Finding a Balance: Complexity vs. Explanatory Power

 

Recently, I read a paper as a part of a weekly meeting of our campus’ Behavioral Ecology and Evolution Research group. From what I understand as a newcomer to the research and literature world, this article is suggesting that ecological research is becoming more complex and, as a result, is losing its explanatory power. These statements were based on Low-Décarie, Chivers, and Granados’ careful analysis of over 18000 articles published by various ecological societies. Based on their desire to gain perspective on the current progress of science, they used the analysis of the articles to come up with three main hypotheses to explain the apparent increase in complexity and decrease in explanatory power of the studies:

  1. Low hanging fruit: Basically, all the simple, broader findings are made early on in a discipline’s development, leaving the harder, more complex questions harder to reach. In addition, more and more scientists feel the need to move away from single species studies, taking on more complex community studies with less emphasis on observational and statistics-independent topics.
  1. True mean explanatory power: By this hypothesis, additional effort by way of more replicates or larger sample sizes could be an accomplice for the decline in the coefficient of determination, R2 (a way of interpreting how much of the data’s variance a model “explains”). Here, publication bias in favor of large R2 values is introduced, based on the observed preference of correlation coefficients (r) rather than the coefficient of determination (R2) when values are low.
  1. Shifting publication bias: Lastly, there was an observed trend that studies that would have been put on the back burner in the past because of their low R2 values are now getting published. Alternatively, studies with higher R2 values may just be gaining publications in journals other than those considered in this study.

In my opinion, the low hanging fruit hypothesis is a viable option. When considering what to research, I feel a subconscious pull away from topics that have already been well covered. All the sexy topics like climate change, ocean acidification, and many others have been so widely studied that it seems to be a daunting task to create an original idea in those areas. Under this light, complexity seems to be the only viable option. However, I have always been told that it is better to do something simple well than to do something complex sloppily.

That being said, taking the low hanging fruit presents its own challenge: larger sample sizes and more replicates become essential to tease out weaker, less obvious relationships with ultimately lower R2 values. Papers with these lower R2 values typically hit a “publishing glass ceiling” of sorts, and have a harder time gaining publications in more prominent journals. However, this study found that there were a substantial number of papers published with less than perfect R2 values. This shift in publication bias could have been due to changes in ecology’s publication attitude or by an “Aha!” moment of sorts, whereby the importance of publishing “negative results” was recognized. Presenting both sides of the scientific story is unarguably vital to reminding researchers that things don’t always work out the way we expect. It is important to understand our pitfalls as well as our successes to forge ahead with improved studies.

Of course, after reading and discussing this article with fellow students and faculty, I was understandably overwhelmed. While some people displayed an apathetic attitude towards this topic, most were driven to share their personal feelings on the matter. Though I am relatively new to this concept in science, I could understand how it might seem like beating a dead horse. Ultimately, however, I do agree it is important to keep these kinds of conversations open and flowing, so that as researchers we never forget to reflect on where we started and how far we have come.

Paper:

Etienne Low-Décarie, Corey Chivers, and Monica Granados 2014. Rising complexity and falling explanatory power in ecology.Frontiers in Ecology and the Environment 12: 412–418. http://dx.doi.org/10.1890/130230

 

This entry was posted in Blog Posts, Featured.
Bookmark the permalink.
Follow any comments here with the RSS feed for this post.
Post a comment or leave a trackback: Trackback URL.

2 Comments