My suggestion would be fill the box with styrofoam with a small space in the middle for the resistor and a thermistor. (Check the insulation resistance of the material chosen of course.)
The idea is to make the thermal time constant from the outside world to the resistor as long as possible. Two advantages:
1. temperature stays steady during measurement
2. protects the resistor from temperature extremes during transport or at least makes the rate of change less violent.
Hi, in this case it's much better to have a big thermal mass (aluminium or oil filled cavity) to stabilize / slow down temperature changes.
Isolation with styrofoam is usually done on self heated applications, like in OCXOs.
The thermometer and the resistor have to be tightly coupled, e.g. assemble them in the same aluminum block.
If you ship such a standard ( Ohm, Volt) over several days in cold weather, a styrofoam isolation does not really help, it just looks as if it would protect from excessive cooling down, but in reality it doesn't. We already made our experiences with "cold" shipped LTZ1000 inside a big isolation package, and they simply shifted anyhow.
Only way around is to monitor the shipment temperature inside the package to find out if the temperature fell below a certain critical threshold (+15°C in this case), or to ship the standard "hot", or to invent a rejuvenation process, to mitigate possible hysteresis shifts (similar to the Fluke 7000 voltage standard).
The bulk metal foil technology also shows hysteresis on big temperature changes. Don't know how much this is on the VHP101 chip, but the Z foil technolgy (VHP202Z) which I use, showed +5ppm on a -40°C trip. I could remove that hysterisis by thermal cycling.
Precision wire wound is much better in this aspect.
Frank