An outlier is an data point that is distant from other observations, it does not fit into the trend shown by the other results. Outliers that are due to variability in the measurement or an error in the experimental procedures should discarded when interpreting results. Repeating the measurement is a useful way to test for outliers that should be discarded.
Outliers that are consistently repeated should not be discarded from results as they show the presence of uncontrolled variables or a flaw in the theory used for the hypothesis.
For example, the number of weeds growing in a patch of grass may decrease with distance from a water source. However, nutrient levels will also have an effect on plant growth and some patches of grass may lie in soil with increased nutrient levels due to decaying organic matter.