The economist has an excellent presentation (less than two minutes) on how assortative mating has influenced income inequality. This closing chart sums it up.
Fifty years ago there weren't many well educated women with high earning potential. In fact, the percentage of adults with college degrees in 1960 (overwhelmingly men) was about 7%. That means 93% had a high school education or less. Fifty years later the percentage of adults with college degrees is 25% and is split nearly equally among between the sexes. People of like education tend to marry each other, thus creating an explosion in the number of households with two high income earners. That stretches the household earning distribution greatly and moves the mean income out much farther to the right on a chart. Thus, even if low-education households have stayed the same in inflation adjusted income, their relationship to the mean will be considerably lower.
We have long known that income inequality shrank from the early decades of the last century until the late 1960s. It has risen more or less steadily ever since. See this graph of the GINI coefficient.
I've seen many people try to link rising inequality to political policies like the 1980s tax cuts. Such actions may have had a marginal impact but the event I see that correlates much better is the rise of the women's movement in the late 1960s and improving earning potential of women in the following decades.
(On a side note, these discussions are usually about household "income." This is income before taxes, transfers, and non-cash benefits. Transfers and non-cash benefits have expanded greatly over the last fifty years. When factored in, some the disparity shrinks. Even the lowest quintile of households has seen improvement, just not as great as quintiles further up the scale.)