If only to contextualize the Harris Poll I mention below, it is worth pointing out that science literacy in this country is fairly appaling. Witness the bi-annual NSF Science & Engineering Indicators (2004), which found that forty percent believe that astrology is either "very" or "sort of" scientific. This drops from 49% to 25% as education rises. It clearly should raise alarm bells if a quarter of college graduates feel that astrology can be even "sort of" scientific. Read the whole report ... it should depress you even if you are an ID supporter.
The demographic for this survey was 7% with less than high school, 53% high school graduate, and 40% college graduate - thus 60% are high school or less - in the Harris Poll this was ~74%, thus perhaps indicating that the NSF survey may in fact be overestimating literacy.
Update: This report indicates a high school drop-out rate of ~30% nationwide. Thus it would appear that both the Harris and NSF data seriously under-represents that group and thus, I imagine, seriously under-represents the problem of scientific illiteracy.
I can't help but wonder: if science education were better (including more engaging), would fewer kids drop out? (Most of the folks I know who dropped out before high school graduation had given up on science, too.)
I'm not sure whether improving science education would increase the graduation rate; it strikes me that there are many other (perhaps more important) factors. However, I will say that science education in general does obviously need to be improved - my students have told me some horror stories over the years!
I should also add, b.t.w., that science education at the university level could probably do with an overhaul as well. Just my 10c :)