The null hypothesis is usually a strawman. People who use a p-value alone in hypothesis testing are essentially making a decision based on the integral of the tail area of a curve built on a strawman. Are there not better decision-making tools?<p>Not only do hypothesis tests not require p-values, but p-values are not even the full problem. Null hypothesis significance testing itself is too often treated as the go-to solution for a statistical problem. Not every stage of a study needs a hypothesis test.
This is a very very short article that doesn't say much! But it quotes Steven Goodman, who has been one of the best and clearest thinkers on foundations of statistics for a long time IMO, so anyone who wants more could follow up on his name.