Answer:
Not using control cases and testing his hypothesis only on one pacient, himself. Not creating or using a method that allowed him to measure the results.
Explanation:
During the investigation Alon had a few methodological flaws. First of all, he based a hypothesis over only one case, his own. He should have tested the product on different pacients under similar circumstances, the more the merrier. Moreover, it is not recommended that the investigator tests the hypothesis on himself, since he can lose perspective.
Second of all, he did not use control cases. This means he should have given placebo pills (pills with no medical effect at all) to a group of pacients without them knowing. Doing this helps eliminating the psicologycal factor of the equation, since pacients have proven to get better, in some cases, just because they thought the medicine would heal them, and not because of the drug itself.
Also, he came to a conclusion based on how anxious and worried he felt after the treatment, but expresses the results with words such as "much more". In this case he should have created and/or used a method to measure the change in anxiety and how much the pacient worried. For example, before starting the test he could have rated his anxiety level from 1 to 10, and then, done the same thing during and after the treatment was over. By doing this he creates data that can be numerically processed and makes the human subjectivity factor smaller.
