I served as a juror on a criminal case, in which the prosecution seemed to want to make a big deal out of DNA evidence found on some clothing. In turn, the defense wanted to dispose of that evidence by having an expert say that DNA evidence was unreliable. The expert they brought in was someone who admitted their knowledge of the field was only based on an article they read.
So the percentage of experts who spend real research on a question appears to be less than 100.
There is a long, sad history of trial courts improperly convicting defendants based on pseudoscience testimony by "expert" witnesses. We now know that forensic analysis of things like hair samples and bite marks is mostly bullshit. Even fingerprint and ballistic analysis has a high error rate. Unfortunately jurors often assign expert witnesses a level of credibility that they don't deserve.
So the percentage of experts who spend real research on a question appears to be less than 100.