Updated: Aug 2, 2020
The state of Texas shows far more threatening rates, peaking close to 19%. But at least it shows a more reliable downward trend since July 12.
A downward trend in this measure indicates one or both of:
Fewer positive test results recorded (lower numerator in the fraction) - Notice that this could be fully or partially accounted for by delayed test results, at least until the backlog clears
More tests given overall (larger denominator)
Both of these measures are directly available in the dataset from covidtracking.com
The positive test results received show a very recent decrease in the 7DayAvg in the last 9 days - too soon to call a trend, but hopeful:
Texas Positive Test Results, Recorded June 1 - July 28
The total number of tests given in Texas increased steadily until just the last few days, then show a slight decline most recently.
Texas Tests Given June 1 - July 29
This complicates the picture somewhat. With BOTH a declining numerator AND a declining denominator, the change in the positivity rate itself depends on the size of the changes in these two underlying measures.
Notice that the Texas positivity rate began its decline around July 15.
At that time, the positive test results were close to flat, but tests given were increasing: Same numerator with increasing denominator = declining positivity rate.
Important questions to ask at this point are WHY are these changes occurring, and WHY are they occurring together. We already have an extensive record of reports of shortfalls in testing availability and delays in tests processing.
This would have an impact on both of the underlying measures, at the same time - you can only observe what you actually measure.
This doesn't mean that these reports fully or even partially account for the positivity rate decline. We can only say that the observed numbers are not inconsistent with that hypothesis.
It is dismaying in its own right that tests administered shows a very recent decline. Hopefully this has occurred only because of the normal weekend dip + hurricane conditions in a highly impacted area.
In order to know what is going on and generate good policy decisions, there is NO substitute for rigorous, extensive measurement of the phenomenon being studied.
The US and Texas both have big problems with observed positivity rates
It is **possible** that conditions are improving slightly, but these trends are too soon to call as of July 30
All opinions expressed in this blog are solely those of the authors, and do not reflect those of University of Texas Rio Grande Valley, or any organizations of which either is a member.