Sabbath’s Tweeter

A scientific examination of the garbage Trump posts on Jared and Ivanka’s day of rest.

Photo illustration by Natalie Matthews-Ramo. Photo by Chris Kleponis-Pool/Getty Images.
Photo illustration by Natalie Matthews-Ramo. Photo by Chris Kleponis-Pool/Getty Images.

Late last month, Vanity Fair reported that Jared Kushner—Donald Trump’s son-in-law and senior adviser, widely acknowledged during the campaign to be a moderating influence on the president—appeared to be losing his power. The piece observed that, among other snubs, Trump had issued his executive order on immigration just before the start of Shabbat. Kushner and his wife, Ivanka Trump, are observant Orthodox Jews. They aren’t supposed to work, use electronics, or babysit world leaders from sundown on Friday to sundown on Saturday, and they were apparently off-duty as protesters massed nationwide at airport terminals where travelers were illegally being detained.

The piece in Vanity Fair set off a wave of speculation about Kushner’s religious observance and its relevance to his role in politics, including a panel discussion on CNN. “When the Jews are away, the goys will play,” crowed Alec Baldwin’s Trump on Saturday Night Live before submitting to the bad advice of a decidedly gentile Steve Bannon. But no one has tried to verify empirically whether Trump behaves differently on Shabbat. This week, I made a first attempt using Trump’s self-created database of insults, outbursts, and propaganda: his Twitter account. I analyzed the Shabbat tweets, I analyzed the non-Shabbat tweets, and I discovered some interesting patterns.

Before I present my findings, I ought to note a few things. Kushner and Ivanka aren’t the first Orthodox Jews to reach these strata of power. Before them were Jack Lew, Obama’s chief of staff and then–treasury secretary, and former Sen. Joe Lieberman, both of whom were willing to suspend Shabbat observance in times of national emergency. Kushner and Ivanka themselves got a rabbinical exemption to ride in a car after Trump’s inauguration. Kushner’s religious observance does not in any way determine his fitness to serve at the highest levels of government, and we’ll never know exactly how he adapts his ritual practices to his job unless he goes on the record to talk about it.

Still, we can be fairly confident that if Kushner exerted any influence at all on Trump’s non-Shabbat tweeting—by toning down drafts, by providing vocabulary words, by simply exuding a calm aura—he wouldn’t have the same effect on Shabbat. Let’s go with that hypothesis for now and see what the tweets say.

As of this writing, Trump has tweeted 1,257 times—plus anything he’s deleted—since he received his party’s nomination for president on July 19, 2016. If we skim off Trump’s copy-and-paste retweets, that number falls to 1,171. According to the conventional wisdom, only tweets sent from an Android device were written by Trump himself. There are 918 of those within this group. Seventy-seven were sent between sundown on a Friday and sundown on a Saturday, according to the National Oceanic and Atmospheric Administration’s sunset estimates at the longitude and latitude of Trump Tower. (Many observant Jews start Shabbat a bit earlier and end a bit later, but those rabbinically sanctioned cutoffs are not easily accessible in my statistical software.) Click here to view a spreadsheet of the tweets as of noon on Friday, compiled slightly later than the dataset I used.

The raw number of Trump’s Android tweets, Shabbat and non-, was high around the Republican National Convention, fell as the election drew nearer, and has steadily climbed since he won. In most months during this period, all but November and January, his tweeting rate on Shabbat exceeded his overall rate for the month, sometimes by far. In September, 15 of his 49 Android tweets, or 31 percent, were posted on Shabbat. You’d expect the proportion to be about half that if he were tweeting at a consistent rate throughout the week.

It’s hard to draw any conclusions about the content of the tweets because the sample sizes are small, and human language is notoriously hard to interpret using math. But a few things do stand out. Since the RNC, Trump has used an average of 17 percent more exclamation points per tweet on Shabbat. That result is statistically significant: The p-value is .04, meaning that the two categories of tweets are structured so differently there’s only a 4 percent chance that the discrepancy is due to random variation. I also tested whether Trump has used more consecutive capital letters on Shabbat (e.g., “FAKE NEWS”). He has—on average, a Shabbat tweet will have 29 percent more pairs of consecutive capital letters—but the result is not statistically robust.

One way to interpret the content of human language, besides math, is reading. Trump’s Shabbat tweets—which include such classic earworms as “So-Called Judge,” “What Is Our Country Coming To (When a Judge Can Halt a Homeland Security Travel Ban)”, and “Fidel Castro is Dead!”—are, to my eyes, clearly crazier than his others. But that isn’t science, so I conducted a poll of Slate employees to see if they agreed. Using a Google Form, I showed each participant Trump’s 15 most retweeted Shabbat tweets and his 15 most retweeted non-Shabbat tweets, in a random order, and asked them to rate the craziness of each collection from 1 (“Not crazy at all”) to 5 (“Unbelievably crazy”). On average, the 39 participants rated the Shabbat tweets 4.49 and the non-Shabbat tweets 3.92, with a p-value of .002. That’s crazy!

Finally, the Shabbat tweets appear to have gotten more retweets during certain periods, but the samples are too small and similar to support a hypothesis about the timing of those periods. In all months, Trump’s handful of Shabbat tweets have gotten either an equal or higher number of retweets on average—up to 74 percent higher in November. If you compare the tweets week by week across two dimensions, Shabbat/non-Shabbat and iPhone/Android, you can see that the Android tweets have consistently overperformed the iPhone tweets, but the Shabbat Android tweets have consistently overperformed the Shabbat iPhone tweets by even more (36 percent more) than the non-Shabbat Android tweets have overperformed the non-Shabbat iPhone tweets. Again, these claims wouldn’t pass peer review.

Trump’s Twitter usage is likely to keep increasing as his feuds with senators, judges, the press, corporations, and reality multiply. So long as he doesn’t convert to Judaism as a ploy to market Ivanka’s fashion line, we’ll have more data soon and will report back to you with any findings.