I think that the first time I encountered the term “pew gap” was in the middle-to-late 1980s, as the side effects of the post-Roe v. Wade era began to emerge.
That was when people started talking about the impact of the Religious Right on the Republican Party and the growing secularization of the elites at the top of Democratic Party structures, where old-school labor union Catholics were being replaced by various kinds of white-collar groups linked to academia and feminism.
At the end of the 20th Century, the “pew” or “God gap” was a given. I know that I have used it before, for this piece of the “Blue Movie” essay that The Atlantic ran in 2003 remains perfect:
Early in the 1996 election campaign Dick Morris and Mark Penn, two of Bill Clinton's advisers, discovered a polling technique that proved to be one of the best ways of determining whether a voter was more likely to choose Clinton or Bob Dole for President. Respondents were asked five questions, four of which tested attitudes toward sex: Do you believe homosexuality is morally wrong? Do you ever personally look at pornography? Would you look down on someone who had an affair while married? Do you believe sex before marriage is morally wrong? The fifth question was whether religion was very important in the voter's life.
Respondents who took the "liberal" stand on three of the five questions supported Clinton over Dole by a two-to-one ratio; those who took a liberal stand on four or five questions were, not surprisingly, even more likely to support Clinton. The same was true in reverse for those who took a "conservative" stand on three or more of the questions. (Someone taking the liberal position, as pollsters define it, dismisses the idea that homosexuality is morally wrong, admits to looking at pornography, doesn't look down on a married person having an affair, regards sex before marriage as morally acceptable, and views religion as not a very important part of daily life.) According to Morris and Penn, these questions were better vote predictors — and better indicators of partisan inclination — than anything else except party affiliation or the race of the voter (black voters are overwhelmingly Democratic).

