Blogs

Bookmark and Share
Search
Skip Navigation Links.

Share

STATISTICA Blogs

Distance between Groups using Scatter with Error Plots

Sep 3

Written by: STATISTICA
9/3/2009 8:35 AM 

The distance between groups is often an important measurement, so it would be useful to be able to plot our data in such a way that we could visually see the distances between the groups. This example shows a method in which we can use Scatter with Error Plots to accomplish this and create the following graph:

Distance Between Groups 1

This example illustrates how to create a Scatter with Error Plot where each group is distinguished by a different color, and the plots will be spaced in a way that represents the physical distance of the measurements. A fictitious data set is used, which contains measurements taken along a river.
 1. Data: Bank is the side of the river on which measurements were taken (North or South), Location is the distance upstream, and Resp is the number of fish that are happy with their living environment.
 2. Select Scatter w/Error Plots from the Graphs > 2D Graphs submenu.
               Scatter_With_Error_Plots_Example2
 3. In the Scatterplot with error bars dialog, click the Variables button.
 4. In the Select Variables for Scatterplot dialog, select Location for X.
      Scatter_With_Error_Plots_Example
 5. Select Resp for Y. (Notice that we aren’t choosing Bank yet.)
 6. Click OK in the Select Variables for Scatterplot dialog.
 7. Now we need to address the fact that we want to have our data grouped by which side of the Bank the measurement was taken on. So, click the By Group button.
 8. In the By Group dialog, click the Grouping Variables button.
 9. In the Select By Variables dialog, select our grouping variable, Bank.
            Scatter_With_Error_Plots_Example
10. Click OK in the Select By Variables dialog.
11. Click OK in the By Group dialog.
12. And, finally, click OK in the Scatterplot with error bars dialog to create our plots.
13. So here we have the plots, but there are some things we want to change.
14. Notice that three plots were produced, one for all the data (All Groups), one for the North Bank, and one for the South Bank. We want to merge the North Bank plot onto our South Bank plot so that we have all of our data in one graph, but there still will be grouping based on Bank.
            Scatter_With_Error_Plots_Example
15. First, let’s clean up our plot some. Double-click on the background of our South Bank plot to display the Graph Options dialog. Now, navigate to the Plot > General node.
            Scatter_With_Error_Plots_Example
16. Make sure that Raw Data is selected for the Plot type option.
17. Clear the Markers check box. This will suppress the data points from being displayed on our plot. 
18. Now, we’ll change the plot line color so that when we merge the two plots we’ll know which set of data belongs to which Bank. So, navigate to the Plot > Spreads node.
19. Select the Mean plot type if it isn’t already selected.
           
20. Click the Line button to display the Line Properties dialog.
21. Click the drop-down menu for Foreground color.
22. Now select a color that you want. I hear that red is a warm color that makes people feel happy, so I’m going to use that.
23. Close the Line Properties dialog.
24. And click OK in the Graph Options dialog to return to our plot.
           Scatter_With_Error_Plots_Example
25. So now we have no points plotted, and our error bars are red. (I removed the points from the North Bank plot as well, but left the color for the error bars as blue.) Now we need to merge the two graphs.
26. From the Tools menu, select Merge Graph to display the Select Graph to Merge dialog.
            Scatter_With_Error_Plots_Example
27. Now we need to note what graph is currently active; in my case it is the South Bank graph.
           Scatter_With_Error_Plots_Example
28. That means we need to add the North Bank plot, so in the Select Graph to Merge dialog, select the North Bank plot.
29. Click OK.
            Scatter_With_Error_Plots_Example
30. And here it is. Overlaid error plots, with groups in colors, and correct spacing depending on the X axis. From here, we can change the format of the plot to fit our needs.

 

Trackback Print
Location: Blogs STATISTICA
Tags: