Here's a fascinating analysis of the accepted scales used to measure scientific knowledge in the public, one that argues that at least a couple of the questions used in these national surveys are actually measuring a dimension of religious beliefs than scientific knowledge (a full pdf may be available to you here).
In other words, asking about evolution and the big bang (the theory, not the very funny television program) may not measure scientific knowledge so much as it measures something called the "Young Earth Worldview," which is the notion the Earth, despite all evidence to the contrary, is only about 8,000 or so years old (for fun, I offer a bit of Inherit the Wind below).
A lot of this paper gets into factor analyses of available survey items to explore whether they measure a single dimension (scientific knowledge) or multiple dimensions. Unless you're a numbers geek like me, it's probably best you avoid the Results section. From a methodological perspective in using these questions, scholars "should take special care to account for the religious loading of the “evolved” and “bigbang” items (and to a lesser extent, “condrift”)." In other words, secondary analysts of GSS data, beware. Make sure you're measuring what you think you're measuring, and don't let religious confusion get in the way of a good, solid index.