1.   Table V shows results of stepwise multiple linear regression analysis of outcome variables from the ventilated infants.

2.   Graphics are generally more readable when explanatory variables are shown along a horizontal axis and outcome variables vertically.

3.   Showing percents requires no more than dots with lines for CIs of a binary outcome variable.

4.   Only when the outcome variable is ordered do cumulative, or stacked, displays make sense.

5.   Boxplots, showing median, interquartile range, and extreme values, are useful for quantitative outcome variables, especially when broken down by explanatory strata and displayed together.