Comments by "Harry Mills" (@harrymills2770) on "This Cornell Scientist Destroyed His Career in One Blog Post" video.

@dshin83 I guess I'm too dumb to understand this. Why not make a new hypothesis if your data support it? It makes sense to me to try to scry what the data are telling you. Then run more experiments to test the new hypothesis. As a trained geologist, that was sort of what we did, and I never met a more honorable and humble bunch of scientists. They understood the limits of their observations and were totally OK with presenting more than one theory, rather than making one theory their mission, and become advocates for that theory, and making things political. Maybe in physics, you hypothesize that the distance a body falls is proportional to the cube of the duration of the time spent under acceleration. Your data disconfirms your hypothesis, but suggests a squared relationship, with a proportionality constant very close to 1/2. While more experimentation is always better, that data is good support of a new hypothesis. Do you deny the world your discovery, because your first guess was wrong? One of the frustrating things to me as a student of geology was how strong the claims being made in psychology (and anthropology) were, compared to the weakness of their data. The softer the science, the more opinions harden, and the acceptance of one theory over another becomes a political fight between 2 sides that wield institutional power to prevail. Great statisticians. Terrible scientists and experimental designers.
56
I consider myself a scientist, and I don't have a problem with changing one's hypothesis if the data support something different. You're not just testing one hypothesis when collecting data. I was taught the method of multiple working hypotheses. As long as you don't abuse or falsify the data, I think that's perfectly OK. If the data destroy all your working hypotheses, you should come up with other hypotheses. I was also trained to be just as proud of null results as "magnificent discovery!" results. Both add to the body of human knowledge, even if the former is much more frequent and much less glamorous.
1
Surveys are notoriously unreliable data. The wording of the question can skew the results enormously. People might answer a particular way for all kinds of reasons. I remember this from the 1980s, when conservative surveys showed strong support for Reagan's "peace through strength" policy, and liberal surveys showed strong support for gutting the military. The key was in how they worded their surveys. None of those surveys were worth the paper they were printed on, but everybody did massively advanced statistics on their garbage data. Very impressive, but it was garbage-in, garbage-out situation in both cases. What compounded the idiocy was how STRONG the claims by both sides were. That's when I decided (2 years into my undergraduate studies) that the softer the data, the more certain the people using that data were about hard results. As a geologist-in-training, I was disgusted by the field of psychology.
1