I must have ranted too long, and my response got eaten. Here's the short version:
I don't think anyone's read the study just yet. All we've got is an abstract from a journal with an editorial seeking submissions.
We can certainly all poke holes in the idea of it. IO might want to point out all the things that don't result in an accident or incident report, and we can concede that; the problem is, if they don't result in something measurable, we can't measure it.
More seriously, as PBL suggests, it's a zero-sum game. Just as everyone dies from something, so the the change in one cause of death will be marked by another. If we cure cancer, we'll see articles decrying the "worsening epidemic of heart disease." If half the population gets wiped out by plague next year, we'll see boasting of how we're winning the battle against AIDS, cancer and heart disease (rather than just the battle against Old Age). So, if one of the factors decreases or increases, the others are affected.
The third problem comes from NA's suggestion that the criteria have changed. Now, his particular statements strike me as being completely false:
Back in those days, there was still a kind of corruption that said: "blame the dead, save the live ones". They blamed anyone who was most convenient to blame, mostly the pilots if they happened to be dead.
I'm not sure that's any more true then than it is now. In societies sufficiently secular where Acts of God are relegated to a sideshow, most of the players in an accident are legally well represented. Individual screwups by pilots are the exception, and they're also an exception to the payout. But they play differently around the world. Some cultures and governments resist the belief that their pilots could make mistakes or commit airborne murder-suicides. And others see a payout from a foreign airline or manufacturer as a good thing.
The pressures on the investigations that produce the report vary by time, culture and government, and they vary considerably. They don't "even out in the wash", and quantitative analyses cannot control for them very well. That's why quantitative analysis of historical data needs to be viewed with suspicion.
As for the other claims:
Also the method of investigations has changed: now investigators look at the chain of events that brings to the crash and reconstructs the whole flight, down to the pre flight events. This has proven to be very helpful in reducing accident factors.
I think you'll find this is the case before 1983 too, where the handling of ground crews who were unable to read the instructions on how to close cargo doors was looked at, the maintenance record scrutinized, even the arguments in Ops duly noted and considered.
Improved media coverage has played an important role in pushing investigators to undertake full investigations to determine the real causes of crashes.
A good example is the Bijlmerramp case, an EL-AL B747 that crashed after take-off in Amsterdam. Journalists found out that investigators were seeding their reports with false information and criticized them widely.
Again, many would disagree. What about a TWA B747 that crashed after take-off in New York? Journalists seeded their reports with false information and criticized investigators widely.
Even leaving apart the questionable notion that media coverage has "improved", that "improvement" does not necessarily imply that it's more accurate.
Most hull-losses occured in the last years happened in emerging countries that have aviation standards that developped countries had 20 years ago.
That is mainly because they can't afford to have the same, modern equipment, and to have the same maintenance standards.
This is just ludicrous. 20 years ago in developed countries, flying "standby" did not mean flying "Standing Room"
Also, I would like to add that though the modern computer systems do not make any decisions, they take over many tasks from the pilots so that these can concentrate more on the decisions they need to make. Add to that, they assist pilots in decision-making by giving them precise data through faster, more precise calculations.
Technological progress does not mean improved conditions. EGPWS and TCAS may help, but the same cockpit automation removed the FE. More precise data and calculations might help, but the same computers allow for peak efficiency of duty rosters, load factors and equipment. And minimum rest cycles were not originally designed to be regular work conditions.
Oh, and speaking of which, ODB: your article needs some sources to it. I also question some of the figures. While it's not disputed that someone operating fatigued can function like someone with a BAC of .05%, nobody functions with a BAC of .8%. Nice cat, though.