Raw data should be checked once it has been entered into the computer and before any detailed analysis is done. A good way to do this is to use graphical methods which show up individual observations. Three examples are given below:
The data in the first figure show the red blood cell counts of C57BL/6 mice given various doses of chloramphenicol succinate. Such a plot provides a feel for what is happening, and any outliers can easily be identified. The two mice with very high counts at the 500 and 2500 dose levels should be checked against original records. This was done, and there was no evidence that they were in error, so they were retained. Note that these plots are only for data screening, so units and means are not shown.
A “Dotplot” from the same set of data shows the haemoglobin levels by dose, again showing individual animals. The two outliers are the same mice as in the first figure, so clearly it is the mice that are unusual, not the individual data points.
The third plot, below, shows a the red blood cell counts plotted against hematocrit. There is a strong correlations, with one mouse a clear outlier. This is the same mouse giving a high value at the 500 mg/kg dose level.