you do not have to respecify all the information. An understanding of how action objects work is important for taking full advantage of the speed and ease with which you can analyze data.
To use action objects, you must be aware of which variables and results are currently selected. StatView provides several ways to monitor selection: (1)You can see the selection handles around selected results; (2) the Results Selected note in the upper right corner of the view reports the number of results selected; and (3) usage markers to the right of variables in the variable browser reflect usage in the currently selected result.
In the view that you just generated, notice that the second set of tables is selected.
• Deselect the tables by clicking in any white (empty) space in the view. Select the first unpaired t-test table (Unpaired t-test for % ideal body wt.) by clicking on it so that only one action object is selected. Notice how the variable browser usage markers and the Results Selected note change as selection changes.
• In the analysis browser, click on the triangle next to Cell Plots. In the indented list that appears beneath it, double-click on Bar Chart (this has the same effect as clicking the Create Analysis button). The Cell Plot dialog box appears.
• The default settings are appropriate, so click OK. A graph appears with two bars, one for the mean of male body weight and one for the mean of female body weight.
The bar chart that appears in the view contains the same variables as the unpaired t test table, yet you did not have to assign variables for the second analysis. You can see that the bar chart visually confirms the results of the unpaired t test which indicate no significant difference between the means of male and female body weights.
• Follow the same procedure to create a second bar chart for % ideal body weight - 3yr.
Next you will see how to use the information in action objects to quickly generate additional analyses for new variables.
ADDING A VARIABLE TO A SELECTED TABLE
You can assign additional variables to an analysis by selecting the result (table or graph) and using the variable browser again.
• Deselect all output by clicking on any empty space in the view. You can confirm that nothing is selected by looking at the Results Selected note. It should be blank.
• Select the unpaired t-test or group info table for % ideal body wt. in order to access the information on the variables, analysis and parameters of these action objects.
Notice the usage markers in the variable browser next to the grouping variable (Gender) and the dependent variable (% ideal body wt.). You can see this information more conveniently if you change the display order for the variable browser to “by Usage.”
• Click on the Order pop-up menu and select “by Usage.” Notice that the variables currently being used by the selected analysis appear at the top of the browser.
• Select Cholesterol in the variable browser and click the Add button.
A new unpaired t-test is created combining the new continuous dependent variable Cholesterol with the original grouping variable Gender. You did not have to respecify either the analysis, its parameters, or the grouping variable.
Notice that these two new tables are selected when they are created. The Results Selected note and the variable browser have updated to reflect this.
SPLIT BY
There are two ways you can easily examine results for subsets of your data. One is to use the Criteria feature, which is demonstrated in Exercise 5. Another is to use the Split By button in the variable browser. When you assign a nominal variable as a split-by variable, the results are generated for each group of the nominal variable. You will now use the Split By button to examine the effects of alcohol use on the body weight of male and female students in this study.
• Select the % ideal body weight-3yr bar chart by clicking on it.
• In the variable browser, select “Alcohol use” and click the Split By button. The bar chart changes to display bars for each level of alcohol use in the study group. The legend to the right of the graph identifies the levels of alcohol use by fill pattern in each bar. There are two missing bars (male >6 and females none), indicating that no subjects fall in those cells.
While this example shows how to use the Split By button with graphs, you can just as easily use it with tables in order to see statistical tables broken down by the groups of a nominal variable. You can do all of this without having to re-order your dataset. You can apply a split-by variable to all the analyses generated in StatView.
<<CONTINUED IN THE DOCUMENT CALLED “STATVIEW 4.0 DEMO DOC 3”>>
