Last fall I taught polarity in a different order than I have before. In every previous class, I introduced polarity in the VSEPR unit: first bond polarity, and then molecular polarity immediately after. Instead, last semester, I introduced bond polarity after Lewis structures and tried to get the students in the habit of identifying polar bonds and drawing dipole moments for those bonds alongside their Lewis structures. Then, after working through VSEPR theory, we brought polarity back into the mix. Anecdotally, they seemed to have a better sense of which bonds were and were not polar in the 3D molecules than my previous classes. (I don't know that the order of topics improved their understanding. I suspect it's simply a matter of having had more time to "digest" the idea and practice it more in 2D before moving to 3D.)

Even though they were quite good at identifying polar bonds, the students still had trouble identifying polar molecules. Specifically, they used the simple rule of thumb that if the bonds were polar, the molecule had to be polar, which fails when the symmetry of a molecule is such that the polar bonds all cancel out (e.g. BF3 or CF4).