Graphing outliers in column plots

Outliers can be a big problem in experimental biology.  Prism provides a method for identifying outliers in column data (new to Prism 6) and a method to identify outliers when fitting curves with nonlinear regression (new in Prism 5). When you use outlier elimination with nonlinear regression, Prism creates a separate data set with the outliers and superimposes these, in red, on your plot so they are clearly visible. But when you identify outliers in a stack of column data, Prism does not automatically identify those outliers on the graph. There are three approaches you can use:

• Let outliers be outliers. They are obvious on the column scattergraph and need no special identification. The problem with this approach is that if you choose to ask Prism to plot the mean or median, it will compute the mean or median from all the values, not just the ones that passed the outlier test. 
• Individually highlight outliers. Right click on outlier in a column scatter graph, choose Format This Point, and choose a color (or symbol shape) for it. Again, this won't affect the placement of the line at the mean or median, which will be computed from all the data including the outliers.
• Do some copy/pasting to create a grouped table, and create a superimposed dot plot that shows the outliers. Details below.

1. Create a grouped table, formatted with enough subcolumns to hold all your data. On the column table, each group is a column and the data are stacked. On the grouped table, each row is a group, and the data are side-by-side.
2. Copy the data, and use Edit..Paste Transpose. Then move the outliers to data set B. The idea is to make a data table like this one:
   
   
3. Create a superimposed scatter graph.
   
4. Double-click on the graph to bring up the Format Graph dialog.
5. For the data set with the good data, choose the symbol and color, and choose to plot a line at the mean (or median). Or choose to plot the mean or median with error bars. These will be computed only from the first data set. 
6. For the data set with the outliers, choose a symbol and color, and choose to plot no line or error bar.
   
7. Polish the graph.
   

If your situation is more complicated, note that you can choose to superimpose one data set over the previous one (but not superimpose others) at the bottom of the middle tab of Format Graph for Grouped graphs. 

Prism file for this example.