Reliable science takes time. But the current system rewards speed

Recently, there have been many headlines about scientific fraud and withdrawal of magazine articles. If this trend continues, it represents a serious threat to public trust in science.

One way to tackle this issue and ensure public trust in science can slow it down. We sometimes refer to this philosophy as “slow science.” Like slow food movements, slow science prioritizes quality over speed and tries to absorb the incentive structures that promote mass production.

Slow science may not represent obvious ways to improve science. This often equates science with progress, and slowing progress seems less appealing. But progress is not only based on speed, but also on important social decisions based on a strong scientific foundation. And this takes time.

Unfortunately, the pressure and incentives faced by modern scientists are almost universal for slower science. There is a shortage of safe and permanent university jobs, and this appears to be exacerbated due to budget cuts.

As a result, the pressure to be made public is more than ever. Certainly, at my annual performance meeting, I am asked how many articles I have published and what the status of the journals I have published.

Fast Science Issues

Our current “fast science” approach creates many problems.

Like fast food, scientists are encouraged to produce as much science as possible in as little time as possible. This means cutting the corners. For example, we know that larger samples are more likely to represent relevant populations, leading to more reliable outcomes. However, collecting large samples requires time and resources.

Fast Science is also related to the gaming of the system. As a hypothetical example, education scientists may gather data to find evidence for the theory that new teaching styles promote better learning. They then look at the data and realize that the intervention did not completely improve learning. However, if you squint, you may have a tendency to drop some of the troublesome outliers that you didn’t see any profit. So they do exactly that.

This is an example of what is known as “suspecting research practices” because traditional standards are not considered entirely fraudulent. Research in many areas suggests that these practices are widespread, with about 50% of scientists saying they are engaged in them at least once.

Fast science is also associated with more clearly unethical practices.

Reports of manufactured data may be attempting to be made public as quickly as possible by potential scientists. The industry was born around scientific fraud. This is known as a “paper mill.” These organizations produce articles on manufactured data and sell authors to those papers.

Why reliable science takes time

So, what does slow science look like and how does it help?

The late British statistician Douglas Altman provided one of the most famous explanations of the slow science mantra. “We need research done for the right reasons, better research, and research.”

In many ways, it is the opposite of fast science: large samples and careful, well documented transparent practices.

Remember the hypothetical example of scientists testing new educational practices. Rather than jumping into data collection immediately, the slow practice is to write a “registration report” first. In other words, scientists write their theory and how they propose to test that theory, and send it for peer review before collecting the data.

The journal follows the usual process of recruiting peer reviews and allowing scientists to modify reports according to those reviews. The authors will then collect the data and are guaranteed to be published in the journal as long as they follow the agreed methods.

Registered reports have two major advantages: This allows peer feedback while still possible to improve research and removes incentives to engage in suspicious or fraudulent practices. Using registered report formats can take some time. However, it is linked to more reliable findings.

Two other slow practices worth mentioning are to conduct research in a reproducible way to correct errors in existing research institutions.

In theory, all science must be reproducible. This means that scientists need to share methods and data so that other scientists can confirm and build their work (develop new recipes to continue food similarity).

Similarly, cleaning up the scientific record is extremely important. For the same reason that chef Gordon Ramsay likes to clean the kitchen before improving it, science needs to deal with what is reliable before building up existing research findings.

This means carefully passing through existing publications to find research that shows signs of being manufactured or unreliable. It is rare among university scientists as they do not usually bring about publications. But that’s very important.

Slow science slowly grows steam

It takes courage to engage in slow science right now.

The university wants to raise the ranking list of universities. These rankings are driven by public disclosure. Therefore, universities hire, promote and maintain scientists based on their publications. This makes it dangerous to slow down.

However, there are several reasons why you would like it. A movement is underway to redefine the quality of research to consider more aspects of slow science.

The Declaration on Research Assessment is a global initiative that moves away from ranking systems that ignore slow science principles.

Grassroots organizations are also creating platforms for more open and rigorous peer review.

And proponents of more careful research practices have recently been appointed to important positions, such as research funders and academic journals.

These developments are worth following and building because society doesn’t need a mountain of low quality science. We need science that is worthy of trust.

This article will be republished from the conversation under a Creative Commons license. Please read the original article.