Winsorizing (defined as revaluing all observations that exceed the 99th percentile or fall below the 1st percentile to equal the 99th and 1st percentiles, respectively) is commonly employed to prevent these observations from having undue influence on the results.