"The BMW X5, Volvo XC60 and Mitsubishi Outlander emitted 28-89% more CO2 than advertised when tested by Emissions Analytics on a fully charged battery in optimal conditions. On an empty battery, they emitted three to eight times more than official values. When driven in battery-charging mode, which could become more common as motorists charge up ahead of using electric mode in low-emissions zones, the PHEVs emitted three to 12 times more."
1) when empty, they emit more CO2 than their average assumed values. this is obvious. its just using gas, no battery.
2) when using the gas engine to charge the battery (i.e. instead of just generate enough electricity to move the car as desired), of course it is the worst performing. you're running the generator at full blast.
I had a Chevy Volt in palo alto where I was able to charge every night. I used it for commuting around (including a regular weekend trip to SF and other trips as far north as berkeley). As I was able to charge regularly (either 220v or 110v), I filled up the gas tank once a year (and that's because the car forces you to not have gas over a year old on average). Now, I don't have the ability to calculate how much CO2 was produced because of my charging times, but in terms of actual CO2 produced by the car, I'd put my experience over any other traditional ICE car.
Similarly, drove the car cross country. While I wasn't able to charge regularly, the electric motor handled the ups and downs of the terrain much better than all the other ICE cars I passed with ease.