Stevenson would have been served well by running against Donald Trump. According to Gallup's most recent numbers, only 42 percent of Americans view the presumptive Republican nominee positively -- and 42 percent view him highly unfavorably. (That figure for Eisenhower? Four percent.)
Of course, Hillary Clinton doesn't do that much better. She's viewed highly unfavorably by a third of the country -- a figure lower than any candidate since 1956, save Trump. (Gallup doesn't have data for 1988, 1996 or 2000, but it seems safe to assume that numbers for candidates those years were higher.)
If we look at the net favorability ratings for candidates past and present, it becomes immediately obvious how this year's crop of candidates are outliers. If we subtract the percentage of people who view the candidates strongly unfavorably from those who view them strongly favorably, we get a net value that we can compare to the results of the elections. There's a loose correlation; candidates that were viewed more strongly positively on net tended to do better on Election Day.
But notice where the two candidates this year are: In no other contest were both candidates in net-negative territory. In fact, only one other candidate ever was in that territory: Barry Goldwater, who got blown out by Lyndon Johnson in 1964. John Kerry was at a net zero in 2004. George McGovern was at plus-1 in 1972.
But the thing about elections is that the candidates run against each other. If we consider the margin of favorability between the two, the picture changes. In every election for which Gallup has data, save one, the candidate whose net strong favorability was higher won -- and the margin of victory correlates fairly strongly to how much more popular the candidate was. (A quick note: The correlation between net strong favorability and election outcomes is actually slightly stronger than the correlation of the outcome and overall net favorability.)
The not-so-stunning conclusion? Being more popular than your opponent is linked to victory. The one exception was in 1980. That year, Jimmy Carter was viewed slightly more strongly positively than Ronald Reagan, but lost anyway.
Clinton, right now, is about 15 points more positively viewed on net than Trump. The two closest previous contests in that range had widely divergent outcomes. Richard Nixon was 13 points more positively viewed than McGovern in 1972 and beat him by a massive 23 points. Carter was 17 points more positively viewed than Gerald Ford in 1976, but only won by 2 points.
So, as with every other bit of data in July of a presidential election year, don't consider this bit of data highly predictive -- especially because of the "right now" in the preceding paragraph. These numbers can and will still move around.
But know that if a candidate suddenly sprang forth from the sea like Venus who had favorability numbers like Dwight Eisenhower, this thing wouldn't be a contest.