Five of the six Republican presidential wins between 1954 and 1994 were landslides in the Electoral College and healthy wins in the popular vote. Yet in that period, the party lost 20 consecutive House elections to the Democrats and won a Senate majority only three times.
FPTP doesn't seem to be a factor given FPTP is used for presidential races too using the same boundaries as the Senate, and inequitable House boundaries/gerrymandering doesn't seem to account for the massive differences in vote-shares.
Sometimes the gap could be huge; Nixon beat McGovern 61% - 38% in 1972, but lost the House race 47% - 52% to the Democrats, polling 14 points above his party. Likewise, Reagan's 59% in 1984 translated into just 47% in the House and 49% in the Senate.
How have political historians and historically-minded political scientists interpreted this phenomenon of sometimes enormously divergent voting patterns, either in general or in specific election years?
Quite often, an legislative election is decided mainly by which candidate is the incumbent. Look at this map of the 1984 U.S. House Elections. Notice that most of the seats stayed the same party because the incumbent was re-elected. A similar pattern can be observed in other House elections of the period, with large mass party changes of House seat makeup, like in 1992, being the exception, rather than the rule.
Generally speaking, over the past 50 years, house seats are won by the incumbent (prior office-holder) more than 80% of the time. As the charts on that same page show, the re-election rate for the senate is only slightly worse, and in 1980 the Republicans were able to turn over a substantial proportion of the Senate seats up for election that year.
In terms of academic study of the topic, it has been observed since the late 1960s that incumbancy re-election rates appear to be going up over time, although exactly why this so is not entirely agreed upon. In 2002, Ansolabehere & Snyder from MIT published "The incumbency advantage in US elections: An analysis of state and federal offices, 1942-2000", which examines in more detail some of the trends I've described above.
