What is Replicable Research and Why Should You Care?
What is Replicable Research?
Research is replicable when an independent group of researchers can copy the same process and arrive at the same results as the original study. Hence, establishing its validity. For example, if someone conducts experiment A and arrives at conclusion B and you attempt to replicate this experiment but end up with conclusion C, the experiment is not replicable.
This is closely related to the idea of empirical generalizations. Empirical generalizations are results that cannot be replicated by independent researchers using valid, but different, methods.
There are three key aspects to the concept of replicability: a finding being replicated, the independent group and the use of valid, but different, methods.
Replication of a single finding
The concept of replicability relates only to a single finding. If a study contains multiple findings, each finding must be assessed separately. While the failure to replicate one finding of a study will typically bode badly for the rest, it is not unusual to have only a subset of findings from a study be replicable.
The independent group
To be considered valid, the group of researchers replicating the study need to be independent of the original researchers. "Independent" means that they have no reason not to be dispassionately objective in their attempts to replicate the findings.
Examples of a lack of independence include people working in the same department, people who have met socially, and situations where the original researchers have considerable influence (e.g., editors of major journals).
There are many potential explanations for different research findings. The most desirable explanation is typically that the research result is consistent with the theory because it is correct. However, there are many other explanations for findings: fraud, flukes, technical errors, and spurious correlations. For this reason, research is typically only regarded as being replicable when the findings have been reproduced using a methodology that is, in some important sense, different to that of the original study. Where a finding is found in multiple different ways it is said to be generalizable.
Why should you care that research is replicable?
In order to take a scientific study or experiment seriously, the results need to be able to be proven multiple times by independent researchers. If something is proven to be true once, why should you trust that it will always be true? Research that has been shown to be replicable affords greater confidence in the results.
It is also crucial that the process of replicating research is undertaken by independent researchers. This will reduce the chances of the research being influenced by selection bias or other dodgy elements.
Two concepts that are closely related to the idea of replicable research are
- Reproducible research. This is research where another researcher can, given the original data, reproduce the findings. Replicability and reproducibility are commonly used as synonyms.
- Repeatable research. This is research where another research can, given detailed instructions, conduct a similar study and get the same results.
Each of these is less stringent than replicability.