1.18 overall and 1.18 excluding New York. Stunning one-day slowdowns in several states, including New Jersey, Michigan, Illinois, and Louisiana. We will see how much the reporting catches up the next two days.
4 thoughts on “3DDRR update, April 19”
Pure weekend effect. Seen it several times now.
Why not use a 7-day interval instead of a 3-day interval? Might that not tend to eliminate or reduce weekend effects, since we’re comparing, say, this Sunday to last Sunday rather than to last Thursday?
Don,
The 7-day is 4 days behind the 3-day. That’s why.
I plotted the daily fractional change in new deaths as r_3=3DDRR^(1/3) and as r_7=7DDRR^(1/7). The graph of r_3 shows a strong 7-day pattern, undoubtedly due to reporting practices that vary with the day of the week. The graph of r_7 shows no such pattern, since we’re comparing, say, this Sunday with last Sunday.
If we’re looking for trends in that daily fractional change, the 3DDRR is going to mislead us: it’ll look like the rate’s falling on weekends, but then climbing again in the early part of the week. The 7DDRR gives us a clearer picture of the overall trend.
Pure weekend effect. Seen it several times now.
Why not use a 7-day interval instead of a 3-day interval? Might that not tend to eliminate or reduce weekend effects, since we’re comparing, say, this Sunday to last Sunday rather than to last Thursday?
Don,
The 7-day is 4 days behind the 3-day. That’s why.
I plotted the daily fractional change in new deaths as r_3=3DDRR^(1/3) and as r_7=7DDRR^(1/7). The graph of r_3 shows a strong 7-day pattern, undoubtedly due to reporting practices that vary with the day of the week. The graph of r_7 shows no such pattern, since we’re comparing, say, this Sunday with last Sunday.
If we’re looking for trends in that daily fractional change, the 3DDRR is going to mislead us: it’ll look like the rate’s falling on weekends, but then climbing again in the early part of the week. The 7DDRR gives us a clearer picture of the overall trend.