The lowest level math course at the Air Force Academy is called Basic Math. Due to their performance on the placement exam, as well as a review of their math backgrounds, about 5% of incoming cadets (students) get placed in it. In my first year on the math faculty there, I determined that these cadets tended to do very poorly in downstream STEM courses that required algebra (Physics, Calculus, Engineering, etc.) and that WAAAAAY to many of these cadets ended up flunking out of the Air Force Academy as a result.
The chain of command made it a very high priority to improve the retention and graduation rates of this cohort of students (referred to by some as the "trailers and failers.") Since I shared some unique ideas for improvement and had succeeded previously at revamping similar math courses at a community college, it wasn't long before they put me (a Physics PhD) and another experienced faculty member (a Math PhD) in charge of the course.
We shifted the focus away from mechanical manipulations (solve this equation) toward modeling (word problems). What had been an 80/20 split mechanical/modeling became a 40/60 split on most homework assignments. Tests were 100% word problems (much better prep for Calculus, Physics, and Engineering which were the same). We dumped all the material that was rarely used downstream by non-STEM majors (sec, csc, cot, completing the square, complex numbers, law of sines, matrices, etc.) to focus on material that was heavily used downstream (
SOHCAHTOA, substitution, lines, quadratics, graphing, parametric equations, coordinate systems,etc.) We also acquired funding and provided opportunities to get these (weakest 5%) involved in publication quality research. Consequently, success rates in downstream courses dramatically increased.
It's not that the junior officers (with Masters degrees in Math) who had been teaching the course in prior years were doing a bad job - they were doing exactly what their chain of command had told them to do - teach a defined list of topics from a defined textbook. But the focus was overly mechanical (solve this equation) and followed typical high school math topics too closely without due emphasis on the material actually needed downstream. The PhDs were able to apply a higher level approach: why does this cohort of students tend to do so poorly downstream, and how do we fix that? We were empowered to identify and fix the problem, and we were bold enough to make the sweeping changes needed to do it. Teaching the same course in the same way would have yielded the same results, even if we were marginally better classroom instructors.
One key in success teaching college is realizing that much more learning occurs in the 2 hours of preparation students should be doing outside of class for each class hour than can occur in the classroom itself. There is more to be learned and mastered in a course than is usually possible in the 45 hours or so total of class meeting time. Directing how students spend that time outside of class and designing what happens in class to support and encourage that is more important than being able to explain each and every concept with perfect clarity. Directing students to the best resources to maximize their out of class productivity is also key. The final key to success is shaking students out of their fantasy that they might succeed in college with the half-hearted efforts that got them through high school.
PhDs are not universally better at these things than those with Masters degrees. But we do tend to be a bit better at seeing the forest for the trees, thinking about the bigger picture, having the boldness to make the required changes, and not being manipulated by all the tricks students pull to attempt to pass without working very hard.