Some factors to consider:
Accuracy – Measurements are near what they're supposed to be. I don't know how to determine this aside from trying out several different devices and perhaps comparing with laboratory-grade spectrometers. Ultimately, you're just going to have to trust that the devices do what they're designed to do.
Reliability (Precision) – Measurements are reproducible.
Speed. Important if you plan to re-profile regularly. If the device is fast enough, you might re-profile just because you can, even if you didn't originally intend to. Based on reviews, newer devices seem to be faster than older ones. Colorimeters seem to be faster than spectrometers.
Price. Often, the exact same hardware are packaged with different software at different price points.
Software. What operating systems are supported? What features does the software have? If you plan to use DisplayCAL, the included software doesn't really matter.
Device compatibility. Some (old) devices are designed to work with CRT, but not LCD. Some work with printers. Etc.
Ambient temperature measurement. Improves color accuracy?
Filter material – Gelatin vs Glass. Glass lasts longer, but gels are reportedly more accurate? I don't really know.
Links of interest:
Nearly all of my output goes to web, where most people view images on uncalibrated monitors, so I just need my monitors to be in the same ballpark as each other. For the occasional print, it doesn't have to be exact, just close enough. Most of my monitors barely changed after calibration, so I just leave them alone (uncalibrated). However, one in particular changed significantly, so that one has to have a profile set. I re-profiled it a few times, and it seemed stable, so I don't bother as long as images look the same as the other monitors.
I have a Spyder2. If I were choosing another device, I'd consider spending a bit more for a faster device with ambient temperature measurements because it would be useful for setting camera white balance. I'd also consider a device that does not require firmware loading. Aside from that, it has served its purpose.
- Accuracy – Seems to do what it's supposed to do.
- Reliability – Re-profiling multiple times produces slightly different results. I use the profile that looks "best".
- Speed – S - L - O - W.
- Price. Cheapest device I could find. This is the main reason I chose this particular device. If it didn't work, I wouldn't lose much.
- Software. I use Linux, so the device had to be compatible with DisplayCAL. Spyders require firmware loading to work.
- Device compatibility. To profile LCDs, requires use of a special filter that's included with the device.
- Ambient temperature measurement – Absent.
- Filter material – Gelatin.