Four potential Republican candidates for president ran into a nasty case of mode effects Tuesday. The results, for them, were not pretty.

The term "mode effects" refers to differing results when surveys are done different ways. In this case it's the difference between a printed ballot - Tuesday's national exit poll - and a telephone survey, and how they handle people with no opinion.

Let's start with the finish: Anywhere from 60 to 64 percent of voters who took the exit poll Tuesday said Jeb Bush, Chris Christie, Rand Paul and Rick Perry alike would not make a good president. That's a lot - enough to make a guy feel practically unelectable.

But an ABC News/Washington Post poll we completed last week asked just the same question - and many fewer likely voters, 48 to 51 percent, said each of these four would not make a good president. If hardly cheering numbers, those still are less-bad places to start.

The numbers saying these candidates would make a good president, meanwhile, are about the same in both surveys, ranging from 22 to 29 percent.

What gives? Undecideds. In our ABC/Post poll, conducted by telephone interviews, anywhere from 21 to 28 percent of respondents had no opinion on whether each of these candidates would make a good president. We know because they volunteered that information to us.

But in the exit poll - given the nature of visual questionnaires - there was no way for them to volunteer that information. It asked whether each potential candidate would make a good president, with options of "yes" and "no." The only way out was to skip the question entirely. Twelve to 15 percent did that, fewer than the numbers who told our interviewers they had no opinion.

Interestingly, this effect did not appear when the question was asked about the better-known Hillary Clinton. Five percent of likely voters in our survey said they had no opinion of her, and five percent in the exit poll skipped the question. That supports the conclusion that some of those who were undecided about other potential candidates in the phone poll simply couldn't express that sentiment in the exit poll.

We learn from both polls. Our pre-election poll makes clear that lots of people have are skeptical of Bush, Christie, Paul and Perry as presidential timber, and that lots of others haven't yet formed an opinion of them. The exit poll suggests that when those without an opinion on these four don't have a chance to say so, they're most apt to pick the negative option. That's worth knowing, but so is the fact that the 60-percent negatives on these candidates in the exit poll are likely not nearly as hard and fast as it may otherwise seem.

A few other items are worth considering, as well. One is the fact that the four Republicans all are much less apt to be seen as presidential material within their own parties than Clinton is within hers. That reflects the highly unsettled nature of GOP preferences - of 13 candidates we tested last month, excluding Mitt Romney, no one exceeded 13 percent support for the Republican nomination. Clinton, by contrast, has a huge Democratic constituency behind her.

Another is the fact that elections ultimately are comparative; what'll matter is not whether any individual candidate is seen as being likely or unlikely to make a good president, but which of those running is seen as best suited for the job. That's also displayed in the exit poll. Voters on Tuesday said by an 11-point margin, 53-42 percent, that they thought Clinton would not make a good president. But she also trailed an unnamed Republican nominee in a 2016 matchup by less of a margin, 6 points, 34-40 percent - with a quarter saying it depends.

Indeed, if the potential GOP candidates don't like their "not a good president" numbers, Clinton may be equally unhappy with her 6-point deficit against a generic Republican opponent. But there's room for reconsideration there, too, since the 2016 electorate likely will differ a good deal from that of 2014. Consider this: In an exit poll question in the 1994 midterms, Bill Clinton trailed the generic Republican by 8 points. Two years later, he won in a landslide.

(Langer Research Associates)