I calibrate equipment for my day job, & I have seen some 30 year old name brand dial indicators that still work great. The ones I've had the most good readings from are Mitutoyo, Starrett, Brown & Sharpe. Most times when they don't move freely it's because they are dirty or the shaft is bent. If it's just dirt you can clean them.
The import ones from china, & India don't seem to last very long. I don't think the materials they are made from are the same, they wear out faster & the movement gets sloppy. The the harbor freight one I had was fine when it was new, but 6 months later with infrequent use, it was off by a whole division when changing directions (Backlash). The e-bay brown & sharpe one I purchased was fine even though the lens wasn't very clear anymore.
I say if you can't afford a brand new indicator look at buying a used name brand indicator over a cheap made in China no name one. If you get a dud try again.
For the best accuracy, & repeatability you have to look at the needle dead on straight at the center of the dial each time. Looking at the indicator from above or the side will give you a different indication, this is called parallax error. The same reason your wife thinks your speeding when she looks over at the speedometer from the passenger seat, & says your going 80 when it only says 70 to you.
You can check your own dial indicator with a good solid stand & mount a surface plate & some gage blocks. Make sure everything is clean & mount the dial indicator to your stand. Make sure that the indicator movement is perpendicular to the surface plate. Move the indicator down until the plunger touches the surface plate & the needle moves at least a few divisions to preload it. Set your zero, then pull up the plunger & set it back down on the surface plate a few times. It should return to zero each time. This tests your repeatability. If it didn't return to zero each time check the rigidity of your setup & try again. Make sure that the plunger doesn't 'hang up' or stick if you clamp on to the outer shaft. It's best to mount it from the back if that is an option. Once you have a repeatable zero indication you can continue. Select a gage block or stack that will put the indication at 3 o clock on the dial. Slide the gage block under the tip (every thing is clean& you should use a round tipto prevent damage. Lift the plunger only enough to clear the gage block(s). It should indicate correctly. Check it again at 6, 9, 12 o clock positions. This checks the dial linearity, some times the indicator will read a little off at some points in it's rotation. Then if the dial goes around more than once check it at at least 25%, 50%, 75%, & near full scale. This checks the linearity of the full travel as the needle goes around & round. Then remove the stack & go back down & check it again at least one other point that you already checked. This time pull the plunger the whole way up & (slowly) lower it. The indicator should give you the same reading you had before when you only lifted the plunger just enough to clear the gage block. This checks the backlash.
Alternatively you could check the indicator against a micrometer if you have one large enough to fit & a way to mount everything in perfectly inline. This method would also work for last word type test indicators set the micrometer to zero & plunger on it & zero the indicator. Rotate the micrometer up & compare the indication to your dial indicator as above at a few points. Move it up & back down & check it again at the same micrometer indication, the needle should return to the same position. If not your dial indicator probably has some backlash. It could be your setup isn't correct (loose, out of square with each other, dirty, etc... but micrometers are usually pretty linear (even the cheaper ones).
Either way you will see the error in your dial indicator, provided your setup was correct. Clean, perpendicular to the measurement surface, & parallel to the micrometer spindle, & mounted securely. If you see that your indicator has excessive backlash then you can still use it in the same direction of travel to minimize the backlash from affecting your readings or if it's not linear you can still use it for minimal deflection checks. If you take care of that tool it could still be useful even if it isn't "accurate".
When I calibrate dial indicators at work I use an Optimar 100 made by Mahr. It is amazing, & shows a zig zag indication on a computer screen so you can see the variation at each test point. I've learned though experience that even the best most accurate dial indicators will still vary a bit at each test point (usually less than half the resolution), but it's more often the fault of the operator reading the indication than it is the equipment.
Maybe one day if there is interest & I have time & motivation, I will post a series of how to calibrate your own micrometer, caliper, dial indicator, level, etc. etc. etc.
How ever like someone else pointed put the resolution is not the accuracy. A lot of times the accuracy if it's even publicly available is worse than the resolution. On top of that the accuracy goes down after the first two & a half revolutions of the dial. Not 2½ from the beginning of the travel, but 2½ from where you set your reference. I believe that this is an international or DIN standard. Also best practice is don't try to take a good measurement in the first 1/4 revolution or last 1/4 revolution of the needle. Sometimes at the ends of travel the indicator results are a little wonky. It's way less, only a couple of divisions with the last test type dial test indicators that only have 1 revolution to begin with.
Lastly as someone else pointed out a metric indicator used for minimal deflection of the needle is a good idea. I have NEVER in 10 years calibrated a metric indicator that did not meet factory specs. They just don't get used much compared to the the inch standard ones, at least here in the states.