Having got both my Vantage Pro 2 system and my standalone indoor temperature/humidity sensor set up and talking to my monitoring and control software, I thought that it might be interesting to look at the longer term correlation between them, and to relate it to whether the weather station is connected to a powered up computer. The attached files show the results of a 48 hour study.
The .jpg shows the 'experimental setup'. The weather monitor and the TH sensor are both attached to wood blocks on the side of a bookcase in my office. The block on which the weather monitor is mounted is about 1.5" high, so most of its body is in (relatively) free air. The TH sensor is held slightly away from its block by a plastic holder that cradles it, but leaves a reasonable amount of room for airflow round it. The software recorded the reported temperatures every 5 minutes over a 48-hour period during which neither device nor any of the weather monitor's connections were physically touched. The weather monitor contains batteries and is hooked up to its standard power supply, which is powered when the computer attached to it (via a USB Weatherlink) is running. That computer is on and running from the time I get up (between 6:30 and 7:30 am) until I shut it down for the night (between 10:30 and 11:00 pm). The TH Sensor is powered by a rechargeable internal battery and was not physically connected to anything during the time of the test - it 'reports' by means of an intermittent WiFi connection, and simply stores data if the computer it reports to is shut down. My thermostat is set back substantially at night, heat is generated by the computer during the day, and the last few nights have been unseasonably cold, so the recorded (by both devices) nighttime temperature drop is entirely consistent with reality.
The interesting findings in the context of this thread are shown in the two PDFs, which respectively show Excel graphs of the readings of both devices over time and the difference between them (Weather Station minus TH sensor) over time. It will be seen that (compared to the TH Sensor), the Weather Station reads a bit over a degree (F) high when the computer to which it is connected is powered on and a bit under 0.5 degree (F) low when the computer is powered off. The 'stepping ' of the TH sensor's line is because its software stores its readings in Degrees C, to one decimal place of accuracy.
I have not compared the readings of either device to a reference thermometer, so I don't know which is more accurate under standard conditions, but it looks as if the Weather Station's inside temperature reading increases by something in excess of 1.5 degrees F when its attached Weatherlink is connected to a powered USB port, which is really something that should have been predicted and corrected at the design stage, since it must be a function of sensor placement and component heat generation, and it is well outside the claimed error bounds of the device.