Subscribe to get an e-mail whenever a new post is published

June 28, 2011

Improving science

© 2011, Coert Visser

Science can be defined as the systematic enterprise that builds and organizes knowledge in the form of testable explanations and predictions about the world. Science is one of humanity's greatest inventions which has the potential of improving our lives and our societies. The core of the scientific process consists of scientists making observations, reading scientific literature, formulating questions, testing ideas through systematic studies, and sharing their findings. The system of science contains principles and sets of rules which help make science self-correcting and cumulative. Scientists are required to share not only their findings through publication but also provide detailed descriptions of their studies so that replication of their studies by other scientists becomes possible. A process of peer review functions as a filter to guarantee that only research that meets scientific standards is published in journals. Replication studies make it possible to test findings using the same methods but with different subjects and experimenters.

By building on robust findings and revising or discarding findings that fail to be replicated scientific theories are built over time. Here is a simplified description of the subsequent stages of knowledge development in science. In early stages, scientists observe associations between phenomena. Then, they are able to make predictions. Next, they gain insight in underlying causal mechanisms. Finally, they are able to control phenomena and to use scientific knowledge.

An example of science at its best is the theory of evolution. Roughly 150 years ago, Charles Darwin, after many years of careful observation and thought, produced a theory that still drives the contempory scientific agenda. Darwin's breakthrough insight was that in organisms whose environment changed nonrandomly and whose reproductive success in that environment depended on inherited traits, evolution became inevitable. Today, the theory of evolution has developed into an extremely well-tested macro theory consisting of a diverse set of micro theories which together yield invaluable technologies.

There is a difference between this idealized description of the scientific enterprise on the one hand and scientific practice on the other. While science as an idea is wonderful there are many aspects of scientific practice which require improvement. When we think about refining and improving the scientific process we usually focus on improving the core of the scientific process. For example, we may discuss whether or not the large scale use of significance testing in social science is justified or not. While important, this is not the only area in which improvement may be achieved. The effectiveness and efficiency of the scientific process is also importantly determined by several other factors, somewhat more in the periphery, which are easily overlooked. The scientific process can be seen as a chain of events. Strengthening the weakest links in this chain may yield the greatest improvements. The figure below is a simplified depiction of this 'chain'.


Here are some ideas about how each of these links in this chain affects the overall effectiveness and efficiency of scientific progress and some (sometimes implicit) suggestions for improvement:
  1. Research funding: in order to be able to do research scientists (usually) have to write grants to find financial resources. Decisions about grants for research are not only often made on scientific grounds alone. Three factors threaten scientific impartiality in grant allocation: 1) commercialization of science: an example is how most medical research is now funded by industry; 2) political agendas: political convictions may have an important influence on whether certain scientific research is funded or not (for instance stem cell research); 3) vested interests of conventional scientists: decisions about grants are usually made by well-known scientists who may be wedded to conventional ideas and approaches and may have an interest in these conventional ideas and practices.
  2. Original research: the quality of the methods and techniques used remains an essential determinant of scientific progress. Ongoing discussions about the proper design of studies and about the correct use of (statistical) analyses remain important.
  3. Peer review: in order to be published papers have to pass the process of peer review. While this is intended to provide a guarantee of scientific quality there is certainly room for improvement in peer review practice. Many instances of peer review failure have been identified which may have to do with limited or no access of the reviewers to the data and/or the details of the method or with bias by vested interest of the anonymous reviewers.  Open peer review has been suggested as a way to improve the transparency and accuracy of the peer review process. 
  4. Replication studies: replication studies play an essential role in the self-correcting nature of science. But in practice scientists are not eager to do replication studies, in particular when they think the results of the study were wrong. Also, journal editors typically prefer to publish groundbreaking new research and are often very reluctant to publish replication studies. Some even have an explicit policy of not publishing replication studies (source). Recently, Daryl Bem reported some very surprising results on so-called precognition. Here is an article about how replication studies were rejected by journals and about some solutions suggested by psychologist Richard Wiseman.
  5. Follow up studies: follow up studies play an important role in the cumulative, progressive character of science. By building on previous research, theories may be tested and refined and, gradually, theories may be developed which are massively supported by evidence. This paper discusses, among other things the critical and perhaps underestimated role that anomalies play in building better theory. 
  6. Further dissemination of scientific knowledge: Scientific knowledge may be spread through popularized science books, magazines, websites like Google scholar, Wikipedia and Youtube, television programs,  and education. An interesting and useful development is also how some universities have started to create chairs for the public understanding of science (for instance Oxford). There are now also peer reviewed scholar sites (scholarpedia.org). And there are more and more researchers uploading pdf files on their websites. Some universities are very generous in sharing their knowledge, too. Despite all these initiatives, there remains a substantial gap between scientific developments and the public's knowledge and understanding of it. Some worry that the gap growths. Even basic scientific knowlegde is unknown to large proportions of the population of many countries. Some people argue for a free open access to all publicly funded scientific knowledge for everyone. 
  7. Science education: education is a specific way of disseminating scientific knowledge and to teach students about scientific methods and making students scientifically literate. Scientific illiteracy of politicians, parents and school managers may pose serious threats to this process. For example, politicians may argue that equal attention should be paid in class to creationism and evolution theory ("teach both sides"), which, wrongly, suggests that there is, among scientists, a controversy about the evidence for evolution (actually, the evidence is overwhelming). Politicians may even go so far as to make libel laws which may practically prevent scientists to criticize claims made by sellers of alternative medicine or other approaches which are unsupported by scientific evidence.  
  8. Application of scientific knowledge: the availability and application of scientific knowledge would be facilitated by improving the practices described above. Scientific progress would happen faster and would benefit more people. 
The scientific enterprise is extremely important and we should continue to improve scientific practice.