In the case of plug-ins, electricity from the grid is the energy *carrier* and not the source. Comparing different carriers (electricity, hydrogen, etc) and different sources (coal, renewable, etc) requires the use of a fuel cycle model, and the GREET model is the popular one right now. Straight conversions on basis of chemical energy or stored on-vehicle electrical capacity don't do the issue justice. If we want to be responsible about our oil dependence and chose fuel efficient vehicles, the 'absolute' model (GREET) should be considered. And it yields some interesting results - primarily that plug-ins are a great solution in the absence of a functioning hydrogen infrastructure.
(To preempt responses from the hydrogen aware, hydrogen is considered a *carrier* and not source by many because, while it does occur naturally, the vast majority of commercial hydrogen currently comes from electrolysis or as byproduct from chemical reactions (refineries, industrial, etc.); we don't mine for it directly. In any case, a fuel cycle model is the best attempt to normalize these different energy pathways for plug-ins.)
For a quick primer on PI-HEVs and the fuel economy issue, take a look at this presentation (slide 9, 10) by Mark Duvall at EPRI which nails the issue on the head. If conflicts between what I said and what this presentation says exist, trust the presentation.