At Northwestern University in Chicago, anyway - and probably externally valid to other institutions, as well.
This is a survey of 194 fourth-year medical students and 141 third-year residents regarding whether they observed or encountered "assurance" practice (extra testing of minimal clinical value) or "avoidance" practice (withholding services from patients perceived as high risk). 65% of medical students and 54% of residents completed the survey - decent numbers, but low enough to introduce sampling bias.
The numbers, of course, are grim - 92% of medical students and 96% of residents reported encountering "assurance" practice at least "sometimes" or "often", while 34% of medical students and 43% of residents had encountered "avoidance" practice at least "sometimes" or "often" - nearly all of those being "sometimes". These behaviors are apparently learned from their superiors - approximately 40% of medical students and 55% of residents were explicitly taught to consider practicing defensive medicine.
Interestingly, medical students, internal medicine residents, and surgical residents all reported nearly identical levels of "often"/"sometimes"/"rarely" regardless of the behavior sampled - although surgical residents were more frequently taught to be defensive than medicine residents.
Must be a tough legal quagmire up in Chicago.
"Medical Students’ and Residents’ Clinical and Educational Experiences With Defensive Medicine"