I don't know about the quality of the data in those specific studies, but it seems to me like it would be very hard to collect high quality datasets on the phenomenon of cyclist-motorist crashes. For one thing, there is a huge reporting bias against reporting minor incidents to the police, and I would bet that many of those are motorist-caused. In many cases, all a motorist has to do in order to escape all blame for an incident is pull a hit and run and drive off, because the cyclist will often not have time to get their license plate number. And incidents in which the motorist is at fault would have a stronger incentive towards hit-and-run, which would bias the reporting against cyclists. For another, police officers may be biased against cyclists in assigning blame (I know we've read about many incidents where that was the case), so depending on their sources and methodology that kind of data could be questionable too. And finally, at least one big thing has changed since the 70's: the prevalence of mobile phone use while driving (other things that have changed are the explosion in popularity of SUVs, which have poorer side and rear visibility, and an increase in traffic). So I wouldn't say this is necessarily wrong, but I question the certainty with which figures can be reported and generalized.