Steve Sailer has been pointing out since at least the 2004 presidential election that the marriage gap is a bigger deal than the much more media salient gender gap. The GSS reveals that Steve's perspicacity has been descriptive extending much farther back than that, since at least 1968.
Parenthetically, this is often the case with Steve's insights. He's portrayed as an extremist by all the usual suspects even though his observations tend to be quite modest and parsimonious given the data--both quantitative and qualitative--he bases those observations on.
The following graph shows the marriage and gender gaps, in black and red respectively, by presidential election. The marriage gap values are computed by taking the Republican candidate's performance in a two-way race among unmarried voters and subtracting it from his performance among married voters. The gender gap values are computed by taking the Republican candidate's performance in a two-way race among female voters and subtracting it from his performance among male voters:
The GSS asks about each presidential election for no more than seven years--three or four survey iterations, as they're conducted every couple of years on average--after that election occurs. So the 1988 election between the elder Bush and Dukakis gets a question in 1989, 1990, 1991, and 1993. Since these responses are being cross-referenced with marital status at the time of survey participation, there's going to be a little noise in the marital status figures. Someone responding in 1993, who may have been single when he voted for Dukakis in 1988 but who married in the interim, will show up as a married who voted Democrat even though he was unmarried when he actually voted.
This will modestly understate the real size of the marriage gap in each election while leaving the gender gap untouched (excepting the handful of trannies who may be hiding in the data sets!). Consequently, the difference in magnitude between the marriage gap and the gender gap is modestly greater than what is depicted in the graph above.
We can thus confidently conclude that in every US presidential election going back to at least Humphrey and Nixon in 1968 the marriage gap has been wider than the gender gap has been.
GSS variables used: PRES68(1-2), PRES72(1-2), PRES76(1-2), PRES80(1-2), PRES84(1-2), PRES88(1-2), PRES92(1-2), PRES96(1-2), PRES00(1-2), PRES04(1-2), PRES08(1-2), PRES12(1-2), MARITAL(1)(3-5), SEX