P/E Ratios. - the spread and outliers

Spread is generally the difference between two values. In this context it is the range of values, or the difference between the largest and smallest value. So if you have three companies in your dataset with P/E ratios of 3, 5.5, and 4, then the spread is 2.5 (5.5 - 3).

An "outlier" in statistics is generally some observation that is an abnormal difference from most other values. So if you had P/E ratios of 3, 4, 5.5, and 10, 10 would probably be considered an "outlier". Sometimes in statistical analysis outliers are excluded from analysis if their value skews certain measurements (like range) in an undesired fashion. One has to be careful when excluding outliers without understanding why the value is an outlier. Is it a measurement error? Is there some extraordinary circumstance that explains the value?

And most importantly what does the term "outliers" mean over here?

Get into a spreadsheet a list (from, say, the VTI ETF) of all companies and their P/E ratios. Sort the list by the P/E ratio and then plot the numbers by frequency (how many stocks have a P/E of 10.1, 10.2, 10.3, etc, etc).

You'll see most bunched in a broad "center", while there are some (with very low or very high P/E ratios) which lie very far out from the center.

As you can guess... those are the outliers.

Eliminating outliers is important when computing an average. The outliers lie really far from the center, and the number of elements in the list is relatively small.

Because of this, the "median" is usually a better statistical measure, since outliers which affect the mean (fancy word for "average") minimally affect the median". (This is also why you hear reference to things like "median family income" instead of "average family income": all those multi-multi-millionaires move the "average income" much more than it moves the "median income".)