I did a test on this a few months ago. Like some have said, all things aren't equal.
I live in a small town, where almost all of the buildings are 1-2 story. The tallest building in town is 10 stories. The test setup was an MFJ simplex repeater at home, with a Comet dual-band antenna in the attic. I did some on-the-fly tests while driving around town with an HT. I then went to a fixed location 3.1 miles (as the crow flies) distant.
Using the stock HT antenna and a Nagoya antenna that was just a bit longer, 2M was essentially unusable. Using the longer Nagoya antenna, 2M was at best 50% copy. With the two short antennas, 440 was almost 100% copy. With the longer antenna, it was much worse. These results were pretty much the same while in the car at various locations, as it was at the fixed location.
As far as the conditions of the test, I was standing behind a single story building, holding the HT in my hand. There was no line of sight. There were several single story buildings between me and the simplex repeater. The terrain was slightly uneven, with a small hill between.
Other factors influencing the test - the Comet antenna had twice the gain on 440 vs 2M. The shorter HT antennas work best on 440. Note that even with the 15" Nagoya, 2M was nowhere near perfect copy.
This wasn't a controlled test, where all variables were the same. However, it did seem to me that even in a small town with single story buildings, 440 worked better in a simplex situation at ground level. Now, if I was talking to a repeater on a mountain (we have a few) and had line of sight, the 2M would probably have been better due to path loss. Also, if I was in the woods, I would prefer 2M.