A curious thing about the polls in 1948: Everyone knows that they predicted a Dewey victory, and that they were wrong. What is odd is that hardly anyone seemed to notice at the time--or to remember afterwards--that the Gallup poll at least showed a steadily narrowing race:
"Yet even a cursory reading of Gallup's results revealed signficant Dewey slippage. From a formidable 17 percent advantage in late September, he had tumbled to 9 percent in mid-October, and to just 5 percent from October 15 to 25. With Dewey fading and a Gallup margin of error of 3.9 percent, if anyone cared to look, a runaway had turned into a horse race." David Pietrusza, *1948: Harry Truman's Improbable Victory and the Year That Transformed America*, p. 363.
What if Gallup had polled the last week of the campaign and its final poll had shown, let us say, only a one or two point Dewey lead? Suppose there had been headlines on the Sunday before Election Day saying "Gallup says presidential race now too close to call." It would probably have been too late for Dewey to make a last-minute change in strategy. What would have changed, though, would be (1) our whole perception of 1948 as an incredible upset that nobody could have foreseen, and (2) the bad name that pollsters had for years after 1948. (Many newspapers cancelled their subsciptions to polls; in 1952 the pollsters were so chastened by what had happened in 1948 that they hesitated to predict an Eisenhower victory that should have been obvious; and really only after 1960, when the pollsters correctly predicted an extremely close JFK-Nixon race, did they regain their lost prestige.)
(In fairness to George Gallup, he kept polling a bit later than Archibald M. Crossley, who stopped polling after October 18, and reported a 49.9 to 44.8 percent Dewey lead in his final poll; or, worst of all, Elmo Roper who announced that his September 29 poll was so conclusive--Dewey leading 44 to 31, with 16 percent undecided--that he would suspend polling on the race, barring some unforeseen development--which did not occur.)
Now you may say: With the polls showing a close race, couldn't tunout have been higher, and couldn't that have affected the final results? The problem with this argument is that it is far from clear which candidate a larger turnout would have helped. Many Republicans after the election criticized Dewey for running a play-it-safe, bland, platitudinous ("America's future lies ahead of it") campaign that failed to excite Republicans enough to get them to come to the polls. (What these critcs often failed to remember was that Dewey *had* run a hard-hitting campaign in 1944--portraying FDR as a tool of Sidney Hillman, Earl Browder, and corrupt urban machines--without success. This of course does not necessarily mean that the critics were wrong; tactics that didn't work in 1944 might have been more successful in 1948.)
But the truth is that Democrats as well as Republicans were unexcited in 1948. Dewey's very blandness--as well as the fact that the polls showed him a sure winner--may have led some potential Truman voters to stay home because a Dewey victory was inevitable and probably wouldn't be such a bad thing, anyway. Some liberal Democrats, like Dorothy Schiff of the then-very-liberal *New York Post*, actually supported Dewey; "I feel,' she argued "that a competent and fairly liberal Republican will make a better president than an incompetent and inconsistent Democrat." David Pietrusza, *1948: Harry Truman's Improbable Victory and the Year That Transformed America*, p. 323. Many Democrats who would not go quite that far must have thought that Dewey's winning was, if not positively preferable, at least a matter of indifference, as well as being inevitable--so why bother to vote? For some others, the fact that Dewey was sure to win anyway meant that there was no harm in casting a symbolic protest vote for Henry Wallace or Norman Thomas (even if Dewey really was worse than Truman, which many of them doubted). Such voters may have been numerous enough to be responsible for Dewey's narrow victory in New York.
Indeed, Samuel Lubell argued that low turnout--particularly from northeastern Democrats--may actually have saved Dewey from a more decisive defeat. ("Far from costing Dewey the election, stay-at-homes may have saved him as crushing a defeat as Landon in 1936." Quoted in Pietrusza, p. 406)
"Yet even a cursory reading of Gallup's results revealed signficant Dewey slippage. From a formidable 17 percent advantage in late September, he had tumbled to 9 percent in mid-October, and to just 5 percent from October 15 to 25. With Dewey fading and a Gallup margin of error of 3.9 percent, if anyone cared to look, a runaway had turned into a horse race." David Pietrusza, *1948: Harry Truman's Improbable Victory and the Year That Transformed America*, p. 363.
What if Gallup had polled the last week of the campaign and its final poll had shown, let us say, only a one or two point Dewey lead? Suppose there had been headlines on the Sunday before Election Day saying "Gallup says presidential race now too close to call." It would probably have been too late for Dewey to make a last-minute change in strategy. What would have changed, though, would be (1) our whole perception of 1948 as an incredible upset that nobody could have foreseen, and (2) the bad name that pollsters had for years after 1948. (Many newspapers cancelled their subsciptions to polls; in 1952 the pollsters were so chastened by what had happened in 1948 that they hesitated to predict an Eisenhower victory that should have been obvious; and really only after 1960, when the pollsters correctly predicted an extremely close JFK-Nixon race, did they regain their lost prestige.)
(In fairness to George Gallup, he kept polling a bit later than Archibald M. Crossley, who stopped polling after October 18, and reported a 49.9 to 44.8 percent Dewey lead in his final poll; or, worst of all, Elmo Roper who announced that his September 29 poll was so conclusive--Dewey leading 44 to 31, with 16 percent undecided--that he would suspend polling on the race, barring some unforeseen development--which did not occur.)
Now you may say: With the polls showing a close race, couldn't tunout have been higher, and couldn't that have affected the final results? The problem with this argument is that it is far from clear which candidate a larger turnout would have helped. Many Republicans after the election criticized Dewey for running a play-it-safe, bland, platitudinous ("America's future lies ahead of it") campaign that failed to excite Republicans enough to get them to come to the polls. (What these critcs often failed to remember was that Dewey *had* run a hard-hitting campaign in 1944--portraying FDR as a tool of Sidney Hillman, Earl Browder, and corrupt urban machines--without success. This of course does not necessarily mean that the critics were wrong; tactics that didn't work in 1944 might have been more successful in 1948.)
But the truth is that Democrats as well as Republicans were unexcited in 1948. Dewey's very blandness--as well as the fact that the polls showed him a sure winner--may have led some potential Truman voters to stay home because a Dewey victory was inevitable and probably wouldn't be such a bad thing, anyway. Some liberal Democrats, like Dorothy Schiff of the then-very-liberal *New York Post*, actually supported Dewey; "I feel,' she argued "that a competent and fairly liberal Republican will make a better president than an incompetent and inconsistent Democrat." David Pietrusza, *1948: Harry Truman's Improbable Victory and the Year That Transformed America*, p. 323. Many Democrats who would not go quite that far must have thought that Dewey's winning was, if not positively preferable, at least a matter of indifference, as well as being inevitable--so why bother to vote? For some others, the fact that Dewey was sure to win anyway meant that there was no harm in casting a symbolic protest vote for Henry Wallace or Norman Thomas (even if Dewey really was worse than Truman, which many of them doubted). Such voters may have been numerous enough to be responsible for Dewey's narrow victory in New York.
Indeed, Samuel Lubell argued that low turnout--particularly from northeastern Democrats--may actually have saved Dewey from a more decisive defeat. ("Far from costing Dewey the election, stay-at-homes may have saved him as crushing a defeat as Landon in 1936." Quoted in Pietrusza, p. 406)