The only conclusion I can draw from these data is that the iPhone 4 3G cell signal reception is really inconsistent. (At least that’s the case where I live.) A few outliers in these datasets appear to throw thing off significantly - check out that 0.3 Mbps download with the full-on death grip. I’d discard the outliers, but the thing is, in every round of tests I get crazy outliers. This, I think, shows the inconsistent results (both in usage and measuring) in real-life scenarios.
All the tests were done using the FCC Mobile Broadband Test app. I sat in the same spot in downtown Portland, away from any obvious sources of interference. A larger dataset would likely flatten out the deviations. This is all I had time for.