糖心原创

From Trash to Treasure: Making Null Results Count in Economics

Empirical studies where the main ‘null hypothesis’ cannot be rejected are called ‘null results’. In the talk, I will start by briefly reviewing the state of affairs in empirical economics and social sciences, with the dominance of NHST and the aversion to null results. I will then introduce a new taxonomy of the types of null results, which will help us evaluate the arguments in favour and against publishing them. Then, I shall discuss a novel methodological approach based on randomised tests for strengthening the inference from null results. I will present randomised test applications addressing each type separately to provide scholars dealing with null results with an option menu for more powerful inference. I shall conclude by arguing that, for most cases, the practice of casually rejecting the contribution of null results is missing an important opportunity for knowledge accumulation and scientific progress.