Confused? Allow 1,000 of our friends to explain:

Here they are! Let's say that

% of them are infected.

K

Infected

d

Not infected

Now, let's simulate an antibody test. On average, this test correctly identifies % of people who K have been infected (sensitivity) and it correctly identifies % of people who T haven't been infected (specificity).

Try a real-life test currently in use:

Roche’s antibody test was authorized by the FDA on May 3. It has a sensitivity of 100% and a specificity of 99.8%.

Cellex, Inc (pdf) made the first test to be FDA-authorized. It has a sensitivity of 93.8% and a specificity of 95.6%.

Premier Biotech (pdf) was used in recent serology surveys in California. It has a sensitivity of 80.3% and a specificity of 99.5%

K

Infected

d
m

Not infected

people tested positive for antibodies

people tested negative for antibodies

That means that if you took the test and got a positive result, there’s a % chance it’s correct. If you got a negative result, there’s a % chance your result is accurate.

Positive results

Our test correctly identified C with antibodies, but said K had antibodies when they actually didn’t.

Negative results

Our test was correct for P without antibodies but F with them were incorrectly told they had none.