Today, "populism" is usually a negative term, basically meaning bombastic, authoritarian, and incompetent (this article is an example). Things used to be different. In 1972, Irving Kristol (a conservative) started an article by asking "What is populism and why is everyone suddenly saying such nice things about it?" Jack Newfield and Jeff Greenfield (liberals) published a book called A Populist Manifesto which started from a similar point--that lots of people were calling themselves populists--before making a case for progressive populism. They held up Robert Kennedy's 1968 campaign as a model for what they were proposing, and Eugene McCarthy's 1968 campaign as a sort of negative model. The idea was that Kennedy tried to understand white working-class voters, while McCarthy ignored or dismissed their concerns. They said that if the Democrats followed McCarthy's example, the white working class would turn to people like Richard Nixon and George Wallace. Similar claims are still made today--that if the Democrats had followed Robert Kennedy's example, they would have had more success in holding on to working-class voters.
In May 1968, a Gallup poll asked about hypothetical races between Nixon, Wallace, and the th(ree leading Democratic contenders, making it possible to compare the support for McCarthy and Kennedy. There were substantial differences--bigger than I expected. First, Kennedy got 86% of the hypothetical vote among blacks, while McCarthy got only 46% (with 36% saying they'd vote for Nixon). I don't know why McCarthy did so poorly--maybe it was because he had challenged Lyndon Johnson, who was still popular among blacks. Because of this difference, I restricted the rest of the analysis to non-blacks.
McCarthy did substantially better than Kennedy among college graduates, although both of them ran behind Nixon (McCarthy trailed by 36%-50%, while Kennedy trailed by 19%-64%). Kennedy did better among people without a high school diploma (who were almost 40% of the sample). Income was measured with 11 categories, so it's more convenient to compute the mean income by vote. The average income of Kennedy supporters was only a little higher than the average of Wallace supporters and undecideds, and substantially lower than that of Nixon supporters. (The differences among Kennedy, Wallace, and undecideds were not statistically significant). The average incomes of McCarthy supporters was closer to that of Nixon supporters than to either Wallace supporters or undecideds.
At that time, the Gallup poll asked about the occupation of the "chief wage earner." McCarthy did considerably better among professionals and managers, and Kennedy did somewhat better among all of the other occupational groups.
So in terms of demographics, the Newfield/Greenfield account holds up. But support for Wal]lace was almost exactly the same in the hypothetical race with Kennedy as in the one with McCarthy (actually a little higher with Kennedy). That is, although Kennedy appealed to the kind of people who were inclined to support Wallace--less educated, lower incomes, lower-status occupations--he didn't seem to appeal to the people who were inclined to support Wallace. 86% of the people who said they'd support Wallace in a three-way race involving McCarthy said they'd support him in a three-way race involving Kennedy. There was considerably more movement between Nixon and the possible Democratic nominees.
Kennedy and McCarthy didn't do as well as Humphrey in a hypothetical three-way race--I wouldn't put much weight on that, simply because people may have been more familiar with Humphrey.* But McCarthy did a little better than Kennedy among whites, almost exactly balancing Kennedy's advantage among blacks. So it looks like Kennedy would have attracted different voters, but not necessarily more voters.
*In terms of demographics, Humphrey's support was in between Kennedy's and McCarthy's.
[Data from the Roper Center for Public Opinion Research