My advice is having (two) on hand. I tested 10 digital units-in a store to find that out of 10- 3 had same relative measurements. 2 or 3 had 2-degree+ differences and I tossed those out. Calibration obviously was not done with care at the factory, but could have also just been 'not-accurate' brand/units themselves.
The lesson learned, I have 2-thermometers that I trust now. One is a turkey fryer style and the other a digital probe. They are both very close to each other, and I have calibrated both to ice-water.
Regardless of if these are true-accuracy, I doubt is a huge concern for most brewers- in that if I consistently produce beer based on what they read, and happy with the results, I can replicate those with the same equipment in the future.