Just to throw my own experience into the mix. I started out with an interest in colour management by purchasing the Spyder2 Pro kit a few years back.
I was happy enough initially, but then I started noticing that there were some slight colour tints (particularly green in the shadows) and sometimes posterisation was quite pronounced.
I tried out
ColorEyes Display Pro after there was an article about it on this site and got much better results using it.
At this point, I had been doing more reading into calibration, finding out what people thought of the various sensors etc. and it was generally considered that the DTP-94 was the best affordable colorimeter on the market at the time, so I went with the ColorEyes bundle, rather than just getting a software license.
This gave me the benefits of smoother gradations with ColorEyes but also a better calibration, particularly in shadow areas (no more green tint) and I sold the Spyder package. (for what it's worth, I sent my sensor into them for replacement, got a new one and had similar issues) So far, I have been really happy with ColorEyes' results.
A few more years down the line and I got interested in home theatre calibration, so televisions and projectors. These don't support ICC profiles like PCs, so you have to make manual adjustments. (I was using the free
HCFR software) Long story short, I wasn't entirely satisfied with the results I was getting on some displays using the DTP-94. With some, greyscale would match perfectly and on others it wouldn't. With no reference for comparison, I really had no idea which was correct. Colour adjustments were somewhat problematic as well.
Well, I did more research on display calibration and the hardware, and concluded that I needed a spectrophotometer. (i1Pro) Basically, the accuracy of a colorimeter is based on how close the primaries of a display are to the response of the colour filters used. So if a colorimeter was designed around displays with an sRGB response for example, it would be less accurate with wider (or narrower) gamut displays. Which means that you really have no idea of how accurate the readings you're getting on your display are. They might be perfect, or they might not.
A spectrophotometer on the other hand, works completely differently and its accuracy isn't really based on the gamut of the display. (well, you might have issues with laser displays due to their spectral characteristics) They have their own issues—they're much slower than colorimeters and aren't as good at low light levels, for example, but should generally be a lot more consistent and accurate.
Initially, it was an i1Pro I had wanted anyway, but it was well out of my budget at the time. It was still very expensive here in the UK, but as I was also looking to upgrade my HT calibration software (in the same way I went from the Spyder 2 software to using ColorEyes Display Pro for more accuracy) I was able to buy an i1Pro from
SpectraCal for about half the cost of the cheapest X-Rite i1 package over here. (though with the extra software and other hardware I bought I think the whole order was about the same cost)
Now, they supply an OEM sensor, which means you just get it and the accessories in a white box without any X-Rite software licenses, so if you want to use the X-Rite software, you have to purchase the individual modules from them (currently it seems to be $99 for a monitor license, $199 for RGB printers, $99 for scanners, $199 for camera profiling) but as I already had ColorEyes Display Pro which also supports the i1Pro, I didn't need to buy any. (I don't do printing/scanning at home or have a need for camera ICC profiles)
Unfortunately they no longer seem to sell the i1Pro on its own (it was around $600 if I remember correctly) so you would also have to purchase their CalMAN software with it which may or may not be of any interest/use (though it couldn't hurt to ask if they'd sell the sensor on its own) but depending on where you live and the price of the X-Rite packages there, along with exchange rates/import duties, it might still be a lot cheaper. Remember of course that you will have to buy additional software for creating profiles, but you may have been planning on doing that anyway. I know if I had bought the X-Rite package I would still want ColorEyes on top of that for my computer monitors.
But anyway, now that I have the i1Pro I'm getting consistent results across all displays I've used it with. That covers CRT, LCD, wide gamut LCD, narrower gamut LCDs (laptop displays) LED Backlit LCD (again, laptops) Plasma, DLP Projection, SXRD Projection. Once they're calibrated to D65, the greyscale is a visual match on them all, whereas it differed when using colorimeters. (I bought a Chroma 5 along with the i1Pro as part of the CalMAN kit, and while it's better than the DTP-94 or Spyder were, it's still somewhat inconsistent) It's not just greyscale that's more accurate though, colour gamut is often measured wrongly with the colorimeters as well.
So if you can afford it at all, I would strongly recommend purchasing an i1Pro package of some kind. I wish I had just bought it right away, rather than incremental upgrades leading up to it, which has cost a fair bit over the years.
The other thing is that with the Spyders, it's my understanding that their factory tolerances/calibrations aren't so strict. I'm not sure how much it applies to the Spyder 3, but the Spyder 2 was basically a lottery as to whether or not you'd get a good one, unless you had a Spyder 2 Platinum sensor. (which was a hand-picked one that you could only get with the purchase of ColorFacts)
SpectraCal are soon going to be selling hand-calibrated Spyder 3s, which suggests to me that it's going to be as much luck getting a good one as it was with the Spyder 2s if you buy a datacolor package.
They place the calibrated Spyder 3 above the i1D2 but below the X-Rite Chroma 5. Well, I own a Chroma 5 (I've had two actually) and as I said, I found it to be better than the DTP-94, but still not nearly as consistent over a variety of displays as the i1Pro.
For greyscale (which is a major part of calibration) I'd say a good colorimeter is maybe 80% of the way there, the i1Pro is 95% there, and then high-end meters (which are orders of magnitude more expensive) get that last 5% of accuracy.
With colour, a colorimeter may as well be guessing. It should still be an improvement over not being profiled at all, but if you want accuracy, you really need a spectro in my opinion. (note: specially calibrated colorimeters that are only bundled with specific displays are an exception to this, they should be very good)
It's also worth mentioning a few other things about the i1Pro. There are quite a few revisions, I think A to D. The only difference between these is the reading speed, they're just as accurate as each other. If you're buying a new meter, you should get the latest anyway. Even it can be quite slow at times though, so I'd hate to think how the others are.
It's not really worth buying one second-hand though (if you had considered it) as they are only certified for a year. (though if they are stored well, I believe they should be good for two or three)
If your primary concern is display calibration or measuring the spectral response of lamps/other surfaces (I used mine to select the most neutral screen for my projector I could find, for example—which was
Carada's
brilliant white, if that's of interest to anyone) you want an i1Pro
without the UV-cut filter.
If you are profiling papers that have optical brighteners, you want an i1Pro that does have the UV-cut filter as it should be more accurate for that. (and if you're doing both displays and prints, well then I'm not sure which you want…)