Three months before the November 6 US midterm elections, the FiveThirtyEight poll aggregate gives Donald Trump a 41.7% approval rating and a 52.5% disapproval rating, for a net approval of -10.8. Despite issues such as Trump’s meeting with Russian President Vladimir Putin and immigration, his ratings have been fairly stable since rising to their present level in early May, helped by the good US economy.
Trump performs poorly when compared to other presidents at this stage in their presidencies. When comparing net approval, the FiveThirtyEight aggregate shows Trump is only ahead of Harry Truman, president from 1945-53. Presidents prior to Truman came before there was much scientific polling.
According to CNN analyst Harry Enten, the majority of presidents approaching a midterm had an overall job approval far higher than their approval of handling the economy. The two largest exceptions were Bill Clinton in 1998, owing to the Monica Lewinsky scandal, and Trump now. In both these cases, the president’s approval on the economy far exceeded his overall approval.
The implication of Trump’s economic vs overall approval gap is that the good US economy is holding up his approval. If the economy does not perform well, Trump’s ratings are likely to suffer a large drop.
Trump has bragged that his approval ratings with Republicans are higher than Abraham Lincoln’s. Lincoln died long before scientific polling started, so we do not know what his ratings were.
The Huffington Post Pollster aggregate has Trump’s approval with Republicans at a high 85%, but self-identified Republicans made up just 26% of all registered voters in 2017 according to the Pew Research Centre, down from 29% in 2016. This suggests that Trump is appealing to a diminishing base.
Trump’s popularity with Republican voters explains why criticism from prominent Republican politicians over issues such as Russia and tariffs has been muted. According to Enten, candidates endorsed by Trump have won Republican primaries, but Trump’s poor overall ratings are a negative for Republicans in a general election.
Republicans barely hold Ohio 12 seat at byelection, and midterm election polling
On Tuesday, the Republican, Troy Balderson, defeated the Democrat, Danny O'Connor, by a 50.2-49.3 margin in a byelection for Ohio’s 12th Congressional District. Provisional votes are likely to narrow the gap, but not overturn the current Balderson lead.
Trump won this district by an 11.3 point margin over Hillary Clinton in 2016, and Mitt Romney won by a similar 10.5 point margin against Barack Obama in 2012. This district is historically Republican; the Democrats have only won it once since the 1930’s, in 1980. A loss by less than one point is a strong result for the Democrats.
In the FiveThirtyEight aggregate, Democrats hold a 47.4-41.2 lead over Republicans in the race for Congress. This aggregate has been more volatile than the aggregate of Trump’s ratings, and the recent move towards the Republicans could vanish soon.
All 435 House seats are up for election on November 6. Owing to natural clustering of Democratic voters and Republican gerrymandering, Democrats probably need to win the House popular vote by about seven points to take control.
35 of the 100 Senate seats are also up for election on November 6, including two Senate byelections in Mississippi and Minnesota. 26 of these seats are currently held by Democrats and just nine by Republicans. Democrats will be defending five states that voted for Trump by at least 18 points.
Despite the tough Senate map for Democrats, RealClearPolitics gives Republicans just a 48-45 lead for the Senate including seats not up for election. Seven states are rated toss-ups; these occur when the polling average has a lead under five points. If toss-up states are assigned to the current leader, Republicans hold a 51-49 Senate lead, suggesting no net change from the current Senate.
Australian polls: Newspoll and Essential both 51-49 to Labor
Last week’s Newspoll, conducted July 26-29 from a sample of 1,700, gave Labor a 51-49 lead, unchanged since three weeks ago. Primary votes were 39% Coalition (up one), 36% Labor (steady), 10% Greens (steady) and 7% One Nation (steady). Three of the four days of this poll’s fieldwork were taken before the Super Saturday byelection results were known.
This was the Coalition’s 37th successive Newspoll loss under Malcolm Turnbull, four more than the previous record of consecutive Newspoll losses for a government. However, the primary vote shift in this poll indicate the Coalition is closing in on a 50-50 Newspoll.
42% were satisfied with Turnbull’s performance (up one), and 48% were dissatisfied (down one), for a net approval of -6, equal with a Newspoll four weeks ago for Turnbull’s best net approval this term. Bill Shorten’s net approval fell one point to -25. Turnbull maintained an unchanged 48-29 lead as better PM.
Last week’s Essential, also conducted July 26-29 from a sample of 1,022, gave Labor a 51-49 lead, unchanged since three weeks ago. Primary votes were 41% Coalition (up one), 36% Labor (steady), 10% Greens (steady) and 6% One Nation (steady). Essential is using 2016 election flows, and this poll would be 50-50 by Newspoll’s new methods.
For more on Newspoll and Essential, including best to lead Labor and Liberal questions and party attribute changes, see my personal website. Turnbull and Anthony Albanese have gained in Liberal and Labor leadership polling.
Greenpeace ReachTEL poll: 52-48 to Labor
A ReachTEL poll for Greenpeace, conducted July 30 – two days after Super Saturday – from a sample of 4,000, gave Labor a 52-48 lead by respondent allocated preferences. Primary votes, after assigning undecided through a forced choice, were 37.3% Coalition, 34.4% Labor, 12.1% Greens and 7.9% One Nation. A Victorian breakdown of this poll gave Labor a 57-43 lead with the Greens on a high 18.0%.
ReachTEL always asks for voting intentions first, but other questions in these polls are skewed towards Greenpeace’s environmental agenda.
Adrian Beaumont does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
Authors: Adrian Beaumont, Honorary Associate, School of Mathematics and Statistics, University of Melbourne