Today, we take the polling data plastered across news channels for granted. These polls are scientific polls, a type of polling that relies on statistics to draw accurate conclusions about a big group from a small group. However, scientific polling only became popular to report on in the 1940s. Before then, election betting odds were used as public polls. Newspapers would report the odds in the weeks leading up to the presidential election. But the first presidential Gallop poll changed the way that news outlets reported election developments.
Why Election Betting Odds Were Popular For So Long
Election betting has a long history that stretches as far back as George Washington’s election. Election betting was popular among the electorate and it grew into an industry all its own. After 1865, specialists founded firms dedicated to election betting.
Over time, Wall Street became the epicenter of election betting. Betting commissioners, or election bookies, would set odds and take a 5% commission on winnings bets. So, the commissioners would profit regardless of the outcome. This allowed commissioners to remain impartial as they set odds.
It also resulted in pricing that market-makers found fair. This resulted in a prediction market that was largely run by analysts who were viewed as impartial. These analysts were experienced at allocating assets and hedging risk. So, their opinions about the standings of presidential candidates mid-race were credible.
This contrasts with prediction markets that are open to everyone. The quality of a market’s conclusion depends on how qualified each person in that market is to draw that conclusion. For example, the Wall Street election markets were viewed as accurate representations of the both candidates’ standings. (Although, these markets didn’t predict the margin of victory well. The mid-October favorite usually won.)
In contrast, Trump voters flooded election markets overseas with enough money to skew the odds to extremes. Those extreme fluctuations made the odds unreportable even if news organizations still used them in election coverage.
Critically, election odds did not decide elections. The electoral college results decided the election, and payouts followed from the results. Even when election betting markets were common, the voters chose the president.
How Scientific Polls Replaced Odds
Election odds weren’t the only ways presidential elections were predicted. Literary Digest also sent out a poll to its readers to estimate the candidates’ chances. Its national reach allowed it to draw from a large sample from across the United States.
But in 1936, a man named George Gallup made two predictions: that Franklin Roosevelt would win the election and that the Literary Digest poll would incorrectly predict Alf Landon to win. Gallup was right on both counts, even estimating Literary Digest’s incorrect margin within 1% of its final poll results.
George Gallup recognized that even though Literary Digest’s poll had a large sample, that large sample missed a crucial voting bloc. As Time reported, the polls that Literary Digest sent to its telephone and auto registration lists “took no account of the low-income voters who had swung solidly behind the New Deal.” Literary Digest’s 1936 poll failed to poll FDR’s most fervent supporters.
To make his predictions, Gallup separated the United States into nationally representative groups and made sure they were in his sample of 3,000 people. This accurate polling methodology also gave reporters an alternative to election betting odds. While Gallup hadn’t invented scientific polling, his public triumph over Literary Digest gave reporters better ways to report on public opinion than Literary Digest and election betting markets.
Regulation Against Election Betting
Throughout the early 1900s, anti-gambling parties tried to outlaw gambling. This would periodically dampen election betting markets. But these markets would shortly recover with the ebb and flow of anti-gambling sentiment.
But in the 1920s, financial regulators began cracking down on election betting. According to Rhode and Strumpf’s 2004 article in the Journal of Economic Perspectives, Wall Street didn’t want its “socially productive risk-sharing and risk-taking functions with gambling on inherently zero-sum public or sporting events.” The New York Stock Exchange and Curb Market forbid its members from participating in election betting in 1924. (That didn’t stop newspapers from reporting the odds “without naming the participants,” though.)
In 1939, New York legalized horse race betting, which gave gamblers a gambling outlet with shorter payout times. Instead of waiting for an election cycle to end, bettors could bet at the horse tracks and receive same-day results.
In the face of scientific polling, anti-gambling regulations, and periodic anti-gambling sentiment, election odds faded from newspapers in the 1940s.
Taken together, it’s easy to see why CNN has a wall of poll results rather than election odds leading up to Election Day.