This post on A Few Things Ill Considered is a great demonstration of how graphs and statistics can be abused. Show CO2 and temperature data from a short stretch of time, with misleading axes, and you can produce the impression that they are unrelated. Present the information in a fair way, and the correlation between the two looks very plausible.
Combine that with the theoretical framework about greenhouse gases trapping energy (in the form of longwave radiation) within the planet system, and you have a hypothesis that is defensible on both theoretical and empirical grounds.
