Yes, regarding the scientific method. I really really want for my birthday and Christmas, for this to be taught not just at college level but at elementary, middle, and HS (because either people never got it, or forgot when they did). Also how research papers are constructed. You have your intro, your methods, your results, and your conclusion. The methods have to be justified and consistent with what you are testing, and after you present your results, you interpret them in the conclusion. Not only are you required not to overstep or overstate what the results actually indicate, you are also required to list any weaknesses the study (lack of power, limitations of the population used, limits to methodology, etc) and what future research should look at. The scientific method is a process. Each paper is incomplete and only part of a larger process or body of knowledge. It may seem disappointing to those who think scientists have all the answers, or that science will provide a definitive unchanging "truth". I also see people read those limitations and conclude therefore the scientists don't know, or "nothing is known". That's also not true.
Honestly, even very well educated people I know don't actually understand how the scientific method works or choose to ignore it. I had to explain to an entire lab of PhD students, many of which are faculty at schools like Georgia Tech, Rice, etc. now, that what they were describing wasn't the scientific method. They believed that by doing a literature review, writing hypotheses and collecting data, then analyzing that data and re-writing their lit review and hypotheses to fit the data that that was the scientific method. I tried to explain to them that that was p-hacking and that the scientific method would include collecting a new set of data after re-writing their hypotheses. They politely told me I was wrong because their professor taught them that what they were doing was correct. Moral of the story is read even peer-reviewed literature with a keen eye. A good portion of it, especially in social sciences was p-hacked. That's what you get from a publish or perish model to tenure.
I'd like to ask a question by analogy.
In physics, a researcher observes a phenomenon, perhaps that the acceleration of falling objects seems to be the same. They will hypothesize something like "gravity provides constant acceleration" and then construct experiments to try to prove or disprove the hypothesis. Then they will try to compare a ball bearing and a feather, and it will seem to disprove their hypothesis.
If the physicist was operating like your students, she might create a new hypothesis that says something like "flat objects fall more slowly than round objects" and she would use the experimental data from the original experiment to determine how much more slowly.
But a more appropriate way to do this would be to create a more appropriate experiment. Instead of having a ball bearing and a feather, use a collection of objects with varying degrees of flatness and try and figure out the relationship between flatness and the acceleration that objects fall.**
In the physics example, the experiments are easy to carry out. But in social sciences the experiments are difficult to carry out because they require the participation of a statistically significant number of people. But too bad - if the experiment is not appropriate for the hypothesis, then you don't have good results.
Have I got it?
** Someone would eventually try the experiment in a vacuum and come up with a better theory...