I tested both cameras just yesterday, courtesy of my local camera store. I've been waiting for some good review of D3200, but since none was forthcoming, finally, I did my own IQ evaluation and chose the D5100, primarily because of the better native IQ (in my view and for my needs) and the articulated screen. Similarly to you, I was using D5000 before as my small camera, and was not planning to upgrade it to D5100, but when I compared the D3200 with D5100, I changed my mind.
Physically, the D5100 sensor is even a tiny bit larger than the sensor in D3200 (1.5 crop instead of 1.6). Considering that D5100 and D7000 have about the same pixel density as D800, I felt that 16MP in DX format at this time is the most effective file size, and that the density of a 24MP sensor in DX body would create for me more problems than the 16MP. And since the D5100 is on its way out, it cost me about 10% less than the new D3200 - $540, plus tax. That's just slightly more than what I paid this week to Nikon for the calibration of my old Nikkor 105/2.8 lens. Now, I may just keep that lens permanently on the D5100 body.
I am aware that most reviewers (when comparing two different file sizes) upsample the smaller file to the size of the larger one, but in my real-world situation, that's seldom the case. Since I seldom print, but the files must pass the scrutiny of agency inspectors, more often I need to downsample those large files to increase the apparent sharpness. I had to employ this practice first, when shooting with the 18MP Canon T2i (same sensor as 60D or 7D), even when using the 50mm/1.4 lens. Quite often, the image was on the soft side, but reduction from 18MP to 12MP usually did the trick. I presume, I would have to do it also with the 24MP DX format.
But when we finally get the D600 with 24MP in FX format, that will be a different story.