Pew has 34.9D-31.6R (
here). You can go to the
General Social Survey and obtain 34.0D-29.8R in 2004 (enter YEAR for row, PARTYID for column, and check row percentages). As Pew points out, people often aren't consistent in their responses from survey to survey. The question wording will influence how many people end up as independents.
Here Pew has a perspective over more time:
http://people-press.org/commentary/display.php3?AnalysisID=95No idea where 40D-40R would come from, except perhaps as a vague indication that presidential races are likely to swing between 60-40 and 40-60.
If you play with the GSS link (try PARTYID as a row variable and PRES00 / PRES96 / PRES92 / PRES88... as column variables, or something like that), you can see that a lot of Dems, especially "not strong Dems," apparently do vote for Republican presidents, depending on the year. (As I've pointed out before, some people misreport their past votes, so don't take these results too literally. Heck, don't take anything survey-based too literally.)