Because science is hard; good science much harder.

Every step of the scientific method has it’s own challenges. Here are some of the biggest problems that follow the steps of the scientific method:

1. Observation: You will need to observe something that has passed notice by others. You might use your imagination, if you are brilliant. More likely, you’ll have to use a technique that is hard to master, expensive and generally not available to most other scientists. An example of the first might be to observe that birds have convexity to their wings (as seen from the top) and realize they might use Bernoulli’s principle as lift. The Wright brothers (and probably others) saw that. An example of the latter is when I was looking at electron micrographs (EM) of muscle and noticed that most mitochondrial cristae run orthogonal to the muscle “Z” bands. Who has an EM machine lying about?
2. Hypothesis: My favorite. You need to not only think of a good explanation for your observation (as in the case of Bernoulli’s principle for the birds’ wings) but one that is testable. My smart students still tend to fail this part. To say, “the little demon inside made this happen” or something of the equivalent, is common and totally useless. There will be no experiment to prove or disprove this. Hence, the worst insult in science is that famously ascribed to Wolfgang Pauli. When asked what he thought of a presentation in physics he sneered, “Bad. Very bad. It was not even wrong.” This may have been apocryphal, but maintains it’s power to teach that failing in your hypothesis has value, but failing to adequately test it does not.
3. Experiment: I’ve published almost 400 peer-reviewed papers, each one with at least one successful experiment. And I don’t think that even once was I able to use data from the experiment that I first attempted. As you do the experiment, things go wrong. The apparatus, or the way you acquire data, or something is obviously problematic and needs retooling. So you do it again, and fix it. And then again, etc. I’ve had fun asking students who recently finished their PhD thesis how long it took (typical answer: 3 years) and how long would it take to do it again now that they knew all the technical mistakes they learned along the way (typical answer: 3 months).
4. Analysis of results: There are so many ways of messing this up. But the most common one that I see is ascertainment bias. In theory, what you test, models the more general universe. But usually, it isn’t really representative. I could test the hypothesis that my average patient sees 3 doctors before they come to see me. What about the patients that never come to see me? Maybe they gave up after 5 doctors. Or got better on their own. Or maybe they died. The point is that I’m only looking at those that came to see me and that may not reflect the general population.
5. Statistics: Actually part of 4 but it’s now become a science of its own. I wish I was better at statistics. But I’m good enough to know that a common problem is cheating in the data analysis by looking at correlations that weren’t originally hypothesized as meaningful (post-hoc analysis). Our new era of genetics gives us gobs of data. So people are publishing that for a certain disease, patients often carry X, Y and Z genetic variants or mutations. But with 22,000 genes from each parent and trillions of possible combinations, there are all sorts of false or coincidental associations that will come up when you let the computer search for association. I’m impressed if you started by hypothesizing the involvement of a certain genetic mutation, and then found it.
6. Not fooling yourself. Human nature, psychology, politics and ego all make it very easy for the scientist to interpret his results with a wish bias. Nature sometimes fools us. But human nature reliably gets us to make conclusions that we wish were true. It’s a never ending battle to avoid this. The good scientist is constantly aware of this problem. The great scientist never trusts himself.
7. Writing the paper: If you did the study but never published it, you didn’t really do the study. Or at least no one will be able to profit and build on your work. It’s sort of a tree that falls in the forest that nobody heard fall down. Many scientists hate writing. Some write poorly. Some can’t handle the criticism and rejection that comes from the peer-review process. I’ve lost many faculty members who had it all together, but couldn’t get past this last step.
8. Funding: Someone has to pay for the work. And getting a government grant (typically NIH for biomedical research) is extremely hard. Months of concerted effort often lead to nothing. Careers hang in the balance. Nervous breakdowns are not uncommon. Sleep, marriages and jobs are often put into jeopardy.

Good science has always been hard. But the reward is the amazing rush; the thrill of extracting from nature a few key secrets and knowing that these will be building blocks for future works, mostly by others, that will lead to something grand. Or maybe it’s as simple as satisfying your own curiosity.