Rick Scott and Beto O'Rourke lead the pack by a wide margin in campaign spending, with $66 and $59 million respectively in campaign this cycle. While Scott is pouring money in TV ads across Florida, O'Rourke is buying up not just TV, but also Facebook and Twitter ads in bulk.
The next-highest spender, Sen. Claire McCaskill, spent just about half as much as O'Rourke, with $31.9 in expenditures so far. Other high-spenders include candidates running in close races, such Sen. Bill Nelson in Florida, Scott's opponent, Rep. Jacky Rosen in Nevada, and Doug Jones, who narrowly beat Roy Moore in a special election for an Alabama Senate seat in 2017.
The dominance of ultra-wealthy, self-financed candidates among the highest-spending House and Senate candidates may give off the illusion that extravagantly rich candidates can effectively "buy" elections, but researchers have actually found otherwise.
Research from political scientist Adam Brown of Brigham Young-University concluded that candidates being able to spend exorbitant amounts of money with their own personal wealth has a relatively weak effect on their vote share compared to external donations.
He instead found the ability to fundraise large amounts from outside donors is a much better predictor of electoral success, arguing that candidates attracting lots of donations signals their attractiveness as a candidate to voters.
And even though Brown's work demonstrates the importance of outside fundraising, it doesn't determine the result of an election.
Jon Ossoff, the highest fundraiser and spender among House candidates this cycle, lost to Karen Handel by 4 points points in the Georgia special election, despite out raising her by a stunning $23 million, most of which came from small donors.
And in Pennsylvania's former 18th district, Democrat Conor Lamb won by a just a few hundred votes against Rick Saccone in the March 2018 special election, despite $11 million being spent to support Saccone (much of which came from outside groups) compared to Lamb's $4 million in spending.