The trend has raised questions from critics about how USDA leaders are making decisions with huge implications for struggling farmers, food stamp recipients and workers in dangerous meatpacking jobs, among other aspects of America’s food system.

“They operate much more on anecdote and ideology than facts and data,” said Rep. Chellie Pingree (D-Maine), a member of the House Agriculture Committee. “I’ve seen a dramatic shift with this administration using less reliance on data, less interest in talking about data, or completely ignoring it when the facts don’t go their way.”

When USDA rolled out a proposal in July to crack down on eligibility for food stamps, there was a key figure absent from the Trump administration’s formal analysis of the rule: how many low-income kids would lose automatic access to free school meals. Lawmakers hounded USDA officials for months to track down those figures, which turned out to be twice as high as USDA initially indicated.

In June of last year, the department’s internal watchdog launched an investigation into whether officials used flawed data to support a new rule allowing meatpackers to accelerate their pork processing lines to high speeds that could endanger plant workers.

And agricultural economists have challenged the calculations USDA used to structure its $20 billion-and-counting trade bailout for farmers, which has been criticized for paying too much to some farms.

Those cases and others reviewed by POLITICO highlight a pattern of questions surrounding the data and analysis behind many of the department’s most ambitious policy moves. The trend has fueled complaints from members of Congress who feel left in the dark, and it’s fed criticism that Perdue and his top officials are making political decisions first and gathering the relevant facts later, according to lawmakers, agricultural research experts and former USDA staff.

“The administration has made moves to reduce the amount of evidence that enters into the policymaking process,” said Rebecca Boehm, an economist with the nonpartisan Union of Concerned Scientists. “It’s obviously political, and special interests come into it. But bottom line is the public loses, farmers lose.”

While every president faces scrutiny over how they set federal policies, and whether those actions stem from objective analysis or political objectives, experts suggest President Donald Trump has ushered in a new era of fact-free decision-making, a sharp departure from previous administrations of either party.

The Center for Science and Democracy, part of the Union of Concerned Scientists, published a study in 2018 analyzing violations of “scientific integrity” under each administration dating back to the 1950s. The researchers concluded that “the Trump administration’s actions reflect a new evolution and escalation” of disregard for science and in some cases were “unprecedented.”

For his part, Perdue has frequently talked up the need for “sound science” at USDA, telling POLITICO last year that “we’re very serious when we say we’re fact-based, data-driven decision makers.”

A department spokesperson reiterated that point, claiming the secretary “has emphasized the importance of ensuring USDA is facts-based and data-driven, especially when creating and developing policies. To achieve this, the department not only relies on data and science from within our agencies … but has also worked to improve data integration so we can measure decisions and outcomes against clear performance standards.”

But longtime agriculture policy watchers say USDA’s actions under Trump don’t match its rhetoric.

“If this administration wants to be transparent and use evidence-based policy, then what we’re seeing at USDA seems not to be in line with that stance,” said Susan Offutt, who led the department’s Economic Research Service for a decade under the Clinton and George W. Bush administrations.

Jobs on the chopping block

Perdue’s short-lived plan to end the longstanding Forest Service job training program left lawmakers on both sides scratching their heads about how the decision was made.

The centers train low-income youth to respond to natural disasters, maintain national forests and work on rural infrastructure projects. Perdue wanted to hand them over to the Labor Department, which already oversees a much larger number of job training sites.

But the move entailed shuttering nine facilities in rural districts across the country and potentially laying off some 1,100 workers — a deal-breaker even for normally supportive members of Congress.

USDA’s hasty rollout didn’t help, either. Lawmakers said they weren’t briefed in advance of the May announcement, and the chief of the Forest Service told her staff she was given just four days’ notice.

After the rocky launch, Perdue’s attempts to justify the changes to Congress fell flat, as Republicans from Newhouse to Senate Majority Leader Mitch McConnell lobbied the administration to back down.

In an interview earlier this month, Newhouse praised Perdue for listening to lawmakers and changing course. “They wanted to make sure that taxpayer dollars were being used as efficiently as possible,” he said. “On paper, it probably looked like the right thing to do to consolidate these centers.”

But the episode points to data problems behind the decision.

In the official regulatory notice last May, the Labor Department defended the plan to shut down the Forest Service job centers by claiming they “suffer from a variety of problems, including operating under-capacity, not achieving long-term student outcomes, and operating in an inefficient manner.” However, no data was provided by either department to support that statement.

The Trump administration has also asserted in budget documents that the USDA-run sites on average were more costly and less effective than other centers managed by the Labor Department — even though their own performance data shows that most of the Forest Service centers scored in the top 25 percent of all job training centers, meaning they significantly outperformed the other sites.

A USDA spokesperson said it’s the Labor Department’s decision to shutter any of the job training centers. But the closures were a central piece of the plan from the day it was initiated by Perdue and announced to startled lawmakers and federal workers.

The spokesperson said officials are now searching for “a pathway that will maximize opportunity and results for students, minimize disruptions and improve overall performance and integrity,” citing the need for the Forest Service to focus on its “core natural resource mission to improve the condition and resilience of our nation’s forests.”

A slaughterhouse overhaul invites safety questions

In certain cases, questions about USDA’s use of data have exposed its policies to bureaucratic and legal hurdles. That includes lawsuits and an inspector general investigation into whether food safety officials relied on faulty data to justify their recent overhaul of pork slaughterhouse inspections.

The final rule released in October removes federal limits on pork processing line speeds — allowing meatpackers to move more carcasses per hour and maximize profits. But labor advocates have long warned that ratcheting up the pace of operations in messy, humid slaughterhouses will further endanger plant workers, who already face higher rates of injury than those in other industries.

USDA’s Food Safety and Inspection Service maintains that worker safety is under the Labor Department’s jurisdiction, so it wasn’t a factor in shaping the final rule.

However, in a February 2018 regulatory notice, FSIS wrote that any evaluation of changing the processing line speeds “should include the effects of line speed on establishment employee safety.” Officials went on to claim that their “preliminary analysis” showed that plants with faster line speeds under a pilot program actually recorded lower worker injury rates than other facilities.

But independent researchers said they discovered flaws in the statistics underlying USDA’s assertion: FSIS relied on a sample size that was too small to draw meaningful conclusions and methodology that would leave statisticians scratching their heads, they wrote.

“It was really an inappropriate analysis,” Celeste Monforton, an independent expert in occupational safety, said of USDA’s claims at the time. “There isn’t enough data to make that kind of conclusion. It just doesn’t exist.”

FSIS argues that the analysis in question was only mentioned in the initial rulemaking notice for the sake of soliciting public feedback. A USDA spokesperson said the agency “did not rely on that analysis for the proposed rule and did not use the worker safety data in the final rule. Therefore, the claim that FSIS used flawed data is not accurate.”

But doubts about the soundness of USDA’s data have already created obstacles for the pork slaughter rule.

First, lawmakers cried foul over the worker safety concerns. Then, an IG probe was launched in June. Now, labor groups are suing to block the final rule. A central piece of their lawsuit is that USDA failed to disclose its data and analysis during the public comment period and rejected public concerns about worker safety “based on a methodologically flawed analysis.”

Some farmers gain, others lose

Complaints about USDA’s statistics extend to Perdue’s signature effort so far: a massive two-year trade bailout for farmers and ranchers bruised by Trump’s tariff wars. The department has doled out $8.6 billion to producers for 2018 production and nearly $11 billion for last year’s losses, so far.

USDA officials have published the economic calculations showing how they determined which regions or commodity sectors get more taxpayer money than others, but some lawmakers, commodity groups and farm economists have questioned the outcome.

Many in the industry complained from the get-go that they weren’t being paid an adequate rate to offset the financial sting from a steep drop in exports and lower commodity prices. Most famously, corn growers were outraged about receiving just one penny per bushel under the 2018 trade aid plan, when industry estimates showed that corn prices for farmers had sunk 44 cents per bushel on average since Trump started his trade war.

Smaller corners of agriculture, like Maine’s wild blueberry growers, say USDA completely failed to address their financial losses from the tariff brawl. They’re facing a 97 percent drop in exports to China since 2017, but were left out of the direct payment program altogether.

Meanwhile, a study published in November by the conservative American Enterprise Institute found that USDA’s calculation of trade damages “likely substantially overestimated the impact for some U.S. producers,” such as soybean growers.

The AEI study, conducted by former USDA chief economist Joseph Glauber, cited six other academic studies that found significantly lower trade damage to soybean prices than the department’s estimate. USDA officials have said they measured “gross trade damage” rather than “price methodology” and didn’t account for other factors that have weighed on commodity prices, but lawmakers and farm economists have taken issue with that approach.

Senate Democrats in November said they found inequities in the bailout program like a disproportionate amounts of aid flowing to Southern states and larger farm operations relative to their trade losses. Michigan Sen. Debbie Stabenow, ranking member on the Agriculture Committee, specifically criticized “USDA’s flawed aid formula.”

A USDA spokesperson pointed out that 68 percent of the total payments have flowed to the Midwest, with farmers in Illinois, Iowa, Kansas, Minnesota and Texas hauling in the most funds so far. “While we appreciate feedback on this program, the fact of the matter is that USDA has provided necessary funding to help farmers who have been impacted by unjustified retaliatory tariffs,” the spokesperson said.

Critics see a degree of political favoritism behind the design of the program, given the influx of funds to key states that Trump hopes could catapult him to a second term in the White House. Even after delivering several long-sought trade wins in recent months, Trump tweeted in January that he hopes “the thing [farmers] will most remember” is the bailout money.

How many kids lose automatic free lunches?

USDA has also been accused of playing keep-away with some of the most important metrics related to its policy plans, like the recent proposal to curb broad-based categorical eligibility under the Supplemental Nutrition Assistance Program.

The change would result in 3 million low-income Americans losing automatic eligibility for SNAP, by USDA’s own count. But anti-hunger advocates quickly homed in on another key figure: the number of school children who would lose automatic access to free or reduced-price school meals under the plan.

That number was nowhere to be found when USDA released the draft rule in July.

The department then declined to answer questions for months about how many students would be affected. House Education and Labor Chairman Bobby Scott (D-Va.) tried to pry the figures from USDA; he claimed that Perdue’s staff had initially suggested during a phone briefing that more than 500,000 children would no longer automatically qualify for free meals.

Months later, USDA’s Food and Nutrition Service published a new analysis in October, late in the day ahead of a House committee hearing with the department’s top nutrition official. That document showed 982,000 students would lose their automatic eligibility for free school meals under the new SNAP rule — nearly double what the department had informally told congressional staff.

Lawmakers at the hearing angrily questioned why it took so long for USDA to acknowledge the extent of the potential impact on low-income school kids.

Rep. Suzanne Bonamici (D-Ore.) said the monthslong delay “makes it appear USDA was trying to thwart oversight.” Deputy Undersecretary Brandon Lipps contended that the analysis was provided “as soon as it was available and ready.” A USDA spokesperson reiterated that point in a statement, adding that “every family continues to have an opportunity to apply for meal benefits.”

To veterans of the department, USDA’s original analysis in June should have painted the full picture. Offutt, the former ERS administrator, said it’s unusual that the department didn’t explore the potential consequences for low-income schoolkids before proceeding with the proposed rule.

“In this case, it was so obvious that the analysis could have been done,” Offutt said. “For the big regulatory changes, certainly there’s a good expectation that there should be a full analysis of the consequences. And that didn’t happen.”

The hazy impact of a food stamp crackdown

House members have pressed the department for more statistics relevant to another piece of its proposed SNAP crackdown: curbing states’ ability to waive work requirements for able-bodied adults without dependents.

Perdue has defended the plan as necessary during a strong economy to promote self-sufficiency. Democrats and anti-hunger advocates argue the rule will harm recipients who need more flexibility under work requirements, including veterans struggling to readjust to civilian life or young people who recently left foster care.

To that end, the House Agriculture Committee has been asking Perdue since early 2019 for detailed information about the population that would be affected by the rule, like how many are veterans or homeless. At a hearing last April, top Democrats, including Reps. Jim McGovern of Massachusetts and Marcia Fudge of Ohio, complained that USDA had yet to turn over any new information months after they requested it.

“They are just making up things,” Fudge said after the hearing, later adding: “They are making a decision with no data whatsoever.”

USDA had already shared a publicly available report including demographic information on a sample of able-bodied adults without dependents, including basic household and income data. However, Democrats were seeking more specific analysis that would capture barriers to working, such as a lack of access to transportation. Nearly a year later, they claim they’re still waiting for Perdue to produce that data.

“The fact is that [the recipients] are a complicated group of people on which we have little data … yet USDA has done no research on how this new rule will impact these vulnerable Americans,” McGovern said after the final rule was published in December. “The Trump administration ought to know more about this population before they literally take food off their table.”

USDA says it’s not authorized to gather data on the specific groups of SNAP recipients sought by Democrats, but Fudge and others have disputed that point. At a recent House Veterans’ Affairs hearing, Food and Nutrition Service Administrator Pam Miller said USDA would undertake a new study to gather more information on veterans who will be affected.

Uprooting a research agency

The branch of USDA that often conducts the sort of scientific research that Democrats requested on SNAP recipients is the Economic Research Service. But the ERS itself remains in a state of flux because of Perdue’s controversial plan to uproot the agency from D.C. and transfer it to the Kansas City region, a move announced in mid-2018.

The agency started hemorrhaging staff ahead of the relocation, in part because of widespread fears that the abrupt move was part of a political crackdown on scientific research that didn’t fit with the Trump administration’s agenda. More than half the ERS employees selected for relocation refused to move, according to USDA estimates in July.

Mass attrition has left parts of the agency paralyzed. An internal USDA memo in September identified dozens of reports that could face “significant delays” because of the relocation and the resulting staff exodus.

One of the projects flagged for potential delays was a study of the number of working-age veterans who lack access to food.

Of course, the relocation itself was perhaps the highest-profile case of a sweeping decision that critics see as based more on politics than science.

Perdue said the move would make the ERS more effective by bringing it closer to farmers in the heartland, though agency employees point out that their primary audience is agricultural policymakers in Washington rather than producers on the farm.

The secretary also promised it would save taxpayer dollars by avoiding pricey D.C. office space.

But when USDA released a cost-benefit analysis in June claiming modest savings over 15 years from the relocation, the findings were disputed by independent economists, including Offutt.

They said the move would actually cost taxpayers an additional $128 million over time, rather than the $300 million in savings Perdue was touting as a primary reason for the change.

The outside economists said USDA’s analysis overstated the costs of keeping the agencies in Washington and failed to consider cheaper locations in the Beltway area. The department also didn’t account for the economic value of agricultural research that would be lost due to veteran employees fleeing the agency in droves.

USDA continues to stand by its cost-benefit analysis and the relocation itself.

Former ERS researchers told POLITICO that the process demonstrated USDA leadership’s indifference to backing up major policy and organizational moves with facts. According to one economist who quit the agency last year, “the decisions are made, and then gathering data is an afterthought.”