I went to the local universities' chemistry supply store, where they sell goggles, lab coats, etc. and picked up 3 thermometers for under $15; all of them read identically and are alcohol based, not mercury. Like Mark says, accuracy to a 1-2 degrees is less important than consistency as it needs to read the same, again and again.
You assume the newly purchased thermometer is correct because it cost a lot but your negatives are consistently thin. Now, it could be your light meter is off but you don't think so. Add time to the developing cycle. If your highlights in the negatives are dense, reduce time from the developing cycle. After X-amount of time, you discover the thermometer is incorrect by 2 degrees but your amended developing times are producing good negatives. DON'T CHANGE ANYTHING! You have calibrated your development times to your personal process and it works. As long as you are getting good negatives on a consistent basis, it does not matter what the thermometer reads. It is important to note that the development times at a particular temperature published by manufacturers are always recommended time, not hard rules. My development process with your negatives will probably not work and vice versa.