Exercise 14: Box plots in Prism

Go to parent GraphPad Prism statistical analyses

For this exercise we will use data of the babies data set. We are going to create a boxplot of babies’ weight for each category of smoking behavior of the mother:

• 0 = never (never smoked)
• 1 = yes (still smoking)
• 2 = recently (quit during pregnancy)
• 3 = long ago (quit before pregnancy)
• 9 = unknown

So we need to create five boxplots in total.

This means we only need the wt (weight of the baby) and smoke column, but we need to get the weights for different smoking categories in separate columns. I will explain it for the first smoking category: 0=never.

Get the weights for smoking behaviour 0=never in a separate table.

To do this you need to prune the rows. Pruning is done based on X-values. So first you need to paste the columns wt and smoke to an XY-table where smoke is the X column (smoke contains the values you want to base the pruning on): Create a new XY table Copy the smoke column Paste Link the smoke values in the X column of the new XY table Copy the wt column Paste Link the wt values in the Y column of the new XY table Give the new table a meaningful name Now you can prune the XY table: Click the Analyze button in the Analyze section of the upper toolbar Select Prune rows in the Transform, Normalize... section Click OK Now keep only the values for which X equals 0 (the mother never smoked): Click OK. This generates a transformed data table containing the weights of the babies from mothers that never smoked: I have changed the title of the wt column to never since this is tha label I want to use in the box plot.

Go back to the XY table containing smoke and wt values. Repeat the pruning for smoke=1, smoke=2, smoke=3 and smoke=9.

Once you have the pruned wt columns for each smoking type, you can start thinking about the box plots. The easiest is to create one table containing 5 columns, each column containing weights of babies for one specific type of smoking behaviour.

Create a new column table containing these 5 columns.
Box plots are linked to column tables: Click the New data table and graph button in the top toolbar. Select to create a new Column table and give it a meaningfull name. Paste each wt column as a column in the new table. Adjust the names of the columns: use labels you want to use in the graph

Create a box plot for this new column table.
Select the corresponding graph of the table. Select a Box and whiskers plot. Select Tukey numbers to plot. This generates a box plot of the data.

Just for the sake of the next exercise (Exercise 15C), we'll also create an individual box plot for smoke=0 and smoke=1.

Make a boxplot for smoke=0.

Click the graph of the table containing the pruned data for smoke=0 in the navigator. This is a XY table so te default plot is a scatter plot not a box plot. Click the Choose a different type of graph button in the Change section in the upper toolbar Set Column as graph type. Here you clearly see that Prism allows you to make plots on different table types than they are intended for. Select a Box and whisker plot. Select Tukey numbers to plot. This generates a box plot of the data.

Do the same for smoke=1.