Thank you for the response, jremington. Yes, I calibrated the thermistor using the test rig I mentioned, reading 3 temperatures and their resistances and inputting them in this online thermistor calculator (I'll post the link later because I'm on my phone right now) and getting the coefficients.
But the Steinhart-Hart equation is not the problem, the temperatures are matching perfectly the resistance read by Arduino (I know this because the calculator also shows that). The problem is that the resistance is wrong to begin with, compared to the multimeter readings.
To calculate the resistance, the code reads the voltage on the analog pin, so there must be something wrong with this reading (that is if the code is fine, which I think it is).