Data Analysis – Linear Regression: Graphically examine the data

	par(mfrow=c(3,2))
	plot.design(pat.df)
	plot.design(pat.df,fun=median)
	plot(pat_dat~Measure+Patient, data=pat.df)
	interaction.plot(pat.df$Measure,
+	pat.df$Patient, pat.df$pat_dat)
	interaction.plot(pat.df$Patient,
+	pat.df$Measure, pat.df$pat_dat)
Six plots of the data
Six plots of the data

Transcript

Now, we look at our data analysis - first, we say, "Well first we're going to graphically examine the data". We want to know if the data itself has any problems.  So the first thing we do is look at the mean and the median of both the patients and the measurements separately, and then we plot the data - the patient data dependent upon the measurements. And then, we look at an interaction plot, which is a plot of the measure and the patients, and then we look at it in reverse a plot of the patients and the measure.

And it looks like this. The first - the top two plots - give you the data mean and the data median for the measurements separately from the patients. And the second set of plots (in the middle) tell you the measurements and the patience and how the data is constructed - so you have box plots to see if there are many outliers or how the data falls. And then, the bottom two plots give you the interactions. And if there is an interaction between the measurements and the patients or between the patients and measurements, it will not - you will not be able to do a successful ANOVA.