[ You are not allowed to view attachments ]
That image is from Stephen Burt's 2009 review of the VP2. This would be relative humidity as measured by the SHT11 (original OEM temp/humidity sensor package) compared against a Vaisala HMP45C capacitive humidity sensor (the dotted black line).
Keep in mind, that measurements on this graph start in January 2009 (red dots) at which point the VP2 is approximately 6 months old as the study commenced in July 2008. The May readings (in blue) are when it's nearly a year old.
Here are his comments on the graph:
5 minute sampled RH for both instruments for a winter and summer month (January 2009 data in red and May 2009 in blue): in all, 17 780 pairs of observations. The thick dashed line marks the trend line that would be expected with a ‘perfect’ agreement; points above the line indicate where the VP2 indicated higher than the Vaisala sensor, and vice versa below the line.
Look how often the sensor runs high in the middle of the range -- often by 5-10% or more. In fact, I don't see one instance of it running drier than the Vaisala until almost 80% humidity. He said over the whole range it was 2-4% higher on average, but that's balanced out by the dryness on the high end (above 80%).
Those results are why he rated the VP2's humidity measurement only as "adequate" (essentially a "C" letter grade) while giving the temperature portion an excellent rating (or "A"). The humidity sensor was acceptable, I suppose, in England, the testing location, where it's above 80% humidity a great deal of the time. Not so acceptable elsewhere where the middle ranges are more common.
In other words, the Sensirion humidity sensor has run wet in the middle for years at least within 6 months from installation. I'm sure Davis is fully aware of that review as it was widely publicized at the time.
It doesn't appear that the issue was really "fixed" with the 31 model. Perhaps out of the gate, the readings are better, but not for long.