(Crappy Brewers palette, Kalshi.)

Over the regular season the Brewers beat the Dodgers 4-3. Brewers won 103 games, Dodgers won 100. Brewers run differential was +214, Dodgers +201. Both finished their last 10 regular season games 8-2. Do those observed data justify a Dodgers advantage?

Of course, the Dodgers won the last two World Series, and of course they swept the Brewers in the 2025 NLCS. Those are powerful data, but that tells us priors at the start of the 2026 season. Updating from observed 2026 performance should give you a different prior for NLCS game 1.

Now what should the posterior look like after game 1 (2-1 Brewers)? After game 2 (7-4 Brewers)? After game 3 (6-2 Brewers)?

Is this too naïve a model? What about pitching matchups, home field, starting lineups? If those more sophisticated models would justify giving LA better odds in games 2, 3, 4, then those models also proved… wrong.

My priors for the World Series? Brewers in 5.

3 1/2 hour to Brewers vs Dodgers.

The other positive of the first term, at least relatively positive, was that Trump’s job approval rose from 2017 to 2018 from the mid-30s to the low-40s. Still not great, but a genuine recovery from a quite abysmal first year. In the second term, there has been steady decline with no such sustained recovery period. As of today his approval is 37.3%, by my trend estimate. That’s where he was at his worst in 2017.

In term two there have been only brief and small moves up, often after self-inflicted dips, as after “Liberation Day” when he subsequently backed off the April 2 tariffs, or after the end of the extended government shutdown in late 2025. There has been no similar rise since the start of the Iran war and a steady decline over the last 100 days.

With this opinion of the economy, and this approval of the president, it is hard to find a positive story with the midterms 39 days away.

News attention

The death of Dolly Parton gained the most attention to news stories this month, with 78% saying they heard a lot about it, while just 5% heard nothing at all. Attention was also widespread with only small differences by party, age, education, or source of news. In contrast, the pennant races in Major League Baseball received the least attention of nine topics, with just 7% hearing a lot and 69% nothing at all. Substantially more, 31%, heard a lot about the start of the NFL season.

Other topics receiving substantial attention include the war with Iran, with 67% hearing a lot; new tariffs on Canada, 51% hearing a lot; and Trump announcing he was renaming Lake Ontario, 50% hearing a lot. Less visible topics include the national debt reaching $40 trillion, 35% hearing a lot, and the projected depletion of the Social Security trust fund by 2032, 22% hearing a lot. Less attention was paid to the announcement by Alexandria Ocasio-Cortez that she would freeze her eggs.

Attention to events varies across sources of news and across topics. Parton’s death was quite visible regardless of news source, ranging from 74% to 89% hearing a lot. Attention to news about the war with Iran varied more, from 92% of MS Now viewers hearing a lot to 73% of Fox News viewers and 69% of those who rely on social media for news. Similar variation is seen for news of tariffs on Canada and renaming Lake Ontario. And news of the national debt passing $40 trillion was heard a lot by 66% of MS Now viewers but only by 34% of the Fox News audience.

About the Marquette Law School Poll

The survey was conducted Sept. 2-9, 2026, interviewing 1,023 adults nationwide, with a margin of error of +/-3.3 percentage points. For registered voters, the sample size is 864, with a margin of error of +/-3.6. For likely voters, the sample size is 581 with a margin of error of +/-4.3 percentage points. The full results of the poll can be found at the Marquette Law School Poll website here.

The public remains strongly skeptical of the building of data centers, with 71% saying the costs outweigh the benefits and 29% saying benefits outweigh the costs. In January, 62% said the costs were higher than benefits.

Views as to whether artificial intelligence is good or bad for society have fluctuated slightly since April, with 64% in September saying AI is bad for society and 36% saying it is a good thing.

Despite the skepticism about AI, a growing majority of the public says they have used an AI app in the last month, with 69% in September saying they are users—up from 61% in April.

Users of AI are considerably more likely to say AI will be a good thing for society, but, even among users, more say AI is a bad thing, 56%, than a good thing, 44%. Non-AI users are especially negative about AI’s impact, with 81% saying it will be bad for society.

The public is also strongly opposed to Flock cameras, which record license plates from passing cars and are used by police agencies across the country. Thirty-nine percent say these cameras are an important tool for police investigations, while 61% say they are unacceptable violations of privacy rights.

The public sees relatively limited benefits from data centers. Asked which two of six possible benefits were most important, 33% said maintaining U.S. leadership in AI, and 30% said the digital economy relies on data centers. Fewer, 29%, selected the creation of permanent jobs at data centers, and 19% said construction jobs were an important benefit. Least-mentioned benefits were local tax revenue at 15% and the establishment of new industry in their state, at 14%. A substantial 23% said none of these six considerations were benefits of data centers.

As for the costs of data centers, the most frequently cited aspect was high water use by 54%, followed by driving up electricity costs, with 51%. Less often cited was replacing rural or agricultural land, selected by 30%. Least-mentioned were tax subsidies, 14%, the creation of few permanent jobs, 14%, and noise from data centers, 13%. Given this list of potential costs, 7% said none of these six were important costs of data centers.

Data centers are not the only potential industrial developments opposed by a majority of American adults. The survey asked about five possible large industrial or commerce developments: chip fabrication plant, automobile manufacturing plant, e-commerce fulfillment center, steel or aluminum mill, or a data center. Each development consumes substantial land and resources while providing jobs and economic development. Respondents could select none, some, or all of the developments for what they would like to see in their area.

Of the five development options, no particular one was supported by even 33%, while 44% said they would not like any of these developed. Fifty-six percent selected at least one of the five options.

The most preferred development is computer chip fabrication plants, with 31% selecting that option, while 26% would like an automobile manufacturing plant and 26% favored development of an e-commerce fulfillment center. Less popular was a steel or aluminum mill, selected by 16%. The least popular option was data centers, at 12%

.Industry or manufacturing has been a backbone of American blue-collar employment for decades, and it is striking that there is very little difference in preference for these five operations between those polled who have or do not have college educations. An identical 44% of non-college and college-educated adults say they would want none of these developments in their area. College graduates are somewhat more in favor of chip fabrication plants than are non-college graduates, while non-graduates are bit more favorable to auto manufacturing or steel and aluminum mills. Neither group likes data centers.

Conclusions

Data centers remain opposed by over 70% of adults, a share of the public than has been relatively stable for several months. Likewise, 64% say AI will be a bad thing for society, despite a growing segment of the population who say they use AI, now 69% and growing. This opposition to data centers is especially impressive when considering other large industrial developments. Data centers rank last, behind chip fabrication, auto plants, e-commerce fulfillment centers and steel or aluminum mills. Fifty-four percent said at least one of these five were acceptable; only 12% said a data center was something they would favor.

Opposition to new tech is also seen in 61% who say Flock license plate readers are an unacceptable violations of privacy rights.

About the Marquette Law School Poll

The survey was conducted Sept. 2-9, 2026, interviewing 1,023 adults nationwide, with a margin of error of +/-3.3 percentage points. For registered voters, the sample size is 864, with a margin of error of +/-3.6. For likely voters, the sample size is 581 with a margin of error of +/-4.3 percentage points. The full results of the poll can be found at the Marquette Law School Poll website here.

Trump approval in second term

Approval of how Trump is handling his job as president declined in September to 37%, with 63% who disapprove. That is down from 40% approval in July, and a percentage point below the previous low of 38% approval in May and June. The course of approval has been downward since the start of Trump’s second term, when 48% approved and 52% disapproved. This trajectory has not been marked by dramatic or sudden drops. The steadiness of the decline from poll to poll has been notable, with approval falling by 0 to 3 points across successive polls, with the exception of a 2-point increase in the July 2026 poll. While successive changes have been small, the cumulative decline is substantial.

The strength of approval of Trump has also shifted over the second term. A substantial share of adults has strongly disapproved of Trump throughout 2025 and 2026, ranging from 40% in February 2025 to the current 47%. More change is seen in the balance of those who strongly approve and those who somewhat approve. Strong approval outweighed somewhat approval by 29% to 19% in February 2025, but as of September 2026, this is reversed, with 22% somewhat approving and just 15% strongly approving. This is a drop of almost half in strong approval since February 2025.

Approval of Trump varies substantially across issues, with his highest approval rating, 49%, for his handling of border security, and 43% approval on the issue of immigration. Each of the other six issues measured has approval ratings below 35%, with overall approval at 37%. Approval is especially low for the economy, war with Iran, gas prices, and inflation and the cost of living, each of which has approval below 30%. While approval on border security and on immigration each has remained relatively stable, approval on other issues has seen significant decline over time. Approval of the war with Iran has declined from 32% in April to 26% now.

Americans are especially negative on Trump’s handling of inflation and the cost of living, with just 19% approving. A substantial 64% say his policies are increasing inflation, while only 18% think they are decreasing it. An additional 18% think Trump’s policies are not affecting inflation. This represents a sharp downturn in public opinion from the beginning of the second term, when twice as many, 36%, thought his policies would lower inflation, though 49% said he would increase inflation.

There has been an especially striking change over the second term among Republicans in their belief whether Trump’s policies will decrease inflation. In December 2024, shortly after his reelection and before taking office, 76% of Republicans thought Trump’s policies would lower inflation. That percentage has fallen by more than half, to just 37% of Republicans who now think the president’s policies are reducing inflation. Among independents, only 28% thought Trump would lower inflation just after his election, but that, too, has fallen to 9% as of September. Democrats scarcely expected Trump to decrease inflation, declining from 8% to 3% now.

Most important issues

Inflation remains by far the most important issue for Americans, with 37% ranking it first among 10 issues considered in the September poll. The second-highest-rated issue is the economy, at 16%, followed by the war with Iran at 10%. The high ranking of inflation has been steady throughout the year, and by a large margin in each survey. Concern over the war with Iran has declined from 24% in April to 10% as of September, while the importance of most other issues has remained relatively stable. The importance of issues, and the changes over the year, are shown in the table.

Gasoline prices rose sharply with the start of the Iran war, and Americans are vividly aware of the change. In January, prior to the war beginning on Feb. 28, half of adults said gas prices had declined at least a little over the previous six months, a bright spot for the cost of living. In contrast, in May, 80% had said gas prices were up lot, and 95% said up a little or a lot. That sense of substantial increases has lessened somewhat as of September, with 15% saying prices have been about the same over the last six months, and 5% saying they have gone down a little. Yet 80% say gas prices have risen over the last six months since the start of the Iran war.

Grocery prices are also seen as rising, but more consistently over the year than for gas prices. As of September, 81% say their bill at the checkout counter is higher than six months ago, up from 70% in January. This consistent pressure on grocery prices is a clear source of concern over the cost of living, beyond the recent surge in gas prices.

Adults are also pessimistic about the course of inflation over the next 12 months, as 68% say inflation will increase in the next year. Only 12% are expecting a decrease, and 20% are expecting to see no change. This is a somewhat more negative outlook than in January, when 61% thought inflation would increase over the year.

More people say they are worse off than a year ago, 38%, than say they are better off, 19%, while 43% say they are about the same as a year ago. This two-to-one ratio of worse-to-better is a considerable shift from January, when 28% said they were better off and 34% said worse off.

Iran war

The public remains very doubtful that the United States has achieved its goals in the Iran war, with just 13% saying the goals have been reached and 87% saying they have not been achieved. This is down from 21% in April who thought the goals had been achieved.

One in five, 21%, say the Iran war has been worth the cost, down from 28% who said so in May.

Concluding note

President Trump’s job approval, and views of the economy, provide a bleak outlook for Republicans in the November elections. While the size of the seat swings in Congress may be limited by the reduction in the number of competitive seats since 2010, the national forces of presidential approval, the cost of living and the economy as the most important issues, and the gloomy feelings about the economic outlook, are a heavy burden for Republicans. Democrats have an opportunity, IF they can capitalize on these negative views.

Among all registered voters, Democrats hold a 50% to 42% advantage over Republicans, a larger margin than the 46% to 44% gap in July’s Marquette Law School Poll, as shown in Table 2.

The survey was conducted Sept. 2-9, 2026, interviewing 1,023 adults nationwide, with a margin of error of +/-3.3 percentage points. For registered voters, the sample size is 864, with a margin of error of +/-3.6. For likely voters, the sample size is 581, with a margin of error of +/-4.3 percentage points.

The Democratic advantages among all registered voters stem in part from stronger Democratic performance among independents, including both those who lean to a party and those who say they are not closer to either party. Among independents who lean to the Democratic Party, 90% support the Democratic Party candidate, while among independents who lean to the Republican Party, 83% say they will vote for the Republican congressional candidate and 10% are reluctant to vote for either candidate. Among independents who do not lean to either party, 41% favor the Democratic candidate and 10% the Republican, but 40% say they favor neither and 8% say they won’t vote.

In contrast, both Democratic and Republican partisans support their party’s congressional candidate at least 93% of the time, with very little crossover voting and with few saying they won’t vote, as shown in Table 3.

The substantial independent preference for Democratic congressional candidates is also seen when the six Marquette polls conducted in 2026 are combined, boosting the sample size and showing the September poll is in line with results all year. In the combined sample, 31% of independents favor the Democrat and 14% favor the Republican, while 46% say they prefer neither. The strong partisan loyalty seen in September is nearly the same in the combined polls, and those independents who lean to the Democratic Party report more Democratic votes than those who lean Republican favor the Republican candidate, as shown in Table 4.

Which party is better

The Democratic Party is seen as doing a better job on six of 10 issues. Republicans are seen as better on three issues, and there is a tie on one. Democrats hold the advantage on the federal deficit, foreign policy, the economy, inflation, Medicare & Social Security, and health care. Republicans are seen as better on national defense, crime, and immigration and border security. The two parties are tied on which would handle taxes better, as shown in Table 5.

The Democratic advantage has grown across issues during 2026, while the Republican advantages have decreased. On each issue where Republicans are seen as better, the size of that advantage has declined since January, while each Democratic issue has seen a larger advantage over time. Table 6 shows the changes for each issue in 2026 polls.

Approval of the parties in Congress has also shown similar trends. In September, 36% approve and 64% disapprove of how the Republican Party in Congress handles its job, and identical totals of 36% approve and of 64% disapprove of Democrats in Congress. Both parties have seen substantial majorities disapprove throughout President Donald Trump’s second term. However, there has been a change in the approval ratings of both parties. Over the term of the 119th Congress, Republican approval has fallen from 45% in March 2025 to 36% in September 2026, while Democratic approval has increased from 29% to 36%. Approval of each party is shown in Table 7.

The Republican Party is seen as too conservative by 51% of adults, and the Democratic Party is seen as too liberal by 50%. About a quarter (28%) say the Republican Party is about right ideologically, and 22% say the Democratic Party is about right, as shown in Table 8.

Those who identify with the parties are more likely to say their party is about right for them, with 62% of Republicans and 40% of Democrats saying this. Those independents who lean to either party are somewhat less satisfied with the party ideologically, and a quarter or less of independents say either party is right for them. Among Democrats, 46% say the Democratic Party is too conservative or too moderate for them, while among Republicans, 23% say the Republican Party is too liberal or too moderate. This shows substantially more desire among Democrats for a leftward shift in the party than among Republicans for a rightward shift in theirs. This is shown in Table 9.

The Democratic advantage in congressional vote among likely voters is aided by higher expected turnout among Democrats than among Republicans. Among registered Democrats, 77% say they are certain to vote in November, versus 69% among Republicans. Likewise, among independents who lean Democratic, 67% say they are certain to vote, compared to 60% among independents who lean Republican. Independents who don’t lean to either party are substantially less likely to say they are certain to vote, with just 41% sure they will go to the polls. This lower turnout among “pure” independents reduces their influence in elections below their size among all registered voters. These results appear in Table 10.

Democrats also enjoy an enthusiasm advantage going into the fall elections. Forty-seven percent of Democrats say they are very enthusiastic about voting in the election, compared to 32% of Republicans. There is a more modest enthusiasm gap among independents who lean to a party, with a 29% to 23% advantage for Democratic leaners over Republican leaners. As with certainty of voting, pure independents are by far the least enthusiastic about voting, with just 14% very enthused, as shown in Table 11.

Among Republicans and independents who lean Republican, turnout is highest among those who are very favorable to the Make American Great Again (MAGA) movement, with 82% of them saying they are certain to vote. This falls to 64% for Republicans only somewhat favorable to MAGA and to 52% among those unfavorable to MAGA. This shows that the turnout issue among Republicans comes from those less attached to Trump’s loyal base of MAGA enthusiasts. These results are seen in Table 12.

As with affection for MAGA, turnout among Republicans is strongly related to approval of Trump. Among those who strongly approve of the president, 81% say they are certain to vote, dropping to 61% among Republicans who somewhat approve and to 56% among those Republicans who disapprove of how Trump is handling his job. Among Republicans and independents who lean Republican, 36% strongly approve of Trump, 40% somewhat approve, and 24% disapprove (either somewhat or strongly). Turnout by approval is shown in Table 13.

The Democratic Party has seen fresh competition from democratic socialists in primaries in 2026, creating new divisions within the party. However, these divisions are not associated with differences in turnout, such as are seen among Republicans. There is very little difference in expected turnout among Democrats regardless of how favorable or unfavorable they are to democratic socialist candidates, as shown in Table 14.

Concluding note

The Democrats have a substantial advantage on the generic ballot, especially among likely voters. While party loyalty is similar across parties, independent voters lean heavily in the Democratic direction on congressional vote. Democrats also enjoy a modest turnout advantage, though the lower turnout among independents mutes that groups influence. Enthusiasm also favors the Democrats. Finally, Republicans who are very favorable to MAGA or strongly approve of Trump are very likely to turnout. But for those less positive to MAGA or to Trump, turnout falls substantially. Among Democrats, feelings to democratic socialists are unrelated to turnout.

In Part 2, I turn to approval of Trump and the most important issues for voters.

In the Marquette Law School Poll’s final survey of Wisconsin voters, July 22-27, Francesca Hong received 38%. Her unofficial vote total is 39.3%. And 34% said they were undecided. The remarkable difference between poll and outcome was that the poll had David Crowley at 7% and he won with 39.8%. Mandela Barnes, who dropped out three days after the poll was completed, had 16% in the final survey. One clear thing is that the vast majority of late-deciding voters went to Crowley and very few went to Hong.

The race was complicated by three candidates dropping out, and Crowley reentering, in the final month.

On the demographics of the race, older voters were both the least supportive of Hong and the most undecided age group. They also said they were more likely to vote than did younger voters, providing a potential advantage to Crowley, though they had not settled on him at the time of the final poll.

Another unusual aspect of the race was the very late endorsement of Crowley by Gov. Tony Evers. In our final poll, more than one-third (36%) of Democratic primary voters said that this endorsement made them more likely to support Crowley. Evers’ job approval with Democrats was also over 85%.

Non-response bias is always a potential source of error in polls. In this case, the age and ideological makeup of the survey samples of primary voters was similar to the pre-primary makeup of Democrats, suggesting that young, very liberal, Hong supporters were not overrepresented. Those groups strongly supported her but were not an unusually large share of the samples. Older and moderate voters were more undecided and less supportive of Hong in both of our July polls but had not settled on the very newly returned Crowley at the time of the last survey.

And Hong’s actual vote of 39% was barely different from her 38% in the final poll, further demonstrating that her supporters were not overrepresented in the survey.

Turnout was much larger than in recent Democratic primary contests: an unofficial 792,009, compared to 640,247 in the 2024 Senate primary and 538,857 in the 2018 gubernatorial primary. We don’t know the makeup of that surge beyond the polling results, but this turnout was considerably larger than expected.

On the substantive side, Hong’s performance was generally impressive during most of the campaign, but the late shifts from poll to vote suggest the limits of her appeal.

Finally, in our mid-July poll (July 8-16), 80% of Democratic primary voters said that it was extremely or very important to nominate a candidate who could win against the Republican in November, and electability became an increasingly common topic in the closing weeks of the campaign. And in the mid-July poll, Hong trailed Republican Tom Tiffany by 3 percentage points, whereas then-candidate Barnes led Tiffany by 4 points (this poll was conducted while Crowley was out of the race).

The chaotic changes of candidates, plus the governor’s late endorsement and increasing electability concerns, all made for a surprising endgame to the race.

Note: Below are the tables from the July 29 release showing survey results of the vote and percent undecided, by age and by ideology.

But there was been one stable point in the campaign: Francesca Hong, a democratic socialist, has been in first place in each of five Marquette Law School Poll surveys since Oct. 2025. Her margin has increased in early July, and again in late July.

The big question is will her vote totals follow the polls, or will we see a closer race, as in Michigan? Check in late Tuesday/early Wednesday to see. The polls in July have all agreed on a significant lead, but, well… you know.

My purpose today is not to give an election forecast but rather to show where Hong’s support comes from, and where it doesn’t come from. The data are from my Marquette Law School Poll of July 22-27. This was done after Milwaukee County Executive David Crowley re-entered the race and was endorsed by Gov. Tony Evers, but before former Lt. Gov. Mandela Barnes dropped out on July 28. If that wasn’t enough to make polling uncertain, the campaign has also been late developing, with the last two weeks more intense than the previous two months had been.

The purpose here is to show relative strength for Hong across important groups. The eventual votes may be stronger for Hong or weaker for Hong. In this poll there were a substantial number of undecided voters. For Barnes’ voters, I’ve used their second choices for the vote.

There will be no exit poll, so we won’t get that measure. For now, these are my best estimates. The pattern of support is the same in my early July poll and the late July poll, though Hong gained strength is virtually all groups in the later poll.

Let’s walk through the usual suspects.

Hong support by age

Age is a strong predictor of Hong’s support, winning a substantial majority of 18-29 year olds, and a majority of 30-44 year olds, while slipping with each older age group. In Wisconsin polling since 2012 I don’t think we’ve ever seen as strong an age gradient as this. (In this and all tables, lighter colors indicate stronger support and dark colors indicate less support.)

Hong support by ideology

If age matters a lot, ideology is an even stronger predictor of support for Hong. Moderate and conservative Democratic primary voters are her weakest group, while she passes three-quarters of the vote among very liberal voters. Note the 36% among liberals is the largest vote for any candidate in this group, due to the number of candidates splitting the vote and the undecided.

Hong support by age and ideology

What happens we we combine age and ideology? Hong support ranges from a low of 8% among moderate/conservative and age 60+, to 88% among very liberal 30-44 year olds. Read down the rows and support declines, read across the rows and support rises, with one exception in each direction. Age and ideology largely reinforce each other.

How Democratic primary voters feel about democratic socialists

Hong has been a member of the democratic socialist caucus in the state legislature, and describes herself as a democratic socialist. This year has seen democratic socialists, and other strong progressives, in a number of districts, though not universal success. How do primary voters feel about the label “democratic socialist?” We asked favorability to the term in the early July poll.

By ideology

Democratic socialists are seen favorably by liberal and especially by very liberal Democratic primary voters, and unfavorably by moderate or conservative voters.

By age

There is a substantial age gradient in feelings about democratic socialists, but not so large as for ideology. Peak favorability is actually among those 30-44, and a bit lower among 18-29 year olds. Notably even among those 45-59 favorability reaches 50%. We haven’t measured favorability to democratic socialists before this year, so don’t have comparisons with the past, but this seems a significant development among Democratic primary voters.

Democratic socialist favorability by age and ideology

Combining age and ideology shows these reinforce each other, though the 30-44 year olds are more favorable to democratic socialists than are 18-29 year olds in each ideological group, as we saw in the simple age breakdown above. There is a bit more noise in this table compared to the similar table for Hong support, but broadly favorability rises with both age and ideology, with the exception of the move from 18-29 to 30-44 year olds.

Likely voters by age and ideology

Opinions don’t count if you don’t vote. If age and ideology are the key predictors of Hong support, how do those predict turnout, measured as being a likely voter, measured by saying certain to vote in the primary? This table also uses the early July poll which measured likely voters.

Turnout rises from left to right and from top to bottom. Age has been a strong predictor of turnout since the Big Bang, and clearly is here. Given Hong’s strength with younger voters, and relative weakness among older voters, this is a potential weak spot for her vote if the strong young supporters fail to turn out.

Reading left to right, turnout increases in every row, regardless of age, with the most liberal voters more certain they will vote, with lower turnout among less liberal voters. Here turnout helps Hong due to her especially strong support among the most liberal voters.

Now it’s up to the voters to decide.

We’ll know by Wednesday morning if the polls got the race about right, or not. Then it will be on the general election and a whole new ballgame.

The national economy is the next highest rated, with 15% saying it is the most important issue. The war with Iran is chosen by 10%, a decline from April, when 24% said it was the most important issue. Medicare and Social Security is tied for fourth place with the proposed SAVE America Act dealing with voting procedures, at 9% each. A three-way tie then follows, at 5% each, among health care, immigration and border security, and affordability of housing; the rating of immigration and border security has declined from 14% in January. Rounding out the list, crime is most important to 2% and abortion policy to 1% of adults.

The full list, for each poll this year, is shown below.

The survey was conducted July 22-29, 2026, interviewing 1,076 adults nationwide, with a margin of error of +/-3.2 percentage points.

The importance of economic issues is apparent when considering what issue people consider their second-most important issue. Once more, inflation and the cost of living, and the economy, top the list of second-most important issues, with the Iran war remaining ranked third on the list. Concern for health care moves up to 11% as the second-greatest concern.

Gas, groceries and finances

Changes in multiple measures of economic circumstances since January explain why the cost of living and concern over the economy continue to top the list of most important concerns. Most of these measures also show slight improvements from May to July, although on all measures circumstances are worse than they were in January. An exception to small recent improvement is family financial situation, which worsened from May to July.

Increases in the price of groceries and gasoline are important elements affecting the cost of living, with perceived prices rising over the past six months. Gasoline prices have been especially volatile since the start of the Iran war. In January, prior to the start of the war, 50% said gas prices had gone down either a lot or a little, and only 7% said they had gone up a lot. This changed dramatically in April, with 75% saying gas prices had gone up a lot, and only 2% thinking they had gone down. Perception of rising gas prices peaked in May, with 80% saying they were up a lot. In July, a smaller percentage of people (57%) said prices were up a lot over the last six months, with 11% saying the prices were stable and 8% saying prices had declined. These perceptions track with average prices reported by AAA, which show prices declining from May through early July, though with a new surge in prices as this survey was commencing interviews.

Perception of grocery prices has not been as volatile as for gasoline, but in July, 48% say grocery prices are up a lot over the last 6 months, with 31% saying they are up a little, 15% saying they are stable and 6% thinking they have gone down.

In total on these cost-of-living measures, 81% say the price at the pump is up a little or a lot, and 79% say the price of groceries at the checkout line is up a little or a lot.

The public has pessimistic expectations of inflation over the next 12 months, with 69% saying inflation will increase, 19% saying it will remain the same, and 13% expecting a decline. Over the course of 2026, the percentage expecting an increase in inflation rose from January through May, with a slight reduction in July.

The percentage saying they are better off than a year ago has declined from 28% in January to 19% in July, with those saying their circumstances are about the same as a year ago rising from 37% to 44%. Those saying they are worse off show a small increase from 34% to 37%.

People’s description of their family’s financial situation took a negative turn in July, with the percent saying they are living comfortably declining 10 percentage points from 43% in June to 33% in late July. The total is down 7 percentage points since January. Those saying they are struggling also increased to 21% in July, up 6 points from both June and January. Those just getting by has remained relatively stable over Marquette Law School Poll surveys this year, with 46% just getting by in July.

Views of the national economy improved slightly from May to July, though they are more negative than in January. In July, 28% say the economy is excellent or good, up from 26% in May but down from 35% in January. Those saying the economy is not so good rose by 1 percentage point from 45% in May to 46% in July. That is up from 40% in January. Those saying the economy is poor rose from January through May, but declined 4 points to 26% in July.

Concern for health care is well down the list of most important problems, though it rises to fourth place among the second-most important problems. The expiration of federal tax credits for insurance purchased through Affordable Care Act exchanges is unpopular with the public. Sixty-nine percent say the tax credits should have been extended, while 31% say Congress should have let them expire. Opinion on this has been stable since January.

Beyond economic concerns?

There is more than just the economy. The Iran war has declined in importance, from 24% most important in March to 10% now. But views of the war are exceptionally negative. I’ll cover that in a separate post.

And I haven’t mentioned Trump’s approval numbers. Hint: not good on economic issues. Also deserves a post of its own.

And one final look at the Wisconsin primary before Tuesday.

It is a busy August. I hope you can take a little time to relax.

“it had been a long and eventful day.” busy, event-filled, action-packed, full, lively, active, hectic, strenuous; momentous, significant, noteworthy, notable, remarkable, outstanding, important, crucial, critical, historic, consequential, fateful, decisive. ANTONYMS dull, uneventful.

Or the term used a lot in Wisconsin, but omitted above: chaotic.

Democratic socialist Francesca Hong has led in each of five Marquette Law School polls since October, though with very high undecideds until recently. Barnes was consistently in 2nd place, followed by Rodriguez, until she dropped out. Crowley in now in second place, after Barnes left.

Here is the trajectory of candidates in the Marquette polls, omitting the percent undecided, which we’ll get to. Saying Hong has led every poll is true, but at very small percentages until July, when her advantage has become substantial. Barnes made modest gains in percentage support, but gained less than Hong from poll to poll. Before dropping out, Rodriguez made some gains, but remained at just 11% the week she left the race. Other candidates, including Crowley, showed little movement. Crowley pops up to 7% after returning to the race, in the Marquette poll. He is higher in two polls taken after Barnes left the race, which we’ll get to below.

My main point from this chart is that Hong’s campaign has succeeded in building support through the year to an extent that no other campaign matched. That is especially notable because Barnes was much better known, having run for U.S. Senate in 2022, losing by 1 point to Sen. Ron Johnson. Barnes also had a bigger fundraising total than the other candidates, though he burned through most of it, ending with less cash on hand than most other candidates. Despite those advantages, the chart above shows he made only modest progress in winning support from voters.

Democratic candidate favorability

Another way to judge the success of a campaign is the extent to which it builds name ID and favorability toward the candidate. As noted, Barnes began the race with the lowest percentage saying they hadn’t heard enough to give a favorable or unfavorable opinion. His campaign was unable to take advantage of that. Over the campaign the percentage unable to rate him barely budged. Likewise is favorable and unfavorable ratings hardly moved.

In contrast, Hong began with over 75% unable to rate her, but by the early July poll she had brought that down to under 50%, the most striking success of any candidate in the field. Rodriguez also gained name ID, but not as much as Hong. The week after reentering the race, Crowley continued to have a high percentage not familiar with him, something that is likely changing in these final two weeks of the campaign, given the endorsement by Evers and a last minute ad campaign.

Hong did not just become more familiar. The percentage of Democratic primary voters with a favorable opinion of her rose sharply over the campaign, while her unfavorables only inched up. Another mark of a successful campaign. (Also a mark of the reluctance of her opponents to campaign against her, at least until this week. Even in the one debate, opponents did little to raise criticism of her.)

The chart below shows how each candidate’s favorables moved, and the percent saying they hadn’t heard enough.

The undecided

Wisconsin primary voters have a history of taking a long time to make up their minds. In six primary elections covered by the Marquette Law School Poll, undecided have remained high into July (with two exceptions from 2012.). The 2026 primary has had a higher percentage undecided until the final Marquette poll, ending July 27. In that poll, 34% remain undecided, but that is in line with four previous primaries, include the 2018 Democratic primary for governor, which has a similarly large number of candidates.

This year’s primary has had remarkably little television advertising until very late, and a single debate on July 28th, though there have been a large number of candidate forums.

Only at the end does the share of undecided fall to the “normal” level of most recent primaries, as shown in the chart below.

All the late polls

As promised, here is a look at all the polls from July and August that included the full field of candidates (at least as of the time the poll was conducted.) As with the Marquette polls, the full set show Hong with a significant advantage, and a rising percentage share of votes. The chaotic shift of candidates over the month makes it hard to know the effect of Rodriguez and Barnes leaving the race and Crowley re-entering. The reentry also came so late that it is a sprint of just over 3 weeks to election day, which is not much time to reach voters, even with a concentrated burst of ads and events. The latest, and I assume last, pre-primary poll, by State Navigator, has Hong at 44%, Crowley at 22%, Roys at 11% and Brennan at 5%. Barnes and Rodriguez remain on the ballot, and on early absentees cast before their exits. Barnes is estimated at 6% and Rodriguez at 3% in the State Navigator poll, with 9% still undecided.

The Crowley campaign released an internal poll by PPP, showing Hong at 42% and Crowley at 24% but did not report the results for other candidates. The Crowley campaign memo argues polls, including their own, have too few older voters, and that when weighted to the campaign’s estimate of the primary’s likely age distribution Hong has 36% and Crowley 27%.

The chart below shows all the polling, but does not include the Crowley campaign’s readjustment for age. See the polls at NYTimes.

Waiting for Tuesday night

The question keeping pollsters awake is “will this be like Michigan” with a much tighter race than the polling shows? As the chart above shows, the polling across different organizations is consistent in showing a decent Hong lead, with reasonable uncertainty about how the very late changes of candidates may ultimately affect the vote for Crowley and perhaps for Roys and Brennan. The math says that Hong doesn’t have a majority, but to catch her there would have to be a coalescence of support for one alternative candidate, and/or a substantial shortfall in turnout among Hong supporters.

Michigan proves polls can be wrong, or the electorate can shift in the closing days. But that caution should not miss the evidence here that Hong has run the most successful campaign of all the candidates, building support, name ID, and favorability over the year to a degree no one else matched.

I’ll close the way I close each final Marquette Law School Poll pre-election presentation. (And yes, I do own that pair of socks.)

I released my Marquette Law School Poll on Wednesday July 29th. That included Barnes. Now here are those results allocating Barnes’ supporters to their 2nd choices. It moves Hong up a little more, though 41% of Barnes’ voters didn’t have a 2nd choice.

As for where his supporters go, most to Francesca Hong, a few to David Crowley and the rest.

My this has been a curious race. When he got in the race I was told by a lot of Democrats that with his name ID and fundraising lists from the 2022 Senate race, Barnes would run away with the nomination. Yet he never was ahead in our 4 polls since he entered the race.

In the late 1950s and early 1960s, Johnny Carson, paired with Ed McMahon, hosted the TV game show “Who Do You Trust?” Aside from uniting the legendary late-night duo, the title poses a question as relevant today as it was then. Society rests on connections between people, and trust is one vital element in social bonds of cooperation, business, friendship, and family. In 2026 America, 64 years after Carson and McMahon moved to “The Tonight Show,” whom do Americans trust and how much trust do we have in each other?

The Marquette Law School Poll national survey has dived into the question of trust in the age of cell phones, the internet, social media and dating apps—long past the age of black and white television.

In four polls conducted with 4,500 respondents in 2026, 77% of those born before 1960 say most people can be trusted, a markedly high level of interpersonal trust. Contrast that with those born in the 2000s, among whom only 35% say most people can be trusted. This dramatic difference is no fluke. Trust steadily declines according to age cohort, as shown in Figure 1, with the numerical data shown in Table 1.

There has been relatively little change in overall trust in Marquette Law School polling since 2021, with trust ranging from 51% to 55%.

While there has been little change in overall trust, the youngest birth cohort has shown a notable increase in trust. Trust among those born in the 2000s increased from 23% in 2021 to 35% in 2026.

Each cohort shows at least some increase in trust from 2025 to 2026, and the 1990s cohort has increased in trust during each of the past three years. None of these shifts has changed the relative ordering of cohorts by level of trust, and only the two youngest cohorts have moved up consistently in the last several years. Despite the recent rise among the youngest cohort (those born in the 2000s), their level of trust has not risen to match that of their predecessor, the cohort born in the 1990s. No younger cohort, as of 2026, has trust as high as their predecessor, although those born in the 1980s and 1990s are close to each other. The trends for cohorts by year are shown in Figure 2.

It is tempting to reach for a handy explanation for the decline in trust, but the fact that trust has declined with each new cohort makes recent changes, such as the growth of social media, not generally persuasive. It is true that heavy users of social media are less trusting, and they are also younger and so a larger proportion of more recent birth cohorts. In 2026, among those who used one social media platform or none, 59% say most people can be trusted, while among those who use five or more social media platforms, 46% say most people can be trusted. This is a significant gap but far smaller than the gap between the oldest and youngest birth cohorts.

It is also true that the young are much heavier users of social media. Only 6% of the oldest cohort use five or more social media platforms, while 32% of the youngest cohort use that many. At the other end of the scale, 41% of the oldest cohort use one or no social media apps, while just 15% of the 2000s cohort use so little. Is it birth cohort or social media use that plays the dominant role in trust?

A closer look at trust by social media use within each birth cohort finds that there is little consistent effect of social media use within each cohort. Members of the oldest cohort are the most trusting on average no matter how much social media they consume. The youngest cohort is the least trusting, regardless of social media immersion. The modest variation within each cohort is inconsistent rather than showing clear trends across social media use; this is shown in Figure 3.

The interpersonal trust gap between the old and the young carries over to more specific questions about trust in people. Among all adults in June 2026, 59% say most people would “try to take advantage of you if they got a chance,” while 41% think most would “try to be fair.” A majority of each cohort except those born before 1960 think most people would try to take advantage, and that majority grows with each younger cohort.

The public is even more inclined to say people just look out for themselves, 64%, rather than saying people try to help others, 36%. Only a small majority of the pre-1960 cohort think people try to help others, and the majority saying people look out for themselves grows to 76% in the youngest cohort.

Despite these majorities who doubt others, a majority of adults say they begin with a presumption of trustworthiness until a person proves otherwise (59%), rather than beginning by not trusting people until they prove they can be trusted (41%). On this question, the majority for a presumption of trust shrinks from the pre-1960s cohort to that of the 1980s, then becomes a minority view among those born in the 1990s and 2000s.

Trust in specific types of people

While generalized trust was low on most questions above, there is large variation in trust in specific types of people. Trust is higher in people with whom we have an ongoing relationship, and highest in family and coworkers. Trust is lowest in strangers with unknown or ulterior motives, such as a stranger who knocks on the door or a telemarketer on the phone.

Most of the types of people who are trusted a great deal or a fair amount are involved in established relationships of varying degrees of closeness. Family stands apart at the top of the list, but 70% or more have a high level of trust in coworkers and neighbors, as well as in “teachers in your local schools” and financial advisers. Teachers represent a role rather than specifically the teacher of a parent’s child. The next tier of trust includes fellow worshipers, businesspeople you buy from, leaders of your religious institution, and casual acquaintances seen a few times a year. Each involves some level of voluntary association and repeated interactions.

Trust is lower for more distant people, especially those whose motives may not be known. Candidates for local offices are more trusted than state elected officials, who in turn are more trusted than federal elected officials. A stranger who strikes up a conversation is not very well trusted but is more trusted than a stranger who knocks on one’s door, or especially a stranger who phones to sell something. Trust in this range of types of people is shown in Table 10.

What about variation in trust in each of these types of people, by age cohort? Do the differences in generalized trust found above also carry over to these specific types of people? Yes: In most cases, the younger cohorts are less trusting than the older cohorts, with only modest exceptions.

The decline in trust from oldest to youngest cohort is clear and substantial among all types of people who are rated most trustworthy overall. This includes family, where the youngest cohort is 15 points lower in trust than the oldest. But these differences are even larger for coworkers, neighbors, financial advisers, co-worshipers, business relationships, religious leaders, and casual acquaintances.

Among the types of people generally rated less trustworthy, differences across cohorts persist, with a small exception for the three least trustworthy categories: federal elected officials, strangers knocking on doors, and telemarketers. In these groups with especially low overall trust, the youngest cohort has slightly higher trust than their elders, though trust is quite low for all cohorts.

The decline in trust also applies to each succeeding cohort, from the oldest to the youngest, with only a few minor exceptions. For an example of an exception, the 1990s cohort is more trusting in coworkers than the 1980s cohort, but the 2000s cohort is less trusting than either. The strong and consistent pattern is that across all these types of people, trust consistently declines as we move from older to younger cohorts. The full pattern is shown in Table 11.

Trust consistently declines as we move from older to younger people across these numerous measures of trust—in people in general, in presumptions about trustworthiness, and in 15 specific types of people. The Marquette Law School Poll also found that the handy explanation of social media use doesn’t stand up to empirical examination. Within each cohort, use of social media is at best inconsistently related to trust, and there is no systematic decline in trust as social media use increases. We are left without a ready answer to the question of what has changed since the 1950s that has produced succeeding generations who have substantially less trust in other people. But the consistency of the decline shows that the newest cohorts are not aberrations. Changes over the past 70 years appear to have systematically reduced trust in people in general and in our friends and neighbors, just as in those whose motives most people don’t trust.

On birthright citizenship, 68% of adults say the 14th Amendment makes all those born in the United States citizens, while 32% say an executive order can limit citizenship. Public opinion in each of these cases was measured in my Marquette Law School Poll national survey May 20-26, 2026, N=1,001 adults.)

The public favors state laws that prohibit transgender girls and women from participating on girls’ and women’s sports teams, 63%, with 37% saying the laws violate the equal protection clause of the 14th Amendment.

The public disagrees with the Court on allowing parties to coordinate with candidates and provide unlimited financial contributions by parties to candidates. A substantial 74% say limiting coordination and contributions is an appropriate regulation of campaign finance, while 26% say it is a violation of the 1st Amendment’s free speech rights.

The Marquette Law School Poll national survey will revisit these and other cases in July, following the rulings to measure reaction to the Court’s decisions and assess public approval of the Court.

For public views on removal powers over the Federal Reserve and independent agencies, and on counting late arriving ballots, see yesterday’s post:

In a Marquette Law School Poll national survey, a 56%-44% majority favors counting of late ballots. (All results are from the Marquette Law poll done 5/20-26/2026, 1,001 adults.)

On the Federal Reserve, 66% say the president may not remove members, while 34% say he has that authority.

An even larger percentage disagree with the Court on firing heads of independent agencies, with 73% saying statutes can set the requirements for removal, while 27% take the Court’s position that the president’s executive authority allows removal.

While carving out an exception for the Federal Reserve, the independent agencies decision expands the power of the president vs Congress in control of regulatory agencies.

Still to come this week are rulings on birthright citizenship, transgender athletes, and campaign coordination between parties and candidates.

Here is a look at institutional trust, measured by confidence in 18 institutions among adults in my June 8-15 national survey.

Among all adults, your doctor tops the ranking by net confidence (the percent confident minus percent not confident.) Second is the U.S. Military, followed by the police. Others with net positive ratings are local schools, colleges and universities, and the people who administer elections in your state. Those six are the only net positive, the other 12 are various degrees of negativity.

Only mildly negative net ratings are given to local news media, the U.S. Supreme Court, and the FDA. All others are negative by double digits.

Confidence in our fellow citizens in electing leaders is quite low, at -29, reflecting bi-partisan negativity.

The bottom four are perhaps not surprising: national news media, Congress, AI companies, and last of all, Facebook.

That Facebook rating is interesting. Not only is it exceptionally negative (worse than Congress!) but Facebook is the second most used social media platform with 71% who say they have used it in the last week. YouTube is first. But among those Facebook users, net confidence is also very negative: -60. I guess the algorithm is working, even if we don’t trust it.

Of course we don’t all agree on who we trust. Party plays a role in that. You’ll not be surprised that Republicans have confident in the Supreme Court and the executive branch while Democrats don’t. Or that Democrats have confidence in state election officials, colleges and universities, and local news media, while Republicans don’t.

But it may be more notable how much agreement across parties exists. Doctors, the military and police get positive ratings from both parties. And the very negative ratings of Congress, AI companies and Facebook are also bipartisan. Others are more closely divided for one party or the other.

Given how polarized we are over policy and the president, it is notably that institutional confidence is actually positively correlated by party, a correlation of +.47. Far from perfect agreement, but on policy issues correlations are usually substantially negative.

I’ll leave it as a bit of a Rorschach test for you to decide for yourself what patterns you see in the chart below, and what conclusions you reach from that.

A quick review of what my Marquette Law School Poll national surveys find about opinion of Democratic Socialists of America among Democrats and independents who lean to the Democratic Party. A question for future research is if people react differently to “Democratic socialists” than to “Democratic Socialists of America”.

The data pool the April and May Marquette Law School Poll national surveys.

The DSA question is

[Democratic Socialists of America] Do you have a favorable or an unfavorable opinion of the following or haven’t you heard enough yet to have an opinion?

All tables are for Democrats and Democratic leaners only.

Topline favorability

Note a lot of Democrats are unfamiliar with DSA. I use those favorable to DSA and lump both unfavorable and haven’t heard enough into “non-DSA” Democrats.

By this measure, 40% of Democrats are also favorable to DSA, while 20% are unfavorable and 40% say they haven’t heard enough to have an opinion.

DSA favorability by age, among Democrats

Young Democrats are more favorable, but so are 45-59s. The age gradient here is not very strong across age groups.

DSA favorability, by ideology among Democrats

The big supporters of DSA are Democrats who call themselves “very liberal”. There are few conservative Dems so I’ve combined them with moderates for this table.

A sizable majority of very liberal Democrats are favorable to DSA, and somewhat more likely to have an opinion of them than either moderate or somewhat liberal Democrats.

DSA favorability, by urban-rural residence among Democrats

Favorability to DSA is only a little greater among Democrats living in central cities than elsewhere, and those in the suburbs of metro areas but outside the counties containing central cities, are at least as favorable as those in the central cities. Rural residents are less positive, but not dramatically so.

DSA favorability, by ideology and age among Democrats

The very liberal people under 45 look like the older very liberal Democrats, though there are fewer very liberal older respondents. To provide more cases for comparison age is collaped into two categories.

As this table shows, it is more ideology than age that drives DSA opinion.

Ideology by age among Democrats

Just to close the loop, the young are more likely to describe themselves as “very liberal” by a smidgen compared to 30-44 and by a lot compared to those 45 and over. Moderation is higher among the older Dems and less among the young.

To the extent the young are more positive to DSA it has more to do with the number of very liberal Democrats under 45 than to age per se.

Addendum: DSA favorability among all adults

Among all adults, 20% are favorable to DSA, half the rate as among Democrats above.)

Across party lines there is an expected gradient. Independents are not very favorable, while independents who lean Democratic are almost as favorable as are Democrats. Note the table above combines Democrats and leaners, so Democrats here do not match the combined Dems+leaners in the tables above.

The ideological gradient among Democrats shows up strongly among all adults as well. Here there is a sizable gap between somewhat liberal and very liberal, which reinforces my point above that the DSA appeals is closely tied to thinking of oneself as especially liberal, a +45 net favorable versus a +10 net favorable among those somewhat liberal.

A reminder of the topline distribution of ideological self-placement among adults.

Very liberal and very conservative groups are almost identical in size, but somewhat conservative outweighs somewhat liberal, a over a third call themselves moderate.

So let’s close with ideology by party for that context as well.

Independents are solidly moderate, and independents who lean Democratic are almost 50% moderate. Very liberal Democrats are not quite a quarter of Democrats, with about a third moderate and a third somewhat liberal.

Republicans, whether partisan or leaners, are more solidly conservative, with fewer moderates than the corresponding Democratic groups. And for partisan Republicans, those very conservative is notably larger than those partisan Democrats who are very liberal.

If I have to end with a moral of the story it would be don’t judge the political environment, or the Democratic Party, from the vote in Manhattan. The DSA performance there is genuinely impressive, and a wing of the party is developing strength. But don’t let that dominate how you understand the party nationally. As I’ve posted in the past, the liberal and very liberal wing of the Democrats have grown over the past decade, but to a quarter of Democrats, which is still less than the size of moderates in the party. It is a heterogeneous party, and therein lies the uncertainty about where it is headed.

As of the latest polls, 68% say the 14th Amendment makes all those born in the U.S. citizens. Sixty-six percent say the president may not remove members of the Federal Reserve. A larger majority, 73% say the Court should rule that statutes can set requirements for removal of heads of independent agencies, rather than allow the president to remove them under his executive powers. A NA% majority say the Court should rule that state laws prohibiting transgender athletes from women’s teams are constitutional. On ballots arriving after election day, but postmarked by election day, 56% say the court should allow such ballots to be counted. On the ability of political parties to coordinate with candidates and make unlimited contributions to those candidates, a 74% majority say the Court should uphold regulations that limit such coordination and contributions.

In the next few days, we’ll see if the Court agrees or disagrees with the public on these cases.

The tables below show the questions asked about each of these cases and the results over time, if an item has been asked more than once. The question wording on birthright citizenship was substantially changed from November when the briefs were filed to April when the case was argued, reflecting the changes in presentation of the case. Other cases have wording changes that are shifts of tense from before to after the case was argued. The full wording for each question in each survey is provided for clarity of the wordings.

Links to SCOTUSBlog for the description of the cases and other details are provided for each case.

Birthright citizenship, Trump v. Barbara

Pending, SCOTUSBlog

November wording:

The Supreme Court is expected to hear arguments asking whether the 14th Amendment’s provision that those “born in the United States, and subject to the jurisdiction thereof,” are U.S. citizens was intended to only apply to newly freed slaves after the Civil War and should not apply to a non-citizen‘s child who is born in the United States today. How do you think the Supreme Court should rule?

April & May wording:

The Supreme Court heard arguments on April 1 asking whether an executive order can legally declare that only children born to at least one citizen parent or to lawful permanent residents are citizens of the United States or whether the order is unconstitutional because the 14th Amendment makes all those born in the United States citizens. How do you think the Supreme Court should rule?

Removal of Federal Reserve Governor, Trump v. Cook

Pending, SCOTUSBlog

April wording:

The Supreme Court is hearing arguments in a case asking whether President Trump can remove a member of the Federal Reserve Board of Governors. A lower court ruling prevented the removal, which the Supreme Court is now considering. How do you think the Supreme Court should rule?

May wording:

The Supreme Court has heard arguments in a case asking whether President Trump can remove a member of the Federal Reserve Board of Governors. A lower court ruling prevented the removal, which the Supreme Court is now considering. How do you think the Supreme Court should rule?

Removal of heads of independent agencies, Trump v. Slaughter

Pending, SCOTUSBlog

November wording:

By federal statute, individuals appointed to serve on independent, multi-member regulatory agencies, like the Federal Trade Commission, can be removed only for “inefficiency, neglect of duty, or malfeasance in office.” The Supreme Court will hear arguments in a case asking whether such Congressional requirements for removal infringe on the president’s executive power. How do you think the Supreme Court should rule?

May wording:

By federal statute, individuals appointed to serve on independent, multi-member regulatory agencies, like the Federal Trade Commission, can be removed only for “inefficiency, neglect of duty, or malfeasance in office.” The Supreme Court has heard arguments in a case asking whether such Congressional requirements for removal infringe on the president’s executive power. How do you think the Supreme Court should rule?

Transgender athletes, West Virginia v. B.P.J. and Little v. Hecox

Pending, SCOTUSBlog

September wording:

The Supreme Court will hear arguments in a case asking whether state laws that prohibit transgender girls and women from participating on girls’ and women’s sports teams violate the Constitution’s equal protection clause. How do you think the Supreme Court should rule?

January and May wording:

The Supreme Court heard arguments in a case asking whether state laws that prohibit transgender girls and women from participating on girls’ and women’s sports teams violate the Constitution’s equal protection clause. How do you think the Supreme Court should rule?

Late-arriving absentee ballots, Watson v. R.N.C.

Pending, SCOTUSBlog

April and May wording:

The Supreme Court heard arguments in a case asking whether to uphold a state law that allows absentee ballots to be counted that are postmarked by election day, but that arrive up to five days after the election, or whether this state law conflicts with the federal law setting the date of elections. Should the Supreme Court uphold the state law to count late-arriving ballots, or should it strike down the state law because it is not consistent with federal law?

Campaign finance, N.R.S.C. v. F.E.C.

Pending, SCOTUSBlog

May wording:

In December, the Supreme Court heard arguments challenging a federal law limiting the amount of money that political parties can spend in coordination with a candidate for office. The challengers argue that the limits violate First Amendment free speech rights. Supporters argue that the law appropriately ensures that contributors cannot use donations to parties to evade limits on contributions directly to candidates.

Patrick Ruffini published on his The Intersection Substack a nice reaction on “unlikely voters”, using my table and having Claude estimate the results for people who are either not registered or registered but not likely to vote.

That prompted me to rerun the tables with added columns that capture various definitions of “unlikely” voters. The first is what I think Patrick was aiming for, people who either aren’t registered or who are registered but not likely to vote. I label these “Unlikely.”

But there are other interesting groups.

Those who are “not registered” are folks who conceivably could register but are generally not very involved in politics. Trump mobilized some of these folks in his campaigns, converting them into registered voters. What does this pool look like now?

Then there are the “not likely” voters— those who are registered but say they may not vote. These are different from “unlikely” above because they are all registered, but they are less sure they will vote than are likely voters.

So with these three additional columns, here are the comparisons across samples. The columns are adults first for comparison, then in order of attention to politics with the not registered, who are least attentive, then unlikely followed by not likely. Last are registered and likely voters, who are the most politically engaged subsamples.

The major takeaways (based on the pooled sample in the second table below) are that the unlikely voters, by any of the three definitions, are less engaged in politics and considerably more independent than adults or especially RVs or LVs. They are considerably more likely to say they would not vote for either party in the congressional elections. They are more disapproving of Trump. They aren’t consistently different in their opinions of the congressional parties or favorability to the parties, though they are more likely to lack an opinion of either party.

They are more likely to say the country is on the wrong track. They are less likely to say they are better off than a year ago, but not much different in saying they are worse off.

They are slightly more concerned with inflation and the cost of living than others and less concerned about the war with Iran.

If they are Republicans they are less MAGA.

(Zoom in on a larger monitor recommended.) These data are from the May Marquette Law School Poll national survey, N=1001.

To avoid using only a single sample for this comparison, I’ve also pooled three Marquette Law School national surveys from January, April and May 2026 to boost the sample size to 2986 and reduce idiosyncratic variation in a single month’s poll. Those results are quite similar, as shown below.

My thanks to Patrick for his comments and for provoking me the run these additional data to look at “the unlikely voter.”

This chart shows data from Pew on satisfaction with how things are going in the country. The partisan differences are obvious and change abruptly with changes in the presidency. A minor note is also that the change happens immediately after the election, without waiting for the inauguration to actually install the new president.

The partisan effect is very clear. The thing I want to point out is how this changed in Bush’s second term and how very low and very flat satisfaction has been among the out-party in the Obama, Trump 1, Biden and Trump 2 years.

In the Clinton years, satisfaction was higher with Democrats than Republicans, but the lines move in parallel though the 8 years Clinton was in office. In the second term Republican satisfaction was higher than it had been for Democrats in the first term.

In Bush’s first term Democratic satisfaction fell from the Clinton years but remained well above 20% until late in the first term, then stayed in the teens throughout the second term. In part that reflected the spike following 9/11 but satisfaction declined relatively slowly through the first term. GOP satisfaction remained high in the first term. In the second term, Republicans reacted to the Afghanistan and Iraq wars and the onset of the Great Recession with a substantial decline in their satisfaction.

In each administration since Bush, the out-party had held satisfaction in the teens throughout, regardless of what happened during the administration. Nothing Obama did, or regardless of events during his term, Republican satisfaction remained in the low teens. For Trump’s first term, Democrats remained in the low teens, and for Biden Republican satisfaction always held at about 10%. Now in Trump 2 it is the Democrats turn to be extremely dissatisfied.

This suggests that in the Clinton years and first four years of Bush, there was some common response across party lines to national conditions. The out-party was less satisfied than the in-party, but both moved together up or down. That broke down in Bush’s second term and has never returned since.

The in-party does show some meaningful variation, with Democratic satisfaction rising through Obama’s two terms, as the economy recovered from the Great Recession, and Republican satisfaction drops with the Covid pandemic. In Biden’s term, initially high Democratic satisfaction declined as inflation rose in late 2021 through 2022, recovering only modestly. And so far in Trump 2 GOP satisfaction is high but has been trending down. None of this moved the out-party in those administrations.

V.O. Key’s classic The Responsible Electorate argued voters were not perfectly informed rational actors, steeped in policy, but instead were “not fools”. Voters, Key said, rewarded good performance and punished bad performance. The modern trends shown here since 2009 (and perhaps since 2006) show that for the out-party there is no reward, only stolid unhappiness. That leaves independents, perhaps, to respond to performance, and the in-party can express disappointment, but a major chunk of the electorate is no longer offering any rewards.

The picture is not quite so bleak if we turn from the direction of the country to reactions to the economy. Here too there are large in-party vs out-party differences, but a bit more parallel movement even in recent administrations.

The University of Michigan Index of Consumer Sentiment is a monthly measure of economic perceptions of the current and future economy. Their data by party is only available since 2006 so offers no details of the Clinton administration or the first term of Bush, but the data otherwise gives a good look during the recent period when satisfaction with the direction of the country has remained at rock bottom among out-partisans.

As with satisfaction, there are large partisan differences, and these change abruptly with each change in administration, as does satisfaction. It is notable that no “real” economic conditions change this abruptly on Election Day or on Inauguration Day, so this is a good measure of how powerfully partisanship distorts our perception of “reality.” It is also clear that these shifts begin before the new president takes office.

What is a bit different here, however, is that the partisan lines (and independents too) continue to move in rough parallel across all these administrations, while satisfaction stopped doing that in 2006 or so. The late Bush second term, when these data begin, shows parallel declines. Through Obama’s two terms all partisans showed increasing sentiment. In the pre-Covid Trump 1 term, the lines are each fairly flat and relatively high, then when Covid arrives all show a sharp drop. In Biden’s term Democrats, independents and Republicans all become more negative about the economy as inflation rises to its peak in June 2022, then all three groups become more positive in most of the remainder of Biden’s term. With Trump 2 so far, independents and Democrats are quite negative while Republicans have remained quite positive, which violates the general parallelism earlier.

This economic data shows that Key’s electorate as the god of reward and punishment may not be dead, at least with respect to the economy. All do seem to respond in similar ways though at different levels. Whether this translates into votes is a topic for another day, but similar responsiveness is a precondition for exacting reward or punishment.

It may be that questions about satisfaction with how things are going in the country are simply so saturated by politics now that they elicit a more explicitly partisan response than was the case in the 1990s and before. A similar trend holds for the “right direction or wrong track” question, a favorite of political pollsters who held it to be a great indicator of whether an administration was in good shape or bad shape. With large majorities saying “wrong track” for administrations since Bush, we see the same out-party nearly all saying wrong track and large partisan divides.

As political division has overrun broader considerations of what it means to be “satisfied with the direction of the country”, we are deeply divided indeed. The remaining responsiveness to economic conditions at least suggests the public may yet respond to good or bad times, if less so that they did in Key’s day, or in the late 1900s. For the 2000s, that responsiveness is at least lessened by universal disapprobation by the out-party.

(Thanks to Pew for making the satisfaction data available. See their report on the sour mood of the country here.)