If the advertised accuracy was +/- 0.2 degrees F, it should display a temperature of +/- 0.2 from the true value. Resolution would be the smallest division displayed (I would assume, with no additional information, 0.1 degree F. There are devices that will measure 0.01 degree F, but there are so many fators other than the instrument itself, that a lab environment becomes a necessity).
When you say this is "off" by 1.5 degrees F, what standard are you using for comparison? From the range, this sounds like it is a "medical thermometer" used for measuring body temperature - if those are inaccurate by 1.5 degrees F (from the true temperature), they are useless.
You speak about a "dial thermometer" - I would assume you mean one of those made by Weston (I have one with a Cibachrome logo) or others - a "dial on a stick" Those are easily calibrated by adjusting the nut holding the stem to the dial. I've calibrated mine aganst a Kodak Laboratory (mercury) Thermometer and it is reliable to one half degree F - or so (never trust the reading of any measuring device to a value less than the smallest division - the resolution).
BTW - What would you cook between 93 and 107 degrees F?