[M T]

You may notice that Santa Clara County's testing results have changed in appearance.  This also corresponds to a change in reporting. (Announcement)

Data Notes: A test represents at least one specimen tested per person per day. Some people will have multiple specimens collected and tested during the same visit but they are only counted once. If a person is tested multiple times on different days then they will show up once for each day they were tested. The positivity rate is the percentage of tests performed that were positive.

Suppose Joe is a healthcare worker who gets tested weekly, and that he's tested negative every time.  Previously, his week-old test got removed from the counts once his newest test came in.   So, the number of negative tests 3 weeks ago would be 3 tests lower than they were originally.   Also, if you look at tests 3 weeks ago, the positive % would be higher than originally reported.  But, if you looked at the cumulative test county, you were seeing individuals.  If 100K tests had been given, that was 100K people.   So you had some sense of the % of people that had tested positive.

Now, he will be in the database of test results multiple times.  So, if "total tests" is 10% of the population of the county, that doesn't mean that 10% of the population has had a test.  The cumulative % positive number (still being reported as 2.72%) is artificially lower.   But the 7-day rolling average will be correct.  The % positive from (say) June 1 will be the same on June 8 and July 8.

They don't say it here, but I think once a positive test is recorded, no more tests are recorded for that person.  If they were, that could drive up the positive % artificially as a patient is tested while still sick.

[M T]

You may notice that Santa Clara County's testing results have changed in appearance.  This also corresponds to a change in reporting. (Announcement)

Data Notes: A test represents at least one specimen tested per person per day. Some people will have multiple specimens collected and tested during the same visit but they are only counted once. If a person is tested multiple times on different days then they will show up once for each day they were tested. The positivity rate is the percentage of tests performed that were positive.

Suppose Joe is a healthcare worker who gets tested weekly, and that he's tested negative every time.  Previously, his week-old test got removed from the counts once his newest test came in.   So, the number of negative tests 3 weeks ago would be 3 tests lower than they were originally.   Also, if you look at tests 3 weeks ago, the positive % would be higher than originally reported.  But, if you looked at the cumulative test county, you were seeing individuals.  If 100K tests had been given, that was 100K people.   So you had some sense of the % of people that had tested positive.

Now, he will be in the database of test results multiple times.  So, if "total tests" is 10% of the population of the county, that doesn't mean that 10% of the population has had a test.  The cumulative % positive number (still being reported as 2.72%) is artificially lower.   But the 7-day rolling average will be correct.  The % positive from (say) June 1 will be the same on June 8 and July 8.

They don't say it here, but I think once a positive test is recorded, no more tests are recorded for that person.  If they were, that could drive up the positive % artificially as a patient is tested while still sick.