#data integrity

Duggar Data loves exact data; however, it isn’t always available…  Say, when a couple tells us that a relevant event happened in “March,” or “Early March,” but doesn’t give an exact date.  Duggar Data estimates the date in these situations, and does so using a set of established rules to ensure “Early March” is always estimated the same way.

I’ve been rethinking my estimating rules recently, and I’ve decided to change them slightly.  It’s honestly really slight…  It probably doesn’t even matter.  But, I wanted to let you know!

Until now, my rules for estimating Month + Year (e.g., March 2020) and Specific Portion of Month [+ Year] (e.g., Early March 2020) were also follows—

Month + Year   Use the midpoint of the month.  (Round up if it’s a month with an even number of days.)

“Mid” Month   Same analysis as above.

“Early” Month   Use the 1st Quartile—i.e., the midpoint of the 1st Half of the month.  (Round up, if necessary.)

“Late” Month   Use the 3rd Quartile—i.e., the midpoint of the 2nd Half of the month.  (Round up, if necessary.)

Applying these rules, here’s the date I’d end up using—

Month + Year   15th (Months w/ 28 or 29 Days), 16th (w/ 30 or 31 Days)

“Mid” Month   15th (Months w/ 28 or 29 Days), 16th (w/ 30 or 31 Days)

“Early” Month   8th (Regardless of # of Days)

“Late” Month   22nd (Months w/ 28 Days), 23rd (w/ 29 or 30 Days), 24th (w/ 31 Days)

I don’t really like this.  Mostly because it implies that “Early” [Month] consists of the entire first half, and it’s just…  Not.  I wouldn’t consider Valentine’s Day to be in “Early February,” would you?

So, I’ve decided to switch things up…  Instead of dividing months into “Early” and “Late,” I will divide them into 3 Parts—Early, Middle, and Late.  Each part gets the same amount of days, to the extent possible.  As a rule, let’s always ensure that “Early” and “Late” get the same number of days; any extra days are allocated to “Middle.”  So, here’s what’s defined as Early, Middle, and Late—

28–Day Months   1–9 (Early), 10–19 (Middle), 20–28 (Late)

29–Day Months   1–9 (Early), 10–20 (Middle), 21–29 (Late)

30–Day Months   1–10 (Early), 11–20 (Middle), 21–30 (Late)

31–Day Months   1–10 (Early), 11–21 (Middle), 22–31 (Late)

… and, just like before, our Estimate Date would just be the midpoint of the relevant range.  Unlike before, let’s round down if the hint is “Early” Month and the Early Range consists of an even number of values.  (For “Mid” or “Late,” we’ll still round up.)  Here’s the date we’ll end up using, after applying this new rule—

“Early”   5th (Regardless of # of Days)

“Middle”   15th (Months w/ 28 or 29 Days), 16th (w/ 30 or 31 Days)

“Late”   24 (Months w/ 28 Days), 25 (w/ 29 Days), 26 (w/ 30 Days), 27 (w/ 31 Days)

If we’re told a Month + Year, but “Early” or “Late” isn’t stated, we’ll just take the midpoint of the entire month, rounded up if necessary.  (Like we did before.)

Like I said, it’s a small shift…  3 Days or Less, actually.  Regardless, the following data will be directly impacted by the rule change.  For all of these, we were told that the event was in “Early” or “Late” [Month], and that’s it…

C. Bontrager–Bowers #2’s Due Date   Said to be due “end of August” 2020.  This is treated the same as “Late August.”  Was using August 24, 2020.  Now using August 27, 2020 (+3 Days).

Samuel Dillard’s Due Date   Said to be in “Early July” 2017.  Was using July 8, 2017.  Now using July 5, 2017 (–3 Days).

Peter Waller’s Due Date   Said to be in “Early May” 2020.  Was using May 8, 2020.  Now using May 5, 2020 (–3 Days).

As of Saturday at 6:35 a.m. ET, the Census Bureau’s webpage about its executive leadership lists George Cook as “Performing the Duties of Director.”

The bureau’s website says Cook “worked at Brevan Howard from 2012 to 2017, BlackRock from 2017 to 2024, and briefly began launching Justinian Advisory, a startup economic research firm, before beginning his public service.”