If you have outliers far to the right of a batch of values, then those outliers will pull on the right tail of the distribution to make it right skewed (aka positively skewed). These outliers will increase the arithmetic mean to make it larger than it should be.
For example, let's say we had the set of values {1,2,3,4} all of which are clustered together fairly close. The mean is 2.5. Then lets add in the number 55 to get {1,2,3,4,55}. The 55 will increase the mean to 13 which is nowhere near any of the original {1,2,3,4} values.
