That could throw the numbers way off since those who volunteered are likely people who thought they had symptoms but could not be tested before.
It's just so hard to process the numbers from this thing because it's on such a short time scale and there is no way to have a control to which to compare the numbers. It's part of what epidemiologists do but it's why epidemiology isn't really a "hard" science. A lot of it is based on the best information we have which is often flawed because you can't do controlled studies on humans. Conclusions are often made based on the overall weight of the evidence even lacking a true consensus.