It is true that the effect a president can have on employment in the United States is limited. The American economy is a sprawling thing that broadly defies the ability of the government to reshape things significantly.

It is also true that presidents nonetheless like to point to economic data as reflections of their political success, when possible. President Donald Trump did this eagerly, touting job growth that was largely in line with his predecessor as a mark of his own capabilities. At least, he did that until the coronavirus pandemic crushed the economy early last year, when Trump went from touting his own success to casting blame on the virus.

With Friday’s jobs report, we now have a complete look at how employment fared under Trump, with the economy seeing fewer people working now than when he took office.

Interestingly, that’s not true for all demographic groups. While the number of White and Black Americans working today is lower than when Trump was inaugurated, the number of Hispanic and Asian Americans working is higher. Employment for every racial and ethnic group was hammered by the pandemic, but job growth for Hispanic and Asian workers was already faring better until employment collapsed in March and rebounded more dramatically. (Over the past few months, job growth has been flat.)

These data are volatile, often estimates of relatively small groups. But the federal data nonetheless show how Hispanics, particularly Hispanic women, were more likely to gain employment both before the pandemic hit and overall for the past four years.

Every group detailed above saw employment growth until last February. After that, hiring overall fell, given drops among White Americans who make up most of the country. But, again, there was growth among Hispanics and, over the course of the past four years, Black men (setting aside concerns about variability in these numbers).

What makes this particularly interesting is that there’s a loose relationship between the change in employment for these groups and the shift in how they voted between 2016 and 2020. Groups that added jobs since Trump took office had a more favorable Trump vote in 2020 than in 2016; groups that shed jobs had a less favorable one.

Let’s take White men and Hispanic men as an example. In 2016, White men preferred Trump by a 62-to-31 percent margin, according to Edison Research exit polls. Hispanic men voted against Trump by a 63-to-32 percent margin. Last year, the margin among White men narrowed to 61-to-38, an 8-point shift to the left. For Hispanic men, the margin narrowed to 59-to-36, an 8-point shift to the right.

Over Trump’s presidency, employment among White men dropped by about 2.7 percent. Employment among Hispanic men increased by about 2.9 percent.

Again, the relationship isn’t perfect, and exit poll data on racial vote preferences are notoriously fluky. But there is nonetheless an existent correlation.

If we look at the numbers before the pandemic hit, the relationship is a bit stronger. Groups for which employment had increased the most from January 2017 to February 2021 were more likely to shift their support in Trump’s direction from 2016 to 2020.

Just as one should not attribute too much of the economy to presidential stewardship, one should similarly be cautious in linking presidential votes to any one factor. It is nonetheless the case that presidential votes have historically been linked to perceptions of the president’s job approval. If a voter feels as though she is doing better under a president, she is more likely to view his presidency with approval — and to support his reelection.

Trump wanted to make the 2020 election about jobs and the economy (when he wasn’t trying to make it about crime and fear). Maybe, at some level, it was.