Judging from my mail, I think some of you think I am back online making predictions. This is not true!

PEC provides aggregation and information to reveal where you can make the most impact. Since almost the start, my reason for operating PEC was to show how such efforts might be optimized. And as I wrote in 2016, I’m not estimating any probabilities this year – that’s why there are only Meta-Margins above. If you want probabilities, I think FiveThirtyEight does a bang-up job. There, now you know I’m serious!

Now…where to give – House, Senate, or governor? Let’s look at current conditions.

Senate. Democratic control of the Senate is somewhat unfavored, but there’s still uncertainty. This is not for want of polls. There are lots. Six races are currently showing poll medians within 3 points in either direction. Since election results can systematically deviate from opinion polls by up to four points (i.e. the polls can all be wrong in the same direction), either Democrats or Republicans could conceivably sweep all of these races. That still leaves a range of 45 to 51 Democratic/Independent seats!

Based on past elections, it’s not a bell-shaped curve; the ends of the range are quite likely. So Republican retention is not certain. Add in the possibility of a 50-50 split (power-sharing) and the fact that this year’s election sets the stage for 2020, and there are many reasons to care about this year’s Senate races.

Therefore the six close races are all worth supporting: AZ, FL, IN, MO, NV, and TN.

House. Here I list three streams of evidence. FiveThirtyEight puts them together, but I think there’s value in inspecting them separately. They all point toward a Democratic takeover, but none are definitive.

The generic Congressional ballot. PEC’s Meta-Margin is defined purely in terms of generic Congressional ballot polls. By this measure, the House has been between 0 and 4 percentage points above threshold for Democratic control (threshold = a 6% margin in the national popular vote, i.e. polls suggest a popular-vote win of 6-10%). The blue band on the beta site displays the 90% confidence band on this measure.

Special elections. These have predictive value, and suggest that Democrats are 3 percentage points above threshold for a takeover. The pink band displays the 90% uncertainty band.

Expert evaluations. Evaluations like the Cook Political Report seem to use district-level polls, funding, and candidate quality. If one takes the Solid/Likely/Lean Democratic seats and adds half the toss-ups, one gets an average of 224 Democratic seats – a handful over the 218 needed for a majority. If one treats the tossups as coin flips, the 90% confidence interval is 215-233 Democratic seats. That is not shown on the graph.

In summary, despite the near-certainty of a convincing popular-vote win, the question of control is still up in the air. The basic reason is gerrymandering, which I estimate gives Republicans about a 3-point bonus.

Based on all of this, one could certainly support close House races. But which ones? There are dozens of them all over the country.

One effective move would be to volunteer in a district near you (with the help of our Competitive District Finder), and donate in selected races. For example, one could give to close House races that are also in a state with a close Senate race. That means Nevada District 3, Arizona District 1, or Florida Districts 15, 26, and 27. All are listed in the ActBlue at left (along with most of the Senate races). On the Republican side, one could give to the National Republican Campaign Committee.

Governor. In some states, a change in the governorship will lead to divided-party control of government. This is important; Democrats are notoriously bad at focusing on state and local races. The closest key race is in Wisconsin, where Tony Evers (D) may unseat Scott Walker (R). The Supreme Court didn’t end Wisconsin’s famous gerrymander earlier this year. A future Governor Evers would be in office in 2021 for redistricting, and would make sure the offense wasn’t repeated.