Given the comedic misinterpretation of the South African testing data offered by @BallouxFrancois (and many others!) last night ... I offer some tips having contributed to the analysis of the testing data for the @nicd_sa since April last year. (1/6)
1. The daily tests announced are those that are LOADED that day. The test may have been conducted several days/weeks earlier. What those daily reports mean is pretty hard to interpret. (2/6)
2. The VOLUME of daily tests is highly periodic, low on weekends, for example. The count of CASES is also highly periodic. But the proportion testing positive is ALSO periodic - so it's not a case of the two periodicities in cases and tests cancelling each other out. (3/6)
3. We know that many negative Ag test results are NOT loaded AT ALL. EVER. AND that delays in loading Ag test results are longer than delays in loading PCR results. (4/6)
It therefore follows that any proper analysis of the South African testing data HAS TO (at the very least)
* analyse by date of collection (not reporting)
* analyse PCR tests distinct from Ag tests
* monitor differentials in public/private testing (5/6)
* analyse by date of collection (not reporting)
* analyse PCR tests distinct from Ag tests
* monitor differentials in public/private testing (5/6)
All of these are offered in the weekly NICD report, published here: nicd.ac.za/diseases-a-z-i… (6/6)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
