OK, you sucked me in.
The problem here is that most of the probes used in these digital thermos are thermistors. Thermistors return to a digital processor a resistance. The computer converts that resistance to temperature. The problem is, the resistance to temperature ratio is not linear, it's curved. Meaning, if the resistance changes by a factor of 2, it doesn't mean that the temperature moves by a factor of 2.
This means that a probe can be accurate on one end of the spectrum and be off on the other side. So I recomend testing on both sides. It's also important to note that when you are taking a measurement, thermistors are slow to respond, sometimes up to 60 seconds. So make sure you give them time to idle to ensure that the proper temp is reached. Also, it takes longer for a probe to read a temperature change from high to low.
When I use thermistors in my program, I disregard the formula to calculate the temperature curve. I create a cross reference table from 40 degrees to 400 degrees. I measure the resistance every 5 degrees and store that value in a database. This is a calibration process I perform and store for every thermistor probe I use.
Geek out.....