What to know about CBS News' 2022 midterm elections Battleground Tracker

The CBS News Battleground Tracker is back — explaining what's on voters' minds and regularly providing detailed snapshots of the 2022 midterm elections in every state and district throughout the 2022 campaign. A key feature is our model estimating how many seats each party is currently poised to win in the U.S. House of Representatives.

Where exactly do our Battleground Tracker numbers come from? Here are five things to know.

1. Analyzing races district by district

We take a district-by-district approach to analyzing races and measuring public opinion, since control of Congress is won across hundreds of individual elections, not by national popular vote. Indeed, relying too much on national polling can be misleading, because the party that wins the most votes nationally doesn't necessarily win the most seats.

The Battleground Tracker estimates individual districts and translates each major party's current support to the metric that matters: seats in Congress. Our approach also gives you a sense of what voters in different parts of the country think about this year's candidates and big issues.

2. More than just a poll

While surveying voters across the country is an integral part of the Battleground Tracker, this is a lot more than your typical poll. At its core is a statistical model fueled by big data. We combine polling, voter files, U.S. Census data, and historical election results to render a clearer picture of what's going on in every state and district.

Here's what the data tells us:

• We learn which party different kinds of voters are supporting from our ongoing polling, which includes much bigger samples — in the tens of thousands — than most surveys.

• We estimate how many voters like them are in each state and district, as well as their turnout history, from voter files and Census data.

• And we know each state and district's previous election results, which enables us to anchor our 2022 estimates to recent history.

Our model combines all this data using multilevel regression with post-stratification. A feature of this technique is that we use trends across the country to inform our picture of a specific district. If we find Hispanic voters across the South shifting support, for instance, we use the information to more precisely estimate specific districts in which Hispanic voters live. The same applies for many other types of voters. The survey lets them tell us what they are thinking, and we map that to how many of them live in each district.

We collaborate on data collection and modeling with YouGov, building on our successful 2018 Battleground Tracker, as well as the 2020 edition that encompassed both the presidential primaries and general election.

3. House landscape informs the model

The landscape of the newly drawn U.S. House districts is a challenge for Democrats and gives Republicans a built-in advantage. Democrats often find their voters packed into fewer districts,  due both to their propensity to live in more concentrated areas and to gerrymandering, which packs voters by design. And despite the population growth among racial minorities over the past decade, many states did not add minority districts. If the national popular vote ended up tied 50-50, Republicans would be expected to win the majority of seats.

Based on 2020 presidential votes, President Biden's performance was sufficient to have carried a majority of the new districts nationwide. However, enough of those districts are close enough so that a small vote shift toward Republicans would flip many seats. And even if Democrats saw voters slightly shift toward them, they simply have fewer pickup opportunities than Republicans do this year.

On top of that, we've seen dissatisfaction with Mr. Biden and the state of the country, lots of Democratic members of Congress retiring, and indications of a Republican turnout advantage — all of which add up to a favorable environment for GOP pickups.

4. Snapshots, not forecasts

We tell you where races stand today, explaining why and what might change. Unlike forecasting, we're estimating each candidate's current support, incorporating all the data we've collected up to this point. There's nothing here to account for forward-looking uncertainty — we don't model changes in economic conditions or future campaign dynamics, for example. We fully anticipate movement before the first vote is cast, so we'll update everything regularly in the months ahead.

For example, if we estimate a party at 230 seats with a margin of error of ±12 seats, we are confident that their support translates to between 218 and 242 seats today — that's different from forecasting their final tally. Similarly, our qualitative ratings of individual races reflect their current status. A race that's leaning toward a party today could be reclassified as a toss-up if it becomes more competitive.

5. Solid track record

While we take a different approach than traditional polling, the Battleground Tracker is based on rigorous methods from the fields of political science, survey research, and statistics. Moreover, we have a strong track record employing similar models at CBS News over the past few years.

Our 2018 model performed particularly well, steadily tracking Democrats' improvement in key congressional races and the eventual blue wave in the House. In fact, our high-turnout scenario nailed the final result, when it came to pass that historic turnout powered Democratic gains.

Our highly accurate race ratings in 2020 were based on a similar model. We estimated that the Democrats had built a lead heading into Election Day, but that Republicans could close the gap with a late turnout surge. Every state we classified as leaning or likely Democratic pre-election wound up going to Mr. Biden, and every state we rated as leaning or likely Republican went to Donald Trump. And of the six states we rated as toss ups, Trump won four and Mr. Biden won two.

Note on Senate ratings

Our Senate ratings are based on a blend of quantitative and qualitative information, including polling, along with information about the candidates nominated, current national environment, and recent electoral history in each state. In states with CBS News polling this cycle, we rely heavily on our survey-based estimates of voters' preferences. A race rating is informed in part by the estimate, the statistical uncertainty around it, historical trends, and the dynamics of the current campaign.

In:
• Redistricting
• United States Congress
• Democratic Congressional Campaign Committee
• Voting
• Democratic Party
• 2020 Census
• Republican Party
• United States Census Bureau