Since you are recording in December, I'm going to suggest that there is a substantial thermal difference between inside and outside. Here I'm assuming you are some place where it gets reasonably chilly in December. If that is the case, we may note that warmer air expands -- fewer molecules / m**3 -- and colder air contracts -- more molecules / m**3. We should expect the infrared transmittance of the air to go up when it is warmer on this basis, because there are fewer molecules to absorb the signal. So, we may indeed see changes in the detector based on temperature with constant ppm of CO2, beyond changes that are simply based on the temperature of the detector itself.
Boyle: amount of substance / Volume = Pressure / (R * Temperature)
As temperature goes up, amount of gas in a fixed volume goes down. Temp is in Kelvin here, so we'd need maybe 30°C to make a 10% difference. The relative concentration of CO2 in the gas should remain constant, so the "apparent concentration" of CO2 in the air would appear to go down a little as the air heats up based on a spectrophotometer like this device for a naive calculation that didn't take this into account.
Now, I'd like to think that a device with a humidity and temperature sensor in it have those things so it can correct for such factors! But, as it claims "Best price / performance", perhaps there wasn't enough engineering $$ spent on that calibration, or this one is defective, or there is a lot of variance between devices, making that sort of calibration difficult without individually calibrating each one which might be too expensive. It seems deeply suspicious to me that the device measures lower concentrations as time progresses over a short period. (Because the IR lamp heats up the detector or the gas? Because the detector has some issue? Because the IR source gets brighter as it warms?) But, I'm not a analytical chemist, so listen to anyone with actual experience with NDIR photometry over me.
There is probably someone in the engineering department of the company that can help you out, but finding that person is going to take some doing. I might see if the detector reaches some steady state if left on for a few hours, and if so, try those measurements instead. And of course, a real scientist might prepare calibrated samples in a closed gas bag and recalibrate the thing himself and see whether he could experimentally account for whatever nonsense the device is putting out and correct for it.
The only other speculation that came to me is that air conditioning might remove some CO2 from the air as it condensed out some of the water from the air. The air recirculates. Maybe it could do it if your home was particularly air tight. <Vigorous hand waving> Of course, in December, this seems unlikely unless you are enjoying summer at this time of year. However, I am really just rampantly speculating at this point and should quit before I get even further behind.
It is certainly observed that bubbling gas through water systems (e.g. ozone treatment of water) pulls dissolved CO2 out of the water driving away some of the hardness. This is something of the opposite of the process:
CO3(2-) + 2 H+ <==> HCO3- + H+ <==> H2CO3 <==> CO2 + H20 (and the CO2 escapes with the gas)
Normally, all these variants exist in solution in some concentration in equilibrium, based on pH. However, when bubbling gas through the system, it pulls out CO2, thereby driving the reaction to the right. This consumes both CO2 and H+ ions. This has the net effect of removing some of the acidity from the water, and because carbonates are often less soluble than bicarbonates, you might observe CaCO2 coming out of solution and coating your pipes.
So, we might imagine that a steady process of condensing water out of the air would remove the CO2 with it. CO2 is far more soluble in water than oxygen or nitrogen. There is also a lot less of it in the air, so if we steadily draw away small amounts of CO2, O2 and N2 with water condensing out of air, the CO2 will be rapidly depleted and the O2+N2 won't change much. Eventually, you might have a fairly pure N2, O2 mixture. The problem is that it would have us believe that the CO2 is not leaking into your house at a comparatively tremendous rate to replace it. Since generally people don't asphyxiate in their homes from CO2 buildup even if they stay inside for days -- you may have recently tried this experiment yourself in the last year -- it seems quite reasonable to expect that the house is in fact not that air tight, and this mechanism really couldn't account for your observations. This former scientist would prefer to go with the explanation that CO2 is in steady state exchange with the outside air. The CO2 ppm inside and outside is going to be very similar and you are just observing miscalibration or other confusion in the device.
...but you know, I haven't actually measured, so confidence is low.
