Charted: Why early midterm results can be misleading

Data: AP via The New York Times and Axios research; Chart: Andrew Pantazi/Axios The first votes counted Tuesday night will almost certainly look redder than the final result, and control of the House and Senate could take days to settle. Why it matters: Early leads can create false impressions of who's winning and suspicion when those leads vanish. Election night is really election week. By the numbers: AP didn't call control of the House until seven or eight days after Election Day in each of the past three elections . The Senate was settled by 10:44pm on election night in 2018 and at 12:06am the next morning in 2024 . In 2022, it took four days , and in 2020, it wasn't called until Georgia's January runoffs , two months later. State of play: Republicans hold 53 Senate seats . Democrats need a net gain of four for a majority. The red head start: In every national election since 2018, the House and Senate votes counted by 9pm ET were 4 to 9 percentage points more Republican than the final result, an Axios analysis of AP and Edison Research vote counts found. It's mostly geography. Fast-counting Florida and other right-leaning states report first. At 9pm on election night 2022, Florida accounted for 40% of all Senate votes counted nationwide. By 11pm, more House races had been called for Republicans than Democrats in all four elections, including 2018 and 2020, when Democrats won the House. The tilt fades, slowly. At 6am Wednesday, the national House count was still 2 to 3 points more Republican than the final, largely because California, Washington and Oregon count for days. Data: Axios research; Chart: Andrew Pantazi/Axios Zoom in: Each state tilts its own way, depending on which ballots and which counties it counts first. States that count early and mail ballots first tend to start bluer than they finish. States that count Election Day votes first start redder. Ohio, Nebraska, Texas, Kansas and Iowa start blue. Sherrod Brown and Bernie Moreno were tied at 9pm in 2