Count the days actually on trial. The calendar put 27 Friday the 13ths between Jan. 1, 2009 and Dec. 31, 2023 [1], inside a window of 5,478 days [2]. A file of more than 250,000 unexpected deaths averages about 46 a day across that window [4], which places roughly 1,230 deaths in the exposed group [3].
That number sets the resolution. Counting noise on a group of 1,230 is near its square root, about 35 deaths, so a real difference has to clear something like 70 before it separates from chance, about 5.7 percent of the baseline [5]. An excess of one death in twenty on Friday the 13th would have surfaced in this design; an excess of one in fifty would not.
The contrast, as Rosenthal describes it, was between Friday the 13th and the overall rate on all other days [6]. Fridays are not interchangeable with Tuesdays for accidental death, and pooling every non-13th day folds that weekday difference into the comparison in the direction that would flatter the superstition, because every exposed day here is a Friday. The count still came in a shade below the rest of the calendar [6]. Holding weekday fixed, Fridays against other Fridays, or the 13th of each month against the 6th and the 20th, would isolate the date itself.
The interview, not the data, explains why the belief survives arithmetic. Rosenthal, who found out from a calendar program he wrote himself that he was born on a Friday the 13th [11], offers Apollo 13, which suffered its explosion on April 13, 1970, a Monday [8]. It gets filed as evidence for a Friday superstition anyway, which is confirmation bias doing ordinary work: the memorable case is recruited and the inconvenient weekday quietly drops out [7].
The paper also looked at full moons, Halloween and time of year, and the interview reports a finding for none of the three [12]. The moon is the better-powered test. A full moon comes round roughly 12.4 times a year, so about 185 exposed days sit inside the same 15-year window, near 8,440 expected deaths, and the same noise calculation bounds a lunar effect at roughly 2.2 percent [6]. If those numbers land where the Friday numbers did, the lunar null is the stronger of the two, and it is the one worth reading in the paper itself.
