Donald Trump leads Hillary Clinton by 50 percent to 44 percent in a two-way race, and by 44 percent to 37 percent in a field including Libertarian Gary Johnson (10 percent) and Green Party candidate Jill Stein (2 percent), according to the new Quinnipiac survey of 612 likely Iowa voters. Trump has gained ground since last month’s Quinnipiac poll showing a small lead for Clinton in both the two-way and four-way races. The polling memo noted Trump’s 52 percent to 26 percent lead among men, which overcame Clinton’s 47 percent to 37 percent advantage with women. Trump also had higher levels of support among Republican respondents (86 percent) than Clinton did among Democrats (83 percent).

Nuggets from the cross-tabs:

• Independents split 38 percent Trump, 33 percent Clinton, 19 percent Johnson, and 4 percent Stein in a four-way race. Against Clinton alone, Trump led 47 percent to 41 percent among independent respondents.

• As expected, Trump has a big lead among non-college-educated whites: 55 percent to 41 percent in a two-way, 48 percent to 33 percent in a larger field.

• Surprisingly, Trump leads among whites with a college degree (49 percent to 44 percent)

• By self-reported party identification, 33 percent of this poll’s respondents were Republicans, 30 percent Democrats, 33 percent independents, and 5 percent other/don’t know. That party breakdown better reflects the 2012 general electorate in Iowa than did last week’s Monmouth poll.

According to Quinnipiac, its poll has a margin of error of plus or minus 4 percentage points–though that number refers to only one of many potential sources of error in opinion polls. I am seeking further details on Quinnipiac’s likely voter screen, which wasn’t explained in the note on methodology. Nate Cohn’s latest post at the New York Times Upshot blog explained how four different approaches to separating “likely voters” from other poll respondents produced four different results from the same set of interviews with North Carolina voters.