For me, I am looking for a basic reasoning in an experiment. 1) They came up with an actual concept to test. 2) They made a hypothesis based on some sort of reasoning. 3) They determined what they actually had to measure to test this hypothesis. 4) They developed a procedure to isolate that measurement from other variables. 5) They took a critical look at the data. 6) They drew a valid conclusion from that data.
These are surprisingly rare things at science fairs. The highest scores I have given have been to projects where their conclusion was that they were unable to confirm or deny their hypothesis. This is the same problem as the premise of this article, that people believe science has to conclude something. No, some of the best science out there concludes that no conclusion can be drawn from the experiments. It doesn't mean you failed, it means that you either need to refine your test or look for other explanations. It is not only a perfectly good result, but it takes a lot of intellectual integrity for a student to admit that their findings were not significant. They have a very hard time articulating this, but it can be coaxed out of them.
Most students just take their data, put it in excel, have excel "best fit" a linear regression (regardless of if their data is linear) through their data and then conclude based on what is just noise in their data recording or worse, they extrapolate that line well outside of the data recorded in their experiment to make sweeping statements.
Maybe statistics should be a required course in high school... Or maybe earlier.