President Trump says that the low unemployment rate among black Americans will help give him another four years in office. (Pablo Martinez Monsivais/AP)

During his speech in Cincinnati on Thursday night, President Trump touted his track record of delivering for black Americans.

“To give former prisoners a second chance at life,” he said, “we passed groundbreaking criminal justice reform that nobody could have passed, except us. President Obama tried, a lot of them tried, they couldn’t get it passed. I got it passed with a lot of help from liberals and conservatives, and a lot of help, and the biggest beneficiary is the African American community. Biggest beneficiary. Right?”

The crowd began to cheer. Trump turned to his right and nodded, joining the audience in clapping and pumping his fist toward a group of black supporters sitting nearby.

Trump often cites prison reform as a reason that the black community will support him in the 2020 presidential election. His most common claim, though, is that the low unemployment rate among black Americans will be reason enough for them to give him another four years.

“Unemployment has reached the lowest rate in over half a century, and unemployment for African Americans, Hispanic Americans and Asian Americans have all reached the lowest rates ever recorded,” he said in Cincinnati.

Have all reached — but are not currently at. New jobs data released on Friday indicates that the black unemployment rate is 6 percent, just above the record low set in May of last year. Among Hispanic Americans, the rate is 4.5 percent, a bit above the 4.2 percent low. Regardless: good numbers.

There’s just one catch. So far, it hasn’t helped Trump with those voters.

When Quinnipiac University first polled Americans on their views of Trump as president shortly after his inauguration, his overall approval rating was 36 percent. In the most recent poll, released earlier this week, he was at 40 percent approval. Over that same period, the unemployment rate dropped from 4.7 percent to 3.7 percent — more than 20 percent.

Quinnipiac started breaking out black and Hispanic approval in August 2017, at which point the unemployment rates in those communities were 7.6 percent and 5.1 percent, respectively. Black approval of Trump was 4 percent that month and 6 percent in the poll released this week. The improvement among Hispanic respondents has been better, from 24 percent to 40 percent in the most recent poll. The number of Hispanic respondents is small enough, though, that the margins of error — and swings in polling — are large. In March, Quinnipiac polling had Hispanic approval of Trump at 20 percent, meaning that approval in that group shifted 20 points while overall approval moved 2 points.

So let’s look at it another way. We took all of Quinnipiac University’s polling since Trump was elected and created a monthly approval value for different groups. To accommodate months in which there were no polls and since presidential approval seems like it would be a trailing indicator to positive economic shifts, the average includes polls from a given month and the month following. (Correlation using this average was slightly higher than looking at the months alone.) We then compared the change in those averages to the change in the unemployment rate in the month.

For each group, we looked at the number of months in which unemployment went down and approval went up and vice versa. We also looked at another metric, the employment-population ratio, which is a measure of how many of those in the labor pool are working. (This allows us to account for population changes.) For both metrics and each demographic, we made charts showing what happened in the 30 months under consideration.

Those charts look like the one below. Darker red boxes indicate more months in which the combination occurred. Looking at the example below, the lower left box shows Metric B going down (-) and Metric A also going down (-). The top left box is Metric B going up (+) and A going down (-). Top right? Both going up. The dashed line shows what a correlation would look like: darker boxes along the line as both went down, stayed the same (the middle boxes) or increased.

(Philip Bump/The Washington Post)

For unemployment, we’re looking for an inverse correlation. As unemployment drops, we would expect approval to go up. As the employment-population ratio increases, though, we would expect a direct correlation.

So, what do we see? We see a mishmash of results.

(Philip Bump/The Washington Post)

Consider the black/unemployment box. It’s darker at upper left and bottom right, which is what we would expect. But it’s dark elsewhere, too. In 12 of 24 months (August 2017 to July 2019), unemployment went down as approval went up or vice versa. In the other 12 months, something else happened, including six months in which unemployment went up alongside approval or both went down.

After the 2016 presidential race, Trump gave a speech in which he thanked black voters . . . for staying home that November. This is another way that the good economy might help Trump: People disinclined to vote for him might also be disinclined to go out and vote for his opponent.

This is hard to measure, but Quinnipiac’s polling gives us data that might help. Included in its approval polling is a question on how strongly people feel about Trump. One thing we might expect to see is that, if people are less worried about the economy, they might also feel less negative about Trump.

So does that happen? As unemployment falls, does strong disapproval of Trump fall with it? Again, the result is not clear-cut.

(Philip Bump/The Washington Post)

In the first poll where Quinnipiac broke out black respondents, 87 percent strongly disapproved of Trump. Now, 79 percent do. In a presidential election, small shifts in views can have big ripples; the drop in turnout among black voters in Michigan three years ago, for example, may have cost Hillary Clinton the state. What we don’t see here, though, is strong evidence that the improving jobs numbers are boosting Trump significantly.

That holds with the population overall, too. There were 10 months in which overall approval and unemployment moved in opposite directions, as we would expect — but eight in which they moved together. There were 10 months in which both approval and the employment-population ratio moved the same direction and five in which they didn’t.

(Philip Bump/The Washington Post)

This is a very complicated way of making the point we raised at the outset: There’s not a lot of evidence to suggest that lower unemployment — itself a continuation of a years-long trend — is making people feel more positively about the job Trump is doing.

While Trump repeatedly disparaged President Barack Obama’s handling of the economy on the campaign trail, for many Americans things have generally continued in the same direction. However you felt about Trump at the outset of his presidency is probably how you feel about him now. Polling suggests that’s certainly true for most black Americans, lower unemployment and prison reform notwithstanding.