How does excel calculate outliers in box plots?

[cfth]

Hi there,
I need to know what calculation excel uses to determine outliers when making a box plot.

The method I know is: an outlier is any value greater than Quartile 3 + (1.5 x IQR) and any value less than Quartile 1 - (1.5 x IQR).

However, this doesn't appear to be the method Excel uses when plotting data on a box plot, as there are points outliers on the graph which fall within this range.

Help!

[MrShorty]

After a day, I will venture a response. I will say that I am a bit handicapped in my ability to look at this because my older version of Excel does not support the built in box plot chart type and, therefore, I cannot see what points Excel is identifying as outliers.

I could not find anything direct from Microsoft documenting the outlier algorithm used by the chart. The few other sources claiming to know how Excel determines which points are outliers all pointed to the same 1.5*IQR that you identify. If you are comparing your implementation of the 1.5*IQR standard to the chart and getting something different, I am unable to see that from your sample file.

If MSFT is not going to tell us exactly how the chart identifies outliers, and I would be strongly disinclined to try to reverse engineer their algorithm, then I would be inclined to not even use the chart's algorithm -- especially if I needed to really know what algorithm is being used. That might mean that you cannot even use the built in box plot chart type, but might need to build the box plot as a standard stacked bar/column chart. Then you will be able to control the outlier algorithm.

Sorry that I don't have any better answers.