Don't compare "household" data over time

Single Americans make up more than half of the adult population for the first time since the government began compiling such statistics in 1976.... In 1976, it was 37.4 percent and has been trending upward since.

-- link

This is why you should ignore any statistic that points at changes in -whatever- per household over time; households are an ever-changing unit of measurement. One example where you often see this mistake is people pointing out the stagnating incomes of the median household over the past few decades. But when a household increasingly refers to one person rather than two people, that comparison just doesn't work.