Is the three-foot bicycle passing law working in Baltimore, Maryland? is a 2012 paper published in Accident Analysis and Prevention.
The 5 subject bicyclists in this study were not “blind;” they were the investigators who collected and reduced their own data.
The investigators misrepresented the 3-foot law, interpreting “not less than 3 feet” of the statute as “greater than 3 feet.” The closest passes were 3 feet but were described as “3 feet or less.” This indicates confirmation bias.
Bicyclist lateral position has an enormous effect on motorist overtaking behavior and Vehicle Passing Distance (VPD), but it was not measured, making the VPD findings of little use. A bicyclist tracking in a lane control position compels a full lane change in motorists, while tracking at travel lane edge encourages motorists to squeeze by within the lane.
To obtain large VPDs in bike lanes, the subject bicyclist investigators likely tracked far to the right in the bike lane. Moreover, there were no 3-foot VPDs in bike lanes, although some would be expected. Similarly, they must have tracked far right in standard lanes even though they apparently recognized the utility of claiming a full lane. These discrepancies suggest subject-expectancy effect bias or deliberate manipulation of data.
Critique of: Is the three-foot bicycle passing law working in Baltimore, Maryland? HERE is a 15 page paper in PDF that details the fatal problems with this research.
Bicycling Matters 