The replication crisis can be defined as the methodological crisis where many studies are difficult or impossible to replicate. What this means regarding the social sciences is that what may be believed to be true at one point is unlikely to stay true. To put it another way – what we know now will likely be changed or expanded upon at a later date. This is because of differences in demographics, public knowledge, world events, and inaccurate/dishonest reporting of results during an experiment’s process. In the social sciences, there is no endpoint – only a point in current time where we have our best understanding on a study or experiment.
While the replication crisis is most notable in social and life sciences, like psychology, other fields are affected as well. As values, norms, and ways of living change, more than likely repeatable science and experiments will evolve and adapt as well. To be sure that all past findings will stay true, or if new determinations will be found out, is yet to be determined. What is not a given is if those initial findings will yield new determinations that completely disembark from knowledge that was gained at first.
It would be interesting to see whether the replication crisis was related to the failure of statistical methods being utilized, or whether some specific failure is responsible within each experiment and study.
When the time comes to recreate past studies – and ultimately when new findings become available from those studies – what does this mean? Do we disrepute those past discoveries, or do we build and alter our ways of thinking to include the advanced knowledge?
There may be ways to find new information and also have AI take into account details that were not recognized at first. In the journal Nature from July 3rd, 2019, new information emerged after certain scientific papers were fed through a machine learning algorithm. The machine learning algorithm could make determinations for future discoveries – which suggests that past scientific work has latent knowledge embedded in it. This knowledge can surface if simply run through a machine learning to figure what would make sense to study next. Using technology to better understand ourselves and the world we live in could lead to uncovering profound discoveries that may have gone unnoticed if not for AI and machine learning.
The replication crisis is a topic that has great potential to better our understanding on what may have been missed from the initial finding or even where to look next. Asking primary questions to form a hypothesis could be better informed by using AI and discovering new ideas from past research.
In conclusion, it bears further scrutiny that the importance of the replication crisis is publicly debated, and even studied further. Truth is – if there is more to understand from something like the Milgram experiment or the Stanford prison experiment, what do we risk losing if we don’t take a step back and dig deeper to discover even more than what those initials findings shared?