Wow, I see we have many Crapple fanboys here that they are also statisticians!
"Outliers can indicate a systematic error or erroneous measurement. Ignoring their impact can undermine the reliability of analysis based on flawed data. Outliers adversely affect the estimated predictive accuracy of the dataset."
This is exactly the case by using Geekbench. You introduce flawed measurements (biased data points) in your analysis, which is systematic. Since it always favors Crapple.
And it undermines the reliability of the analysis (and the analyst).
So yes, keep using Geekbench!